File: ChangeLog

package info (click to toggle)
python-stem 1.2.2-1~bpo70%2B1
links: PTS, VCS
area: main
in suites: wheezy-backports
size: 4,576 kB
sloc: python: 20,108; makefile: 126; sh: 3
file content (18524 lines) | stat: -rw-r--r-- 731,048 bytes
parent folder | download | duplicates (2)
commit 2ef492467d13a121a76ac8ebfdc2dfbaa4b20e1d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 13 18:15:41 2013 -0700

    Stem release 1.1.0

commit 216937c0821e27301f60dd77a1addfb286b1b3ea
Merge: 242569a 6c81b10
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 13 17:40:51 2013 -0700

    Correcting test issues with python 2.6 and 3.x
    
    Generally I just work with python 2.7, so other interpretor versions just get
    love prior to releases. Most of this is issues with our tests, but @lru_cache
    introduced a lot of legitimate regressions for python 3.x.

commit 6c81b101dd69cf9f4b9170aa089d98e38220f383
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 13 17:25:34 2013 -0700

    Python 3 compatability for mapaddress test
    
    Quite a few byte vs unicode gotchas. Also, the behavior of indexing into byte
    strings changed in python 3 (it provides ints rather than characters).
    
    ======================================================================
    ERROR: test_mapaddress
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/control/controller.py", line 813, in test_mapaddress
        test.network.negotiate_socks(s, '1.2.1.2', 80)
      File "/home/atagar/Desktop/stem/test/data/python3/test/network.py", line 321, in negotiate_socks
        request = "\x04\x01" + struct.pack("!H", port) + "\x00\x00\x00\x01" + "\x00" + host + "\x00"
    TypeError: Can't convert 'bytes' object to str implicitly

commit 992c94bc6065cee6281266df6e482a6367dd56c6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 13 16:39:28 2013 -0700

    Fixing AttributeError in use_directory_mirrors()
    
    Correcting a couple issues, one in the tests and another in the remote
    descriptor module, causing integ failures.
    
    ======================================================================
    ERROR: test_use_directory_mirrors
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "/home/atagar/Desktop/stem/test/integ/descriptor/remote.py", line 62, in test_use_directory_mirrors
        downloader.use_directory_mirrors()
      File "/home/atagar/Desktop/stem/stem/descriptor/remote.py", line 419, in use_directory_mirrors
        if Flag.V2DIR in desc.flags:
    AttributeError: 'str' object has no attribute 'flags'
    
    ======================================================================
    ERROR: test_using_authorities
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "/home/atagar/Desktop/stem/test/integ/descriptor/remote.py", line 34, in test_using_authorities
        for authority, (address, dirport) in stem.descriptor.remote.DIRECTORY_AUTHORITIES.items():

commit bfd5a667f150d4fef77c6a730432494ee930610b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 13 13:20:08 2013 -0700

    Fixing python 3 testing issue with key certs
    
    Test file wasn't being opened in bytes mode, causing...
    
    ======================================================================
    ERROR: test_metrics_cert
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/descriptor/networkstatus.py", line 178, in test_metrics_cert
        cert = next(stem.descriptor.parse_file(cert_file))
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/__init__.py", line 200, in parse_file
        for desc in file_parser(descriptor_file):
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/__init__.py", line 241, in _parse_metrics_file
        for desc in stem.descriptor.networkstatus._parse_file_key_certs(descriptor_file, validate = validate, **kwargs):
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/networkstatus.py", line 252, in _parse_file_key_certs
        yield stem.descriptor.networkstatus.KeyCertificate(bytes.join(b"", keycert_content), validate = validate)
    TypeError: sequence item 0: expected bytes, str found

commit a303b137a9ada7cc9a31a57ba2dcbeae89dd83a3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 13 12:55:24 2013 -0700

    Fixing python 3 compatability for remote descriptor unit tests
    
    ======================================================================
    ERROR: test_query_with_timeout
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "/opt/python3.3/lib/python3.3/unittest/mock.py", line 1079, in patched
        arg = patching.__enter__()
      File "/opt/python3.3/lib/python3.3/unittest/mock.py", line 1132, in __enter__
        self.target = self.getter()
      File "/opt/python3.3/lib/python3.3/unittest/mock.py", line 1294, in <lambda>
        getter = lambda: _importer(target)
      File "/opt/python3.3/lib/python3.3/unittest/mock.py", line 988, in _importer
        thing = __import__(import_path)
    ImportError: No module named 'urllib2'

commit ab86475d484c9c727ba17b9cf39e1661d0e72a8a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 13 12:49:22 2013 -0700

    Fixing python 3 issue with orphaned annotation exceptions
    
    ======================================================================
    ERROR: test_annotations
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "/opt/python3.3/lib/python3.3/unittest/mock.py", line 1087, in patched
        return func(*args, **keywargs)
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/descriptor/server_descriptor.py", line 224, in test_annotations
        self.assertRaises(ValueError, list, desc_iter)
      File "/opt/python3.3/lib/python3.3/unittest/case.py", line 570, in assertRaises
        return context.handle('assertRaises', callableObj, args, kwargs)
      File "/opt/python3.3/lib/python3.3/unittest/case.py", line 135, in handle
        callable_obj(*args, **kwargs)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/server_descriptor.py", line 157, in _parse_file
        raise ValueError('Content conform to being a server descriptor:\n%s' % '\n'.join(annotations))
    TypeError: sequence item 0: expected str instance, bytes found
    
    ----------------------------------------------------------------------
    Ran 31 tests in 0.028s
    
    FAILED (errors=1)

commit 1ed6e2b7e553414736913b3cd440843d66ccea94
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 13 12:40:46 2013 -0700

    Hash functions for descriptor classes
    
    Ok, finally found out what this python 2 vs 3 hashing oddity is. If you define
    an __eq__() method in python3 then it doesn't inherit its parent's __hash__().
    This makes some sense, though it's unintuitive as hell. Oh well...
    
    http://stackoverflow.com/a/1608882/1067192

commit eaa89a67ebf6339a68972e705b76a8e01677d7b9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 13 12:21:06 2013 -0700

    Fixing python 3 compatablity for tordnsel test
    
    ======================================================================
    ERROR: test_parse_file
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/descriptor/tordnsel.py", line 55, in test_parse_file
        descriptors = list(_parse_file(io.BytesIO(TEST_DESC)))
    TypeError: 'str' does not support the buffer interface
    
    ----------------------------------------------------------------------
    Ran 1 test in 0.001s

commit e1bf7cbacb8797845c1bc2a778baa93d9748f4c9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 13 12:10:38 2013 -0700

    Fixing a couple python 3 network status document TypeErrors
    
    ======================================================================
    ERROR: test_duplicate_fields
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/descriptor/networkstatus/document_v3.py", line 247, in test_duplicate_fields
        if not is_consensus and lines[index].startswith('dir-source'):
    TypeError: startswith first arg must be bytes or a tuple of bytes, not str
    
    ======================================================================
    ERROR: test_misordered_fields
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/descriptor/networkstatus/document_v3.py", line 225, in test_misordered_fields
        elif not is_consensus and lines[index].startswith('dir-source'):
    TypeError: startswith first arg must be bytes or a tuple of bytes, not str
    
    ----------------------------------------------------------------------
    Ran 35 tests in 0.234s

commit efe5fbdd56b28b3f76de144e71901c0c75ebe5ac
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 12 20:14:59 2013 -0700

    Caching hash values
    
    Well, even if we can't use @lru_cache we can still do it the old fashioned way. :P

commit d32d4ab45d27776385036b1a57d9329f270354fc
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 12 20:11:23 2013 -0700

    Dropping @lru_cache for Version hashing
    
    The @lru_cache calls hash() so if ever called this would infinitely recurse.

commit 208c03cded57507f835265d7e4d2b4677d0e30e5
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 12 20:07:52 2013 -0700

    Fixing python 3 issue concerning exit policy compression
    
    Oops, forgot about the byte vs unicode differences when doing compression for
    the ExitPolicy.

commit 4ef5ee1ee8fbecabeaee872f2681cdda1fb86584
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 12 20:18:12 2013 -0700

    Adding hash functions to exit policy classes
    
    Huh, didn't expect that. To work the @lru_cache requires that all arguments
    (including the class in the ase of methods) are hashable. This makes perfectly
    sense, after all the cache is a 'argument => cached value' dictionary.
    
    Under python 2.x all seemed to be well, but under python 3.x the @lru_cache
    complained that our classes weren't hashable (which... well, they aren't). Why,
    then, did it work under python 2.x? Turns out there's a subtle difference where
    our object parent provides a __hash__ method for our id, but in python 3.x it's
    a little different (object still does, but for reasons I haven't been able to
    figure out it doesn't for our classes).
    
    The ExitPolicyRule can certainly be hashable. The ExitPolicy, however, is a lot
    trickier due to its lazy loading of rules. For now just opting for the same
    behavior as python 2.x and using the address.

commit 677ee1570cc1378b8abadb4071627ff53510cb8a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 12 19:23:04 2013 -0700

    Python 3 lru_cache isn't a collections annotation
    
    Oops, it's in functools, not collections.

commit 1dfa353ecfbef39ee606231e745b9b6318fd3a31
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 12 17:35:55 2013 -0700

    Using @lru_cache for exit policy's str()
    
    Oops, missed one of our instances of manual caching.

commit e89c829fb75331fdc0d5aa03475c751371cb6c7e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 12 16:57:50 2013 -0700

    Using python3's unittest.mock module when available
    
    Guess it's been a while since I last tried to test under python3. From python
    3.3 on up the Mock library is built into the unittest module. Accounting for
    this so we can test under python3 again.

commit d09bd8b365819f4ea0d0a7d6b782b1021dba91c1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 12 16:37:02 2013 -0700

    Fixing python3 issue related to unicode exit policies
    
    Our recent exit policy caching caused a regression under python3...
    
    Traceback (most recent call last):
      File "./test/data/python3/run_tests.py", line 25, in <module>
        import test.runner
      File "/home/atagar/Desktop/stem/test/data/python3/test/runner.py", line 51, in <module>
        import stem.connection
      File "/home/atagar/Desktop/stem/test/data/python3/stem/connection.py", line 109, in <module>
        import stem.control
      File "/home/atagar/Desktop/stem/test/data/python3/stem/control.py", line 145, in <module>
        import stem.descriptor.microdescriptor
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/__init__.py", line 546, in <module>
        import stem.descriptor.server_descriptor
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/server_descriptor.py", line 91, in <module>
        REJECT_ALL_POLICY = stem.exit_policy.ExitPolicy("reject *:*")
      File "/home/atagar/Desktop/stem/test/data/python3/stem/exit_policy.py", line 164, in __init__
        self._input_rules = zlib.compress(','.join(rules))
    TypeError: 'str' does not support the buffer interface

commit c10eef255f0948bcc3cd86eab4221e2b335615eb
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 12 17:13:56 2013 -0700

    Fixing python 2.6 compatability issue with lru_cache
    
    The recepie is using set syntax only available in python 2.7 and above...
    
    Traceback (most recent call last):
      File "./run_tests.py", line 18, in <module>
        import stem.prereq
      File "/home/atagar/Desktop/stem/stem/__init__.py", line 420, in <module>
        import stem.util.enum
      File "/home/atagar/Desktop/stem/stem/util/enum.py", line 43, in <module>
        import stem.util.str_tools
      File "/home/atagar/Desktop/stem/stem/util/str_tools.py", line 21, in <module>
        import stem.prereq
      File "/home/atagar/Desktop/stem/stem/prereq.py", line 29, in <module>
        from stem.util.lru_cache import lru_cache
      File "/home/atagar/Desktop/stem/stem/util/lru_cache.py", line 37
        fasttypes = {int, str, frozenset, type(None)},
                        ^
    SyntaxError: invalid syntax

commit 242569af36ebb1f30a5d2629254b55320cfd5298
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 12 13:11:11 2013 -0700

    Replacing manual caching with @lru_cache
    
    Replacing instances where we do...
    
    def get_stuff():
      if self._stuff is None:
        self._stuff = ... calculated stuff...
    
      return self._stuff
    
    ... with a @lru_cache().

commit 94b4924a3970fe3877b276ad2d07ccd149ed8d3a
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 10 08:51:59 2013 -0700

    Revert "Chaching version comparison and hash function"
    
    Wow, shame on me for not even running unit tests. This causes version
    comparisons to infinitely recurse - hot.

commit e28032aebb14043c2149fe949ef49ddb82083c21
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Oct 9 08:55:00 2013 -0700

    Mocking pydocs referenced functionality we've removed

commit 281db911a5912836ceb63247b55d8f8300b639ec
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Oct 8 09:11:48 2013 -0700

    Chaching version comparison and hash function
    
    Versions are intended to be read-only objects, so we might as well add caching
    for these methods.

commit ba9799e20e8046ef1316616a3c665f8e4558665a
Merge: d1218e0 1920baf
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Oct 8 00:42:46 2013 -0700

    Reducing memory usage of descriptor content
    
    Handful of improvements to reduce the memory usage of stem's descriptor content
    (with a special focus on server descriptors). This drops memory usage by
    roughly 20% without having a sizable impact on runtime.
    
    Script used for testing:
    
      import resource
      import time
    
      from stem.descriptor import reader
    
      print "memory (initial): %s" % resource.getrusage(resource.RUSAGE_SELF).ru_maxrss
    
      start_time = time.time()
    
      with reader.DescriptorReader(['/home/atagar/.tor/cached-descriptors']) as descriptor_reader:
        server_descriptors = list(descriptor_reader)
    
      print "memory (after read): %s" % resource.getrusage(resource.RUSAGE_SELF).ru_maxrss
      print "read time: %0.2f seconds" % (time.time() - start_time)
    
    Before:
    
      % python scratch.py
      memory (initial): 7468
      memory (after read): 50572
      read time: 6.40 seconds
    
    After:
    
      % python scratch.py
      memory (initial): 7532
      memory (after read): 40816
      read time: 6.31 seconds

commit 1920baff1ad557a2a556dcb0cb46609f50f3cde1
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Oct 8 00:37:11 2013 -0700

    Compressing unparsed exit policy content
    
    Exit policies constitute a rather substantial chunch of server descriptors. We
    already optimize our runtime by lazily parsing policies on demand. On
    reflection though, we can improve our memory usage quite a bit too by
    compressing the unparsed content. This drops our memory usage by roughly 6%
    without impacting runtime.

commit 2ce003ef33c8693234f7dce562fdd3cff4e37e07
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Oct 8 00:15:51 2013 -0700

    Caching parsed tor versions
    
    Reusing parsed tor versions saves us an impressive 3.7% memory usage without
    any noticable impact on parsing times.

commit dbee38d1b95d943435cbb6bc989a9a5024b4e8e8
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 7 23:09:04 2013 -0700

    Referencing a single reject-all policy
    
    Oddly, the only exit policy I can find with greater than five occurances is the
    reject-all policy. Maybe a scripting bug. Reguardless, referencing a single
    reject-all instance saves us a little memory (1.3 MB with the present
    consensus, which is 1.6%).

commit a0ce8664bc5758030c912456fcf2ff33a619373f
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 7 22:11:44 2013 -0700

    Storing a single IPv6 default exit policy
    
    Each server descriptor constructed its own default value for the IPv6 exit
    policy, which was foolish. ExitPolicy instances are read-only instances, so all
    descriptors might as well use the same default. This saves roughly 2.3 MB of
    memory when downloading all of the current server descriptors (or roughly 3%).

commit 19cca3a04b9b02e94d22ebb877af3ca2ca9ad5e4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 7 21:14:00 2013 -0700

    Using @lru_cache for exit policies
    
    The ExitPolicy and ExitPolicyRule classes are a perfect fit for @lru_cache.
    They're read-only classes that already do a fair bit of caching. The annotation
    lets us avoid doing this ourselves.

commit d1218e00feeb1d4afc93a105b9a99d38a4ffa841
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 6 23:15:05 2013 -0700

    Switching to the https endpoint of www.atagar.com
    
    Now that we have SSL support there's little reason to point to the http
    endpoint.

commit d9fbdc968479b0d413d1ff875ca19c855e7ce2ae
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 6 17:32:10 2013 -0700

    Adding python 2.x backport of @lru_cache
    
    Python 3.2 added a memoization annotation to python's functools module. This is
    very, very handy, allowing us to avoid the caching boilerplate I do way too
    often...
    
      def get_foo():
        if self._foo is None:
          ... stuff to calculate self._foo...
    
        return self._foo
    
    With a memoization function this becomes...
    
      @lru_cache()
      def get_foo():
        ... stuff to calculate self._foo...
    
    This is a MIT licensed backport from...
    
      http://code.activestate.com/recipes/578078-py26-and-py30-backport-of-python-33s-lru-cache/
    
    Looking forward to when we require python 3.2 so we can use the builtin!

commit e60a033941c63885cb349d3bc181bbbf66227acf
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Oct 2 09:19:38 2013 -0700

    Tests can fail with TypeError when unable to start tor process
    
    Traceback (most recent call last):
      File "./run_tests.py", line 404, in <module>
        main()
      File "./run_tests.py", line 245, in main
        println(exc, ERROR)
      File "/home/atagar/Desktop/stem/test/output.py", line 59, in println
        msg = term.format(msg, *attr)
      File "/home/atagar/Desktop/stem/stem/util/term.py", line 96, in format
        return (CSI % ";".join(encodings)) + msg + RESET
    TypeError: cannot concatenate 'str' and 'UnsupportedOperation' objects

commit bcd849f0cbf3e5a823049dcea51f9158baf897b9
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Oct 2 07:32:11 2013 -0700

    Adding a hash method to our Version class
    
    Our Version class lacked a __hash__() method, causing it to behave
    unintuitively in dictionaries and sets...
    
      >>> from stem.version import Version
      >>> foo = set([Version('0.2.4.9-alpha')])
      >>> bar = set([Version('0.2.4.9-alpha')])
      >>> foo.difference(bar)
      set([<stem.version.Version object at 0xb71fae0c>])
    
    This is because python opts for the object's identity when a hash method does
    not exist.

commit a5f9e9cd5b4662bc5d34e19621fa9667cf71a07d
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Sep 30 14:39:29 2013 -0700

    Halo for windrose icon
    
    Adding a blue halo for the center of the windrose. This doesn't address the
    main thing I dislike about the icon, which is the pixelation of the black
    lines, but still it makes it look a bit nicer. I've tried several things to
    address the black line pixilation (blur, thickening, etc) but this is the best
    improvement I've yet found.

commit 4866bd336cd70eedcd00a685df26737fd7c37939
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 29 20:07:33 2013 -0700

    Adding utilities tutorial to the menu
    
    Oops, forgot one of the spots we need to add new pages...

commit b0be16b7e93f56e6189cea6b0b991cded1db6a02
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 29 20:01:16 2013 -0700

    Adding connection resolution tutorial
    
    Introducing a new tutorial section called 'East of the Sun & West of the Moon'
    (... I love that fairytale) that introduces users to our utility modules.
    Presently this just has connection resolution, a spiffy new feature I added
    last weekend.

commit ec24345427ac804640a425bbb900bf41aae3be83
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 28 16:49:25 2013 -0700

    Linking to the DocTor repository
    
    Oops, our example page was still linking to my tor-utils repository rather than
    DocTor's.

commit 276fb65ef4dbf0536be91e3fcb1912363d5ada73
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Sep 23 12:03:25 2013 -0700

    Only report connection resolvers that are in our PATH
    
    Having get_system_resolvers() check that resolvers are in our PATH before
    returning them.

commit 7c0002d40db176d02c6a57e400ec53b5fb7a63a4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Sep 23 00:28:01 2013 -0700

    Proc unit tests didn't reflect protocol addition
    
    Oops, our unit tests didn't account for proc now providing the protocol. Caught
    by our jenkins tests.

commit 1082349b59eb6ecd89eca1a1f2aa2ab1df007315
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 22 23:38:17 2013 -0700

    Connection resolution support
    
    Adding long overdue support for process connection resolution...
    
    https://trac.torproject.org/7910
    
    This is a highly popular capability of arm, and stem's counterpart for it is
    quite a bit cleaner (with tests!). That said, this still doesn't get around
    tor's annoying DisableDebuggerAttachment feature which screws up proc
    permissions. That's something I'll need to figure out before our next arm or
    stem release...

commit 87b0d512e2b83a2dcced709e33128eaed10cc439
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 22 12:25:00 2013 -0700

    Adding function for getting the system's resolvers
    
    Adding a get_system_resolvers() function that provides the connection resolvers
    that are likely to work on a given platform.

commit aee7de80db712ab1003eb6d8ac52e78d5f0f18cc
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 21 17:46:02 2013 -0700

    Adding a is_private_address() function to stem.util.connection
    
    Helper function for determining if an IPv4 address belongs to a private range
    or not.

commit f40c912ecb8fdb542a7bd12e7a440e5301ea9798
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 21 17:41:13 2013 -0700

    Adding a get_vote() method to the DescriptorDownloader
    
    Stem is frequently being used for answering questions concerning directory
    authorities, so adding a method to more intuitively get their votes.

commit 1d2fe8a218768eb5a1713cd2daa6d54df0d6c7ca
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 21 12:05:25 2013 -0700

    Disable caching of HiddenServiceOptions
    
    Stem's controller has pretty straight forward caching for tor's configuration
    options. If you call 'GETCONF foo' then the value of foo is cached until you
    call either 'SETCONF foo' or 'RESETCONF foo'.
    
    In general this is all well and good, but HiddenServiceOptions behave in a
    different fashion. With those the caller might call 'SETCONF HiddenServiceDir'
    and expect 'GETCONF HiddenServiceOptions' to change. As such, opting out of
    caching hidden service options.
    
    This was a bug uncovered by wayzard on...
    
    http://stackoverflow.com/questions/18777837/tors-stem-module-python-loading-but-not-unloading-config
    https://trac.torproject.org/projects/tor/ticket/9792
    
    Prior to this fix...
    
    >>> from stem.control import Controller
    >>> controller = Controller.from_port()
    >>> controller.authenticate()
    >>> controller.get_conf_map('HiddenServiceOptions')
    {'HiddenServiceOptions': []}
    >>> controller.set_options([('HiddenServiceDir','/tmp/my_hidden_service'), ('HiddenServicePort', '1236 127.0.0.1:1236')])
    >>> controller.get_conf_map('HiddenServiceOptions')
    {'HiddenServiceOptions': []}
    
    Now it behaves as expected...
    
    >>> from stem.control import Controller
    >>> controller = Controller.from_port()
    >>> controller.authenticate()
    >>> controller.get_conf_map('HiddenServiceOptions')
    {'HiddenServiceOptions': []}
    >>> controller.set_options([('HiddenServiceDir','/tmp/my_hidden_service'), ('HiddenServicePort', '1236 127.0.0.1:1236')])
    >>> controller.get_conf_map('HiddenServiceOptions')
    {'HiddenServicePort': ['1236 127.0.0.1:1236'], 'HiddenServiceDir': ['/tmp/my_hidden_service']}

commit cfec958ddec6c1b104def1fdb40cacbb4e73b79d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 21 11:32:03 2013 -0700

    Adding default and constraints for Support022HiddenServices
    
    Reflecting constraints for a new param in network status documents...
    
    https://gitweb.torproject.org/torspec.git/commitdiff/6c974c5

commit e2cff0be96640b584fa1e7044878891ff4bb69b0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 14 17:25:09 2013 -0700

    Dropping config's _raw_contents reference
    
    On reflection we never do anything with the _raw_contents. In fact, we don't
    even have getter methods for it, and never use it within the Config class
    itself. Dropping it to both save a little memory and make it easier to purge
    values.

commit fb3907913b09770ed6de5ff8914553e424fa8f47
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 14 13:53:33 2013 -0700

    Logging tor's output during integ test runs
    
    Saving tor's logging output to disk to help with troubleshooting issues.

commit 00b97f1d711d0d713d03a9ace72ba42cf0e3003e
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Sep 9 12:19:04 2013 -0700

    Changing 'DocTor' spelling
    
    Karsten has been pretty consistent about calling the project 'DocTor' rather
    than 'Doctor' so I should do the same.

commit 7679276642ddf3a64aa8473f2c6de68549aea6df
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Sep 9 12:16:21 2013 -0700

    Adding TorPS to examples page

commit 032df08a5591afaf810dabdbca9177fcc528a7ee
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Sep 2 13:16:00 2013 -0700

    Adding stem.util.system.get_name_by_pid()
    
    Function for getting the command running under a given pid. This is the inverse
    of get_pid_by_name().

commit 26dadd28e70efc5fac826b62bf4f4b35b7acadeb
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Sep 2 09:30:34 2013 -0700

    Replacing DIRECTORY_AUTHORITIES constant with get_authorities()
    
    On reflection a DIRECTORY_AUTHORITIES constant is not something I want to vend
    in our next stem release. That would make it tricky to be flexible about where
    we get that information. For instance, we might want to get this from tor
    itself (if that ever becomes an option), or possibly a config file rather than
    hardcoded dict.

commit e5bfee6dc5ec66f46560e2b9bae68340569e1d77
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 1 16:24:48 2013 -0700

    Typo correction in tutorial
    
    Caught and fixed by irregulator.

commit ec0c7eb0c3b96fc50ddc117bd83f4f6477c13f4c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 1 11:52:21 2013 -0700

    Minor improvements for connection examples
    
    Changes include...
    
    * more conventional example using the 'with' keyword
    * disconnecting the controller afterward
    * corrected indentation

commit 53068b2f3066df1a01be5f13a3578263e21fac54
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 1 10:23:49 2013 -0700

    FAQ entry for 'How do I connect to Tor?'
    
    While code reviewing ra's RTT Prober work I realized that our site didn't
    explain the various ways a user can get a Controller. Adding a FAQ entry with
    examples that explains it.

commit 3288f61da61932664c677619fe303d1d076dbe83
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Aug 30 14:14:22 2013 -0700

    Fixing case for new faq entry
    
    Oops, FAQ entry titles shouldn't be all caps.

commit fdd84baa1edd30cf11d47ad7b5d1c490024c2990
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Aug 30 10:34:00 2013 -0700

    FAQ entry for alternate tor controllers
    
    Entry for our FAQ linking to other controller libraries. This came out of the
    following tor-dev@ thread...
    
    https://lists.torproject.org/pipermail/tor-dev/2013-August/005340.html

commit e34b8c943a7e59ac8d8cbe9a77424b44e60e35d5
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Aug 30 08:58:36 2013 -0700

    Minor typo correction
    
    Little typo I spotted while answering...
    
    http://stackoverflow.com/questions/18372615/tors-stem-module-python-failing

commit a7ec700c1b366b5cd96ba894916ad0a6e6fa0b4c
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Aug 29 10:13:48 2013 -0700

    Explaining python's 'with' keyword
    
    We've had a couple stem newcomers that saw the 'with' keyword in our examples
    then misused it in their scripts. Adding a little tutorial on the keyword and
    the common mistake they've been making.

commit 303b4dfdac10ecb6c348dafbb08f27cae9799347
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Aug 27 22:37:56 2013 -0700

    Adding tor directory authority information
    
    To work our remote descriptor fetching module needed the addresses and dirports
    of voting authorities. However, on reflection it's useful if we reflect the
    rest of the authority information from tor's config.c as well. For instance,
    the v3ident is handy if our users want to query authority key certificates.

commit 847d2377846bf1a64881934e74c10193ec3fb130
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Aug 26 21:27:50 2013 -0700

    Using remote module for last descriptor tutorial
    
    Using the remote descriptor fetching module for the last descriptor tutorial
    rather than the controller. Also updating our unit tests for the tutorials.

commit d5c02be90b3a9d200773e4e128303bd4cee062a8
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Aug 26 09:22:50 2013 -0700

    Adding remote descriptor fetching to our tutorials
    
    Replacing the "Where can I get the current descriptors?" subsection with an
    example using the remote descriptor fetching module. This is the simplest way
    for our users to get descriptor data without the headache of piggybacking on a
    tor instance.

commit 0416b071d4f7b8518ed4e8d36c0afeb8b01e327f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 25 18:12:45 2013 -0700

    Adding doctor to examples page
    
    Swapping our 'Descriptor Monitors' sub-entry with an entry with Doctor.

commit f1e771a37828f85e3bd8631ad7234f50b0efac6e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 25 11:45:08 2013 -0700

    Support for TRANSPORT_LAUNCHED events
    
    Adding support for George's new TRANSPORT_LAUNCHED events...
    
    https://gitweb.torproject.org/torspec.git/commitdiff/48f6dd0

commit 4a84b47b6ebbd3a740884fd8e24f499b79582028
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 24 10:57:59 2013 -0700

    Adding support for extrainfo descriptor's 'bridge-ip-transports'
    
    Support for a new bridge-specific extrainfo descriptor field...
    
      https://gitweb.torproject.org/torspec.git/commitdiff/a01bb8e8e
    
    These are pretty much identical to bridge-ip-versions lines. Like those this is
    part of the ExtraInfoDescriptor class rather than BridgeExtraInfoDescriptor
    because the spec doesn't prohibit its appearance in public relays (though it
    doesn't really make sense there).

commit c076038e82fcd5f5188ec7f889c9c9dc5d5eafd3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 24 10:45:14 2013 -0700

    Handling bridge-ip-versions as a general extrainfo field
    
    In commit 0d01ad6 I added support for the 'bridge-ip-versions' lines. These are
    only relevant to bridge descriptors, and ineed don't presently appear in the
    extrainfo consensus. However, that once could be said about the 'bridge-ips'
    field too.
    
    It's wrong for bridge attributes to appear on public relays, but there's
    nothing in the spec prohibiting it. Hence moving this attribute's handler to
    the common ExtraInfoDescriptor parent class.

commit 003bbd7e8f0ad01300d4b0365302b31a921da37c
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Aug 23 09:08:18 2013 -0700

    Assigning stem's copyright to the Tor Project
    
    Revising the copyright notices throughout stem so both the Tor Project and I
    have copyright. This has a couple advantages...
    
    1. If I get nailed by the bus or otherwise somehow disappear then I want the
       Tor Project to have copyright. This will make life easier for any future
       stem maintainer by allowing them to share code with our other (mostly BSD
       licensed) projects.
    
    2. Contributors are understandably antsy about giving copyright to an
       individual rather than an organization. Thus far I've been able to largely
       sidestep those concerns by asking for contributions to be in the public
       domain, but there was one occasion where that was a no-go...
    
       https://trac.torproject.org/projects/tor/ticket/7566
    
       Hopefully assigning copyright to the Tor Project is something that'll make
       everyone happy.
    
    I've contacted Sean and he's fine with assigning the copyright for his stem
    contrutions to the Tor Project...
    
      Yes, I am still interested in and willing to assign my existing and
      future contributions (unless I explicitly say otherwise) to Stem to
      the Tor Project.  If the previous sentence is not legally binding
      enough, let me know.
    
    Tweaked the FAQ entry a bit to reflect this change.

commit d092810bb9810f4dfeb244b71ae10bd6031b52b3
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Aug 22 09:30:22 2013 -0700

    Adding TorDNSEL docs to the site

commit d7833f31c392623f8dad5a30e7b16713cff510aa
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Aug 22 09:20:20 2013 -0700

    Minor formatting changes for DNSEL addition
    
    Nothing functional, just rearranging a few things to satisfy my OCD. :)

commit 4631228400e0cca43d8c7ba514c40cbcbf2bda34
Author: Arlo Breault <arlolra@gmail.com>
Date:   Mon Aug 19 22:49:40 2013 -0700

    Parse exit list entry.
    
    Published by DNSEL or TorBEL to indicate what ip address exit relay X
    had at timestamp Y.
    
    See #8255

commit 4863c2281c26660a249378e65354a00e406fffbc
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Aug 19 18:15:47 2013 -0700

    Validating that votes only have a single authority entry
    
    Vote documents should only contain an entry from the authority that issued it.
    I was checking this in my consensus-checker monitor but Karsten made the good
    point that the check belongs in stem.

commit bba4069fdf70d422e4d6f4cbd8a50c0be6747732
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 17 12:48:49 2013 -0700

    Don't fail tests upon lingering thread
    
    The chief cause of our present jenkins failures are lingering threads...
    
    https://trac.torproject.org/projects/tor/ticket/8896
    
    This is certainly an problem. It indicates either a bug with stem closing
    threads, or that we aren't closing a controller in the tests. However, I'm
    tired of the false alarms from the jenkins tests, besides which this is a very
    minor issue.

commit 5be9904711cc7f15f438cad9456dfcc31309c73d
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Aug 16 08:41:45 2013 -0700

    Propagating additional descriptor keyword arguments
    
    Adding a 'kwargs' argument to all methods for fetching descriptors. This
    argument is passed along to the descriptor contructor. I just ran into a need
    for this because the NetworkStatusDocumentV3 class has a 'default_params'
    argument that I couldn't access through parse_file(). We'll likely be adding
    more constructor args in the future so this will help to make those... well,
    useful.

commit e9f23a1996356bef3edf22c143898c1f78498486
Author: Arlo Breault <arlolra@gmail.com>
Date:   Mon Aug 12 23:36:59 2013 -0700

    Typo in FAQ.

commit ced3c4883056c23fa43d3975850880841e1677c7
Author: Arlo Breault <arlolra@gmail.com>
Date:   Mon Aug 12 22:13:05 2013 -0700

    FetchDirInfoExtraEarly requires FetchDirInfoEarly.
    
    Tor complains,
    
    [warn] Failed to parse/validate config: FetchDirInfoExtraEarly requires
    that you also set FetchDirInfoEarly

commit bb3de90a82ff927c5005fbadce3b2ea975a15aae
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 11 13:49:30 2013 -0700

    Dropping document_handler from get_consensus()
    
    Oops, the document_handler is already a query parameter. There's no benefit to
    having a default value for it in get_consensus(), and that had the drawback of
    preventing us from asserting a default value.

commit 4fc65393374ed32a9c67c11758b0c5b94ed6f5f5
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Aug 7 09:49:25 2013 -0700

    Issuing DNS lookups via tor in client usage example
    
    Suggestion from Julian to issue DNS lookups via tor. This is based on...
    
    http://www.talesofacoldadmin.com/2012/11/13/proxying-dns-with-python.html

commit e0956015a1589dba607dc41ffa486e20c00f863b
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Aug 7 09:09:14 2013 -0700

    Correcting definition of a 'dirty circuit'
    
    A circuit becomes 'dirty' once it has serviced traffic, not when it expires.
    Thanks to Roger for the terminology correction.

commit 0a79385a89fabd14721ce3fd0e4a49c63c1c13da
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Aug 7 09:00:21 2013 -0700

    Dropping routers attribute from get_network_status_document_v2()
    
    We only have a single test using this mocking module function. The 'routers'
    attribute was both unused and unimplemented so simply dropping it. We can
    implement it later if we add a test that needs it.

commit ccf03a04fcc160a2b87fd0361e7b4c31c562f1b1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 4 16:49:14 2013 -0700

    Additions for our examples page
    
    Adding Robert's RTT Prober and updating the entry for my descriptor monitors.

commit e78f1b74efac02ed09f963c65a652b87ea80f5f2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 4 16:25:09 2013 -0700

    Having get_server_descriptor() warn when using microdescriptors
    
    Tor presently gets microdescriptors rather than server descriptors by default.
    More than one person has been confused by this, so adding a warning to our
    Controller when server descriptors are unavailable...
    
    >>> controller.get_server_descriptor('9695DFC35FFEB861329B9F1AB04C46397020CE31')
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "stem/control.py", line 1207, in get_server_descriptor
        raise ValueError(SERVER_DESCRIPTORS_UNSUPPORTED)
    ValueError: Tor is presently not configured to retrieve server descriptors. As
    of Tor version 0.2.3.25 it downloads microdescriptors instead unless you set
    'UseMicrodescriptors 0' in your torrc.
    
    This was suggested by Robert in...
    
    https://lists.torproject.org/pipermail/tor-dev/2013-July/005196.html

commit e7c165ca7122a12fd5baade33221887176de3fe7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 4 12:04:22 2013 -0700

    Adding a 'content' attribute to the Query class
    
    Including a Query attribute so the caller can get our raw descriptor content.
    I'm actually doing this to simplify the run() method's usage. It had provided
    an iterator so...
    
      query = Query(my_resource)
      print list(query)  # this would print the ist of descriptors
      print list(query)  # this would be an empty list
    
    The reason the second call is an empty list is because we already iterated over
    the query. This is confusing, especially since...
    
      query = Query(my_resource)
      query.run(True)
      print list(query)  # also an empty list
    
    ... due to run() returning a list under the covers. By making run() provide a
    fresh iterator each time I also now had the downloaded content handy so making
    it public made sense (it'll commonly be wanted by our callers).

commit 5c5c59f1e7b16ed1f7a5131ec072e5646d086e44
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 4 11:38:28 2013 -0700

    Adding a 'block' argument to the Query class
    
    I'm finding it to be pretty common to run...
    
      query = Query(my_resource)
    
      query.run(True)
    
      if not query.error:
        # do something
      else:
        # report error
    
    Adding an argument to the Query constructor so we can drop the following run().

commit faae224eb833d7612e2779e34d0f1010f8cb2d08
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 4 11:31:05 2013 -0700

    Query's run() method didn't block
    
    Well, that was confusing as hell. Say you wrote the following script...
    
      query = stem.descriptor.remote.Query(
        '/tor/server/all.z',
        timeout = 60,
      )
    
      query.run(True)
    
      if not query.error:
        print "Results downloaded: %i" % len(list(query))
      else:
        print "Got an error: %s" % query.error
    
    You'd expect run() to block, right? You should - that's what it was documented
    as doing. Instead it returned an iterator and we don't block until the
    list(query) call. This means that we would *never* report an error. This has
    been causing stacktraces for my monitors every time it encountered an error.

commit cf0a4b2f7cff17ff6f819c91e35b2a86ffcb6a81
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 4 11:21:22 2013 -0700

    Defaulting Query's fall_back_to_authority to False
    
    When a user opts to use the Query class direcly they're doing so to use
    specific endpoints. We'll probably cause more confusion then help by falling
    back to the authorities here. We'll still fall back by default in the
    downloader, though.
    
    This also includes a tweak so users can override the downloader's endpoints.

commit 539cd19919a3f5fe0e5ab39259441f736c1a8579
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 28 13:34:23 2013 -0700

    Style correction for event listing tutorial
    
    Oops, minor PEP8 correction for one of our tutorial examples. Not sure why I
    did camel case for this function...

commit b7ecd5951f1b3c1f47f4249c4d4822c00aa041d0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 28 13:15:24 2013 -0700

    Adding a couple missing changelog entries
    
    Entries for the remote descriptor fetching module and frontpage UI overhaul.

commit 7e786104dc1704d8a378918de08474c23c1d47e0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 28 13:05:07 2013 -0700

    Using a red download icon
    
    I opted for ColorCons' 'smoke' colored download icon because I thought it would
    be more subtle. However, all the other section icons are pretty bright primary
    colors so that just made it stand out more. Opting for red so it'll match the
    bug tracker icon.
    
    Also very minor alignment adjustments for the changelog and tutorial icons.

commit 8d39e338bedc1e6c67338feff5176c697e4e08bd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 27 19:39:44 2013 -0700

    Replacing buttons for our frontpage
    
    As a few people have noted our present blue buttons on stem's frontpage looks
    pretty aweful. It clashes with the site's color scheme and is pretty
    in-your-face. Swapping to another color scheme that's hopefully less
    offensive...
    
    http://www.atagar.com/transfer/stem_frontpage/before.png
    http://www.atagar.com/transfer/stem_frontpage/after.png

commit 5eda329ed6a288575f61f26ac2a457299c2f9d04
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jul 26 09:17:07 2013 -0700

    Inverted conditional for reporting bugs for password exceptions
    
    Issue caught by str4d while porting stem to I2P.

commit d7b68d01aabc6fdcd544a479cfa2cac377eaddeb
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jul 26 08:13:58 2013 -0700

    Having use_directory_mirrors() return the consensus
    
    Originally I was planning an on-disk cache so calling get_consensus() after
    use_directory_mirrors() would result in an on-disk read. However, that has been
    put on hold for now so having use_directory_mirrors() return the consensus it
    fetches in order to get the mirrors.
    
    We unfortunately already need to read the full consensus into memory before
    processing it due to how urllib2 works, so this shouldn't cause a memory usage
    hit. We might someday want to change this though if we get a constant time
    download/parsing scheme.

commit 7f050ebfa475aeb39f6b1b4a0f7ad53620931646
Merge: b6c23b0 713b046
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 21 18:39:17 2013 -0700

    Remote descriptor fetching module
    
    What started as a tor-dev@ thread has finally accumulated in a surprisingly
    simple module. This provides the ability to download tor descriptors from
    directory mirrors or authorities, and is made of up of two classes...
    
    * Query
    
      The core of the module. This is a request to download a given descriptor
      resource from a pool of authorities and/or mirrors. This retries failed
      requests and can be iterated over to get the descriptor results.
    
    * DescriptorDownloader
    
      A configurable class for issuing Queries. This provides only two things:
      simple methods to get common descriptor types, and an easy way of specifying
      default Query parameters.
    
    https://lists.torproject.org/pipermail/tor-dev/2013-May/004924.html
    https://trac.torproject.org/8257

commit 713b04632e6ca1120c6d5c2661206fd84c01904b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 21 18:37:48 2013 -0700

    Adding remote descriptor sphinx docs to site
    
    Tidying up our pydocs and including it in our site.

commit 4d122b1a417fe318f0bb3d169688d25f3e749754
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 21 17:10:26 2013 -0700

    Implementing a get_key_certificates() method
    
    Method for fetching key certificates for the authorities. This included a
    little work so parse_file() could return multiple certificates when they're
    concatenated together.

commit e95e00dd6431001000a79e2bfc009a7f3abbde80
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 21 16:31:00 2013 -0700

    Adding a query_args parameter to all downloader methods
    
    Simple tweak in case callers want to do something dynamic (such as varying the
    timeout with each request).

commit 6f12acc9e54842371f52203550a6ef2f9cbfb835
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 21 16:25:58 2013 -0700

    Simplifying DescriptorDownloader class
    
    The DescriptorDownloader is a very, very simple class that simply issues
    Queries on your behalf. Why should it pretend to be anything else?
    
    Dropping most of its attributes (and with it a lot of copy-and-pasted pydocs).
    Also making the query() method public.

commit 90ec648ed075245f6ed2bae42dd4c6968e15fab1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 21 15:57:47 2013 -0700

    Support for compressed descriptors
    
    Ummm... well, that was easy. When I was initially looking into tor's directory
    responses I had trouble getting valid compressed results. Kostas suspected
    that the responses might have been truncated. Oh well, maybe a wget thing. In
    either case adding compression support to our module was trivial.

commit d5b3ec93f44de01b21b27264e761fe8f09ec8012
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 21 10:45:56 2013 -0700

    Fixing "cannot unmarshal code" errors
    
    Damnit python, your import scheme is stupidly confusing.
    
    The descriptor's __init__ module has a circular dependency with its contents.
    This is because the parse_file() function calls the constituent modules, while
    those modules need the Descriptor class from __init__.
    
    So far so good. Only trouble is that python's support for circular dependencies
    sucks. To address this I did lazy imports in __init__, so we imported within
    the parse_file() function.
    
    On the surface this seemed to work. All the tests certainly passed. The trouble
    is that this style of python import is buggy as hell. Turns out that lazy
    imports leave the module in question in a unexecutable state so this *only*
    works if you've also imported the module another time during the interpretor
    execution. Our tests did this, hence passing tests.
    
    I first encuntered "cannot unmarshal code" while writing the remote descritpor
    tests (both unit and integ). I was content to hack around this with superfluous
    import statements while this only manifested within the tests, but now I'm
    seeing it during general usage too...
    
    >>> from stem.descriptor.remote import DescriptorDownloader
    >>> list(DescriptorDownloader().get_microdescriptors('jzcx+1fHsi47Tu+vQIcyItgn4lKs6aKnFshQ0lZ2JTg'))
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "stem/descriptor/remote.py", line 311, in __iter__
        for desc in self.run(True):
      File "stem/descriptor/remote.py", line 300, in run
        for desc in self._results:
      File "stem/descriptor/__init__.py", line 154, in parse_file
        import stem.descriptor.server_descriptor
    RuntimeError: cannot unmarshal code objects in restricted execution mode
    
    Joy. After much head scratching and forum reading it sounds like there's
    something magical about 'from' imports so switching the descriptor modules to
    that, and moving the __init__ imports to the end. I'm not entirely clear on the
    magic going on here, but its elmiminated the errors.

commit 60ad230d88b383a259c531649fdba78d0acd6493
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 21 09:03:39 2013 -0700

    Expanding remote descriptor pydocs
    
    Adding documentation and examples for the remote descriptor API. This includes
    some minor API tweaks and a missing import statement.

commit 25951d4f12a16a1fa6e42ea6b7479959d636139b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 20 10:00:04 2013 -0700

    Making descriptor validation optional
    
    Adding a 'validate' argument that can be passed thorough when parsing
    descriptors.

commit 749c8d37fd54ea6e41dfca04495c35b47a755f6a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 20 09:52:07 2013 -0700

    Adding a broken get_microdescriptors() method
    
    Huh. Not really sure why this isn't working. Adding a get_microdescriptors()
    for how the spec seems to indicate that the method works but tor gives an empty
    reply.
    
    % curl http://217.13.197.5:9030/tor/micro/d/6dCl6ab8CLo0LeMjxi/MZgVJiZgWN8WKTesWPBMtyTo
    
    Checking with karsten on irc t see if he knows what's up, and moving on to
    another part.

commit 4005a88becb68858ec4742455ea4d80e1ff9f827
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jul 19 10:12:39 2013 -0700

    Implementing a use_directory_mirrors() method
    
    Method so we can balance load against directory mirrors rather than hammering
    the authorities. We can either request this during construction (which fails
    silently) or call this method explicitly (to get the exception).

commit 5514b2cfd7d313f429bfe1bb8a13c975f7c3413f
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jul 18 09:59:30 2013 -0700

    Implementing a get_consensus() method
    
    Originally this was gonna be a get_network_status(fingerprint) method but
    evedently we can't request individual router status entries. Understandable
    since signatures are for the whole document but still a bit of a pity from an
    API perspective. Oh well.

commit a014b9b967e8b825a9a4c99514ce2098ab6ba614
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jul 17 09:58:35 2013 -0700

    Implementing get_extrainfo_descriptors() method
    
    Very similar to the get_server_descriptors() counterpart.

commit 871a957f23097807f67e502fd6e2c9a9b9a6c456
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jul 16 09:28:18 2013 -0700

    Rejecting requests for more than 96 descriptors
    
    In our tor-dev@ discussion Karsten mentioned that we could request at most 96
    descriptors at a time when polling by their fingerprints...
    
    https://lists.torproject.org/pipermail/tor-dev/2013-June/005005.html
    
    I've emailed him to clarify where this limitation comes from (the url length or
    something within tor?), but in the meantime adding a check for this to our
    get_sever_descriptors() method.

commit ac694d3a98ce1e56c674893f8006efe785eb41c9
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jul 16 09:16:06 2013 -0700

    Implementing get_server_descriptors() method
    
    Implementing our first downloader method and added an ONLINE integ test.
    Initially I did a get_microdescriptors() method but turns out that tor didn't
    implement what we need...
    
      https://trac.torproject.org/9271

commit 7d04653d908f0c62e197bafa27a5cd94634cbb53
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jul 15 09:16:54 2013 -0700

    Moving retry functionality to Query class
    
    It's far better if the Query class handles retries rather than the
    DescriptorDownloader. Advantages include...
    
    * The DescriptorDownloader no longer has a reason to support blocking queries.
      There's no advantage to them since the Query knows if/when the request fails
      (and hence can retry it on our behalf).
    
    * The Query class is easier to test. The more functionality we can push down
      into it the simpler the DescriptorDownloader will be.
    
    * More advanced use cases will be using Query instances rather than
      DescriptorDownloader. By having retry functionality there it'll be easily
      available to them.

commit 0b1c13553f842088fd5f5ae21600d28e25e6b633
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jul 15 08:43:41 2013 -0700

    Initial pass at the DescriptorDownloader
    
    Implementation of the DescriptorDownloader's _query() helper method which will
    be the basis of its other methods. On reflection I should move retry
    fruntionality and probably caching down to the Query class.

commit d87f35c7c0f99d52a530cdf12d03ef6abd9cadb0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 14 14:06:12 2013 -0700

    Performing integ test queries in parallel
    
    What good is designing an asynchronous descriptor fetching API if you poll
    things serially? Trivial tweak to reduce this test's runtime from 2.4s to
    0.4s.

commit dc6632b5afba37111b9b2d6fb0c316cca1d9e046
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 14 11:44:57 2013 -0700

    Integ test to exercise each of the authorities
    
    Adding an integration test that fetches moria1's descriptor from each of the
    directory authorities. This obviously requires network activity so it's run via
    the ONLINE target.
    
    This uncovered a couple interesting finds...
    
    * This is our first time actually using urllib2 (it was mocked previously), and
      its 'file like object' is incomplete. It lacks tell() and seek() which means
      that our parsers don't have random access. As such we need to read this all
      into memory and use a BytesIO.... damn.
    
    * The moria1 authority has an extra newline in its descriptor content. Why?
      Haven't a clue. Maybe it's running a different version of tor or maybe this
      is an artifact of fetching its own descriptor. Whatever it is, it's the only
      authority to do so. Simply stripping the content to avoid having our parser
      complain about this.

commit 61f00819dc3f18b17af0d8d9d3f4f90791364e39
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 14 11:07:40 2013 -0700

    Dropping Tonga from remote descriptor fetching
    
    Karsten reports that we shouldn't use authorities that lack 'v3ident'. At
    present that's just Tonga.

commit 762b7239e4866ba1b98212259e9751723e0865df
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 6 10:54:58 2013 -0700

    Workaround for screwy RuntimeError related to mock causing imports
    
    The following isn't used by this directly, but we're still importing it due
    to a screwy aspect of how mock works. If patched() results in an import that
    we haven't done before then we can fail with...
    
      ======================================================================
      ERROR: test_query_download
      ----------------------------------------------------------------------
      Traceback (most recent call last):
        File "/usr/local/lib/python2.7/dist-packages/mock.py", line 1201, in patched
          return func(*args, **keywargs)
        ...
        File "/home/atagar/Desktop/stem/stem/descriptor/__init__.py", line 156, in parse_file
          import stem.descriptor.networkstatus
      RuntimeError: cannot unmarshal code objects in restricted execution mode
    
    I've only seen this arise when running this test directly (without the rest
    of the unit tests, some of which import networkstatus on their own).

commit 762f0ad396dd752f0f807871e0715c142cdbf475
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jul 5 19:18:02 2013 -0700

    Class for issuing remote descriptor queries
    
    Adding a Query class which does most of the heavy lifting for downloading
    descriptors.

commit 65924aff93ef70c879175510e9cd2307e8507a9e
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jun 20 08:57:36 2013 -0700

    Stub for a remote descriptor fetching module
    
    Starting a module for remote descriptor fetching. Presently this just includes
    a no-op DescriptorDownloader class. Guts of this sucker are coming soon!

commit b6c23b0c2934b6c1b1c51530754cc48f6ef4e38e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 7 13:35:50 2013 -0700

    Typo in descriptor reader pydocs
    
    Correction from Roger. Honestly I had to look this one up. Both 'run' and 'ran'
    sound right when I read them to myself. Oh well, grammar is most certainly not
    my strong suit.

commit 5cd89263f0acc5a7b6fcdd4021efa09a4f1634ab
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jul 4 13:41:22 2013 -0700

    Pyflakes and PEP8 corrections
    
    After upgrading both they each complained about a few new things. Mostly
    pyflakes getting confused by pydoc comments, but there were a couple fair
    catches.

commit d0ce002d568627876c59c3ff23d6d86fbb049aa9
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jul 4 13:34:56 2013 -0700

    Checking for testing prereq of mock 0.8.0
    
    We had a requirement on mock 0.7.0 but it turns out that we're using a 0.8.0
    feature too. Spotted this while trying to use ubuntu jaunty's apt-get
    version...
    
    Traceback (most recent call last):
      File "./run_tests.py", line 405, in <module>
        main()
      File "./run_tests.py", line 184, in main
        for test_class in test.util.get_unit_tests(args.test_prefix):
      File "/home/atagar/Desktop/stem/test/util.py", line 130, in _get_tests
        module = __import__(module_name)
      File "/home/atagar/Desktop/stem/test/unit/tutorial.py", line 23, in <module>
        class TestTutorial(unittest.TestCase):
      File "/home/atagar/Desktop/stem/test/unit/tutorial.py", line 24, in TestTutorial
        @patch('sys.stdout', new_callable = StringIO.StringIO)
    TypeError: patch() got an unexpected keyword argument 'new_callable'

commit 8662332c844827b2be4ab7ab1c77027738e5723d
Author: David Triendl <david@triendl.name>
Date:   Sat Jun 22 09:03:50 2013 +0200

    Fix parsing of IPv6 address:port pairs in stream events (fixes #9181)

commit 00d78f244bda322f30ecb5d31f3278eeeaa96056
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 30 10:46:11 2013 -0700

    Adding Slackware to our download page
    
    Markus just added stem to slackbuilds (hazaa!)...
    
    http://slackbuilds.org/repository/14.0/python/stem/

commit 003fa8e9e4b38304ccbb28e75a3f646ab6ba1ab2
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 25 08:57:42 2013 -0700

    Switching get_network_status() to provide v3 entries
    
    Turns out 'GETINFO ns/*' provides v3 directory information rather than the v2
    documented in the spec. Now that the spec is fixed swapping us over to take
    advantage of it...
    
    https://trac.torproject.org/7953
    https://gitweb.torproject.org/torspec.git/commitdiff/d2b7ebb
    
    RouterStatusEntryV3 provides a superset of the attributes of
    RouterStatusEntryV2 so backward compatability shouldn't be a concern.

commit 5f65f786dee22089323febc899bf2c85a53c99aa
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jun 24 09:23:16 2013 -0700

    Clarifying why test's load_conf() expects an exception
    
    For a long while I've been puzzled why this load_conf() call asserted an
    exception. When I ran it in a cusom prompt it ran fine - turned out that it was
    relying on us having a custom DataDir. Tor balks (rightfully) when we'd shift
    our DataDir out from a running tor instance.
    
    Doing a better job explaining what's up in the comment.

commit 042fcfce009e5abc86f066c8ef0788cef871998f
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jun 21 09:27:45 2013 -0700

    Test runner tried to use non-existant print_error() function
    
    In commit e751397 we removed the test.output.print_error() function in favor of
    something less verbose. Somehow we missed its usage in test.runner - weird that
    pyflakes didn't catch this...
    
    Traceback (most recent call last):
      File "./run_tests.py", line 405, in <module>
        main()
      File "./run_tests.py", line 212, in main
        integ_runner.start(target, args.attribute_targets, args.tor_path, extra_torrc_opts = test.util.get_torrc_entries(target))
      File "/home/atagar/Desktop/stem/test/runner.py", line 310, in start
        self._start_tor(tor_cmd)
      File "/home/atagar/Desktop/stem/test/runner.py", line 718, in _start_tor
        test.output.print_error("  failed to start tor: %s\n" % exc)
    AttributeError: 'module' object has no attribute 'print_error'
    
    Also, the runner had a couple calls of...
    
    test.runner.get_runner()
    
    Those should obviously just be 'get_runner()', and only worked because of our
    test.output import (damn python's transitive imports are stupidly confusing).

commit 456bc64f2fbd41918dd2db711eaf259a90f200e3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 16 14:44:30 2013 -0700

    FAQ entry about how to reload your torrc

commit d921dfdfe2f085cd7de986ed16956b2a7361e32a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 16 12:50:50 2013 -0700

    Renaming FAQ categories
    
    Trivial change to rename 'Usage' to 'General Information', and 'Tasks' to
    'Usage'.

commit e8a278cfed828be68d83486deefbff86fb548b0d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 15 20:40:01 2013 -0700

    FAQ entry about how to get information about your present exits
    
    Answer for a question I like on Stack Overflow...
    
    http://stackoverflow.com/questions/9777192/how-do-i-get-the-tor-exit-node-ip-address-over-the-control-port

commit 69ff2a6eec5b99a7b13a57f09de2e2abbd87b91d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 15 19:26:15 2013 -0700

    FAQ entry for how to request a new tor identity
    
    A common question on Stack Overflow reguarding tor goes along the lines of 'how
    do I get a new identity?'. For instance...
    
    http://stackoverflow.com/questions/17013707/how-can-i-change-my-tor-process-endpoint-in-stem
    http://stackoverflow.com/questions/16987518/how-to-request-new-tor-identity-in-terminal
    
    Tor doesn't support cycling your IP, so adding a FAQ entry explaining why and
    telling them how NEWNYM works.

commit 792557ccdb475cdd95be09b71a69d38f7770b3c7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 15 15:45:02 2013 -0700

    Requiring mock version 0.7.0 or later
    
    Our jenkins tests presently have mock version 0.6.0, but we're using mock's
    patch.dict which was introduced in 0.7.0. This causes the tests to fail with...
    
    Traceback (most recent call last):
      File "./run_tests.py", line 394, in <module>
        main()
      File "./run_tests.py", line 173, in main
        for test_class in test.util.get_unit_tests(args.test_prefix):
      File "/srv/jenkins-workspace/workspace/stem-tor-ci/test/util.py", line 130, in _get_tests
        module = __import__(module_name)
      File "/srv/jenkins-workspace/workspace/stem-tor-ci/test/unit/version.py", line 20, in <module>
        class TestVersion(unittest.TestCase):
      File "/srv/jenkins-workspace/workspace/stem-tor-ci/test/unit/version.py", line 22, in TestVersion
        @patch.dict(stem.version.VERSION_CACHE)
    
    Making run_tests.py balk if we're using a version of mock prior to 0.7.0.

commit a5dc68b06d049df62f29f308f2029bdbd4521507
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 15 15:16:28 2013 -0700

    Warning that the twitter example is broken
    
    Twitter has deprecated the 1.0 version of their API, which we used on...
    
    https://stem.torproject.org/tutorials/to_russia_with_love.html#reading-twitter
    
    Warning our users about this. The issue is being tracked on...
    
    https://trac.torproject.org/9003

commit 101cf0b71bab55d72a831eff0236dcef5edbff0f
Merge: 65846e8 2b634a9
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jun 13 09:48:30 2013 -0700

    Migrating to the mock library
    
    Our homemade mocking framework has served us well, but over time it's taught me
    one very important lesson: writing a mocking framework is hard. On the surface
    it seems pretty simple: apply and revert a set of monkey patches. But how do
    you monkey patch class methods? What about alias imports like the os module?
    And god forbid you want to mock python's open() function.
    
    I'm finally taking a lesson from one of my coworkers and using a library for
    this. Python has several options but the most common is PyPI's mock module,
    which became part of the standard library in Python 3.3...
    
    http://www.voidspace.org.uk/python/mock/

commit 2b634a98e16c3b2437cab27c7198acd6a738fd13
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jun 13 09:30:07 2013 -0700

    Accounting for GetInfoResponse entries having bytes
    
    Oops. I forgot to update this test when we switched the entries attribute to
    bytes in commit a94f1f0.

commit c7ac81546dcdf1fb9639fdbf0e4997e8ee39ba38
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 12 21:59:41 2013 -0700

    Python3 mock compatability
    
    Ahhh, more unicode vs bytes joy.

commit 35b44e3c7d5090806922cd0775775bb2985f8a9b
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 12 21:48:44 2013 -0700

    Dropping homemade mocking functions
    
    Done replacing our mocking methods with the mock module. It's nice to finally
    be able to drop these...

commit da365bc6ea5312d467c33f14942695c66c6f4e10
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 12 21:39:25 2013 -0700

    Using mock for system integ tests
    
    Unfortunately I'm not spotting a nice method via the mock module to get the
    original function. The patch() function has a get_original() method which looks
    promissing, but I didn't spot a method for accessing it. Oh well...

commit f38959c05a5dbedcf4a90ed5fc6c1f9809ca5d62
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 12 19:53:35 2013 -0700

    Using mock for protocolinfo integ tests

commit 98fd433b09c2449e5ea6dc6d381e729ba32a6099
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 12 19:44:34 2013 -0700

    Dropping mock from server descriptor integ test
    
    Well, that's odd. We revert mocking in these integ tests but never use it. Oh
    well.

commit 21cd260745a1b824c908e3eacac4083fe16750a6
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 12 19:42:48 2013 -0700

    Dropping mock from descriptor reader integ test
    
    I was gonna migrate this call over but it looks to be unneeded. Test passes so
    guess we probably replaced the problematic descriptor content with a real
    descriptor.

commit 625472df56748195c5b0474209278dd57a66c338
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 12 19:38:11 2013 -0700

    Using mock for process integ tests

commit 752a957698550afc112ded236e3a7cceea500245
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 12 19:29:23 2013 -0700

    Using mock for descriptor export unit tests

commit 320d3ba383ab32ec4cecfc4158e9c7336df11c1e
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 12 19:26:06 2013 -0700

    Using mock for server descriptor unit tests

commit 6c0af9db6169b7edb4eeb83f3555b9e5d6cc2d6e
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 12 09:32:37 2013 -0700

    Using mock for descriptor reader unit tests
    
    We get to replace some icky open() mocks. Yay!

commit 33eeb9bbcc4a877d02f227d9f9b76a4631f3053e
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 12 09:14:43 2013 -0700

    Using mock for system unit tests
    
    Pretty much a drop-in replacement for these.

commit 298a08c4b5488bf756c4dbce75ee0ebf4eeb2b0b
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 11 09:27:08 2013 -0700

    Dropping tempfile usage from controller unit test
    
    Now that I have a better understanding of the magic behind mocking open,
    there's no need for us to make a temporary file in our unit tests.

commit 8e90c1029e650536cf07ee42ddeb06300f0aa002
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 11 09:09:05 2013 -0700

    Using mock for proc unit tests
    
    Now that I have a rough understanding of the open mock it's quite a bit nicer
    than our prior version (which got especially icky for python 3 compatability).

commit 8303c8d99ac6728367dabec3f6b2c9e43bb68c69
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 11 08:43:05 2013 -0700

    Using mock for controller unit tests
    
    Another heavy mock user. Nice improvements, I'm really liking this library!

commit 5089ded80aaa4fd5bc260c2b7fb5bf03e69b2133
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jun 10 08:50:44 2013 -0700

    Using mock for protocolinfo unit tests
    
    Oooh, I like this change.

commit f79278eec35de5520158923563256c45aa00f066
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jun 10 08:42:16 2013 -0700

    Using mock for event unit tests
    
    We don't really need a mock in this test, guess we do so to be more faithful to
    the example.

commit 1007a0a2d1963f350444c3b762cab6be8bda6eda
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jun 10 08:30:16 2013 -0700

    Using mock for authentication unit tests
    
    Mostly sorting out another thorny test that exhaustively exercises all
    authentication use cases.

commit 9c0f1d37c0febbaacd2b6b73a2d094d3583869ba
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 9 18:20:49 2013 -0700

    Using mock for tutorial unit tests
    
    The unit tests for our tutorials are some of the ugliest we have since they
    make heavy use of mocks. Hence a good next step for mock.
    
    The mock module makes the controller mocking considerably nicer. That said,
    mock's open() mocking is crap. Their mock_open() docs...
    
    http://www.voidspace.org.uk/python/mock/helpers.html#mock.mock_open
    
    Tripped me up for well over an hour because their examples are hardcoded for
    the __main__ namespace (so it works when I do it in an interpretor, but not the
    tests). Even with that fixed the mock object it returns is buggy as hell (it's
    documented as supporting readlines() but that's a lie). They'd get a lot more
    mileage if they used an io.BytesIO instead.

commit e50a62fd56aec4f5b9f96bf0828338356cc903e9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 9 14:28:43 2013 -0700

    Using mock for tutorial unit test stdout
    
    Mock's patch() docs include an example for mocking stdout. Using it for our
    tutorial unit tests...
    
    http://www.voidspace.org.uk/python/mock/patch.html#mock.patch

commit 9109b6c078496c56b8ba1b85b0932b405be7d953
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 9 14:10:01 2013 -0700

    Using mock for version unit tests
    
    First replacement of our mocking module with mock. This is actually both quite
    a bit cleaner, and also checks that our caching is working.

commit 606ccfbb5001d5a13bce8b7a51cea283c074f3c1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 9 12:14:56 2013 -0700

    Adding the mock module as a test dependency
    
    Mocking is hard. We've gotten a lot of mileage by writing our own mocking
    helpers, but they're clunky and complicated. It's about time for us to use a
    more standard, mature option.
    
    The pypi mock module is not only well documented, but looks to be the direction
    the python world is going. As of python 3.3 it's now part of the standard
    library.
    
    As the first step of swapping us over I'm adding it as a dependency for
    run_tests.py. This warns the user if it's unavailable...
    
      atagar@morrigan:~/Desktop/stem$ ./run_tests.py --unit
      To run stem's tests you'll need mock...
    
      https://pypi.python.org/pypi/mock/
    
      You can get it by running 'sudo pip install mock'.

commit 65846e81628040779e4d28041d1400f53719468c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 8 13:22:12 2013 -0700

    Correcting Controller pydoc links
    
    Handful of mistakes in the stem.control module that prevented sphinx's internal
    linking. Pity sphinx doesn't provide warnings about this...

commit 2257f7954a0d9dba46905a4813b828fcc0b52086
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu May 30 08:43:10 2013 -0700

    Minor fixes for ONLINE testing fixes
    
    Just some minor tweaks to fix the style checker and s/range()/xrange().

commit b5273658f6470a643b70b7ef96654248788eef2f
Author: Jeremy <saturn5971@gmail.com>
Date:   Wed May 29 23:53:11 2013 -0400

    Made attachstream and mapaddress tests more robust
    
    Modified the attachstream and mapaddress integ tests so that they don't
    fail as easily in the face of minor network failures unrelated to stem
    (like temporary issues building a circuit, or an exit node having a
    different IP from the one Tor reports).
    
    This is to address...
    
    https://trac.torproject.org/8692

commit d7d14a82e7d28ff863c71b4b027c9922e17a6661
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 29 20:44:23 2013 -0700

    Methods to query user tor is running as
    
    Couple additions to make it easier to determine the user tor is running as:
    
    * stem.util.system.get_user(pid) => determines the user a process is running as
    * Controller.get_user() => provides the user that tor is running as

commit f5bdd944df426410fe6b21088280cf3779709ed1
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 29 08:53:12 2013 -0700

    Checking is_caching_enabled() when using cache
    
    Adding a is_caching_enabled() call to our getter/setter for the Controller's
    cache. This has a couple advantages...
    
    * It prevents us from accidently caching stuff when the cache should be
      disabled. In at least one point we were using the cache without checking
      is_caching_enabled(), which was probably an error.
    
    * Lets us simplify our code a bit by making is_caching_enabled() optional for
      the cache getter/setter callers.

commit 341091492edf7c060609b73b8619b18eda322ebd
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 29 08:44:44 2013 -0700

    Supporting get_pid() remotely via GETINFO
    
    One of get_pid()'s resolution methods (GETINFO process/pid) works remotely, so
    attempting it before checking that tor is running locally.

commit 31456707624847f8b0404b2e84ebcc2dc7fe22fd
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 29 08:19:54 2013 -0700

    System module's call() function ignored exit statuses
    
    Well... oops. Our call() method was documented as raising an OSError (or
    returning the default value) when the call fails but it completely ignored the
    exit status. In practice the function only behaved as documented when Popen
    failed - otherwise it returned an empty list. Fixing this behavior and updating
    most of our callers since they expected call() to not raise any exceptions.

commit 18aa77f4d6074e54db84f93e630aaacd5e3fbcb3
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue May 28 17:32:11 2013 -0700

    System utility for getting FreeBSD jail paths
    
    Adding a util to better support FreeBSD jails (a rather common way to run tor
    on that platform).

commit 24d9377cdc2d1ee3531238b825621934053bae9f
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue May 28 09:32:13 2013 -0700

    System utility for getting process start times
    
    Adding a get_start_time() to the system utils to provide the unix timestamp for
    when a given pid started.

commit 292ef5c36e38dd2820ed00e34ab270f2bb28ecb6
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 27 21:28:55 2013 -0700

    Performing pid resolion via getinfo when able
    
    Oops, forgot that I had added a 'GETINFO process/pid' option to tor for just
    this case. Using it when it's available.

commit 0f7d5d118d83db000dd57b646ba91d4152c44e91
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 27 21:20:42 2013 -0700

    Controller method to query tor's pid
    
    Adding a get_pid() method to the controller to make it simpler to figure out
    its pid. This attempts resolution via the PidFile, process name, control port,
    and control socket file.

commit 0efbdf80f94574805b0fa080965bb509d535a4db
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 26 21:27:43 2013 -0700

    Allowing ControlMessage.from_str() to include conversion
    
    For txtorcon or anything else using our ControlMessage.from_str() function the
    following is very, very common...
    
    msg = stem.response.ControlMessage.from_str(getinfo_content)
    stem.response.convert("GETINFO", msg)
    
    Adding a 'msg_type' argument to from_str() so the two calls can be combined.
    This addresses...
    
    https://trac.torproject.org/8605

commit 7b396bdcebe7165df6ea6aaa911707e74974fa38
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 26 18:14:15 2013 -0700

    Dropping validation on bandwidth-weights contents
    
    Our validation of network status documents included checking for the
    bandwidth-weights specified in the spec. This is because, according to the
    spec, they were all mandatory and new weights could not be added. Tor has
    loosened this restriction...
    
    https://gitweb.torproject.org/torspec.git/commitdiff/ec3eaca
    https://trac.torproject.org/6872

commit 08c4d273c67e1bb4d273ba5de188703d27a0568c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 25 20:57:12 2013 -0700

    Noting additions in the change log
    
    Adding items from our commit history I forgot to include with the change log
    earlier.

commit 03f4313b04285745f0c46cd83ac26fcc57bf0a6f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 25 20:26:11 2013 -0700

    Adding Ubuntu to the download page
    
    Stem is now on Launchpad, evidently pulled from Debian for the Saucy Salamander
    release of Ubuntu.

commit ad82c905347f3e5849930fdb3e63721b294d38b4
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue May 21 21:30:49 2013 -0700

    Processing controller descriptors as bytes
    
    When descriptor content was fetched by the Controller it was first converted to
    unicode, then back into bytes. As a result descriptors with content that
    couldn't be converted (usually on the contact line) got mangled, and this in
    turn caused our check of its signature to fail.
    
    Caught thanks to aj00200 on...
    
    https://trac.torproject.org/8755

commit a94f1f0a51faacf1fda5350cdfd489d45b41adc4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 20 09:35:45 2013 -0700

    Support for getting byte content from get_info
    
    Adding an optional flag to get byte content from get_info rather than a str.
    This only effects python 3.x.

commit 0362564cc8e1d48ebf7b61c75550fa9e1c4fd321
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 19 17:29:43 2013 -0700

    Normalize ControlMessage content to be bytes
    
    The socket module's recv_message function converted read content to be a str
    (bytes in python 2.x and unicode in python 3.x). In 99.9% of the cases this is
    exactly what we want, but in a few edge cases we actually want byte content
    under python 3.x.
    
    For instance, in #8755 descriptors with non-unicode contact lines are being
    altered by the socket module, causing our validation of their signature to
    fail.
    
    Functionally this commit should have no effect besides adding the option to get
    byte content from the ControlMessage. The tests seem happy so fingers crossed!

commit 525f2f9cee559fb8b77cb8daf46c68737487c63b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 19 14:59:41 2013 -0700

    Using a real descriptor for test_non_ascii_descriptor
    
    Our test_non_ascii_descriptor test used an altered descriptor with katakana
    (Japanese characters) in its contact line. When we use doctored descriptors we
    can no longer exercise _validate_content since the signature won't match.
    Replacing the example descriptor with the one that's causing issues in #8755.

commit ba50df27f2b69e05ce3470c4c1127974420a93bc
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 19 14:11:39 2013 -0700

    Mocking's sign_descriptor_content() incompatible with python 3.x
    
    I didn't have a python 3.x copy of pycrypto installed, so there were some code
    paths I didn't previously encounter. Unit tests failed due to
    sign_descriptor_content() not explicitly using bytes at several points.
    
    ======================================================================
    ERROR: test_mirror_mirror_on_the_wall_4
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/tutorial.py", line 169, in test_mirror_mirror_on_the_wall_4
        exit_descriptor = mocking.sign_descriptor_content(exit_descriptor)
      File "/home/atagar/Desktop/stem/test/data/python3/test/mocking.py", line 984, in sign_descriptor_content
        public_key_string = public_key_string[:64] + "\n" + public_key_string[64:128] + "\n" + public_key_string[128:]
    TypeError: can't concat bytes to str
    
    ----------------------------------------------------------------------

commit b816137fc07d139808dd80a410eba04fd662fdba
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 19 12:52:30 2013 -0700

    Minor simplifications for dependency module
    
    When we dropped python 2.5 support our _check_version() only had a single
    caller, making it not a very helpful helper. We were also missing pydocs for
    is_crypto_available().

commit b626f0c5156e8f5d2a89558ce54890b931648758
Author: Jeremy <saturn5971@gmail.com>
Date:   Sat May 18 22:08:57 2013 -0400

    Revise approach to owning Tor in integ tests
    
    The integ tests now only try to own the Tor process via control
    connection if controllers have a way to access Tor. This is needed to
    successfully run under the RUN_NONE target.

commit 29086a71cd508614d247f70888df884028d51185
Author: Jeremy <saturn5971@gmail.com>
Date:   Thu May 16 23:23:35 2013 -0400

    Asserting ownership over the integ tests
    
    Ticket #8634. Integ tests now assert ownership over the tor process, but avoid
    accidentally terminating tor mid-test.

commit 13dc7861ae7bfa280fda8536ca39fcf6135d7b56
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu May 16 09:20:11 2013 -0700

    Integ tests resulted in temporary directories
    
    Our integ test runner created temporary directories for our control socket but
    failed to clean them up. On reflection it's simpler for us to use a UUID rather
    than tempfile.mkdtemp() here so our start() and stop() methods can perform the
    creation/cleanup.
    
    Tested by running our integ tests with a control socket...
    
    ./run_tests.py --integ --target RUN_SOCKET
    
    ... and confirming both that a temporary resource is made with our control
    socket and that it's cleaned up afterward.
    
    Issue caught thanks to help from Ashish on...
    
    https://trac.torproject.org/8622

commit ffd7274c8bc97557d4027257b958b388911cb429
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue May 14 08:30:54 2013 -0700

    Re-enabling test_event_handling
    
    Oops, I confused this test with another. I bumped the timeouts on
    test_reattaching_listeners rather than this...
    
    https://gitweb.torproject.org/stem.git/commitdiff/a5f8168
    
    Re-enabling the test with similar timeouts to see if jenkins is still happy.
    However, this is definitely still code stink indicating that we need to do
    something smarter for these tests.

commit 73b5ed39da299109c8515bdbd6c18397a18a9634
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 13 09:59:42 2013 -0700

    Client usage example for polling twitter
    
    Adapting a script from Ashish for polling a twitter feed as another tutorial
    example...
    
    https://pastee.org/ghbhp

commit 1b965a4c81d0734d7f94a0283dbdd4b8793c4fa3
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 13 09:31:33 2013 -0700

    Making SocksiPy work with official version
    
    Ashish's SocksiPy was based on a fork of the main project...
    
    https://code.google.com/p/socksipy-branch/
    
    This caused my attempts to run it to fail with...
    
    Traceback (most recent call last):
      File "twitter_script", line 12, in <module>
        socks.wrapmodule(urllib)
    AttributeError: 'module' object has no attribute 'wrapmodule'
    
    Making this work with the official upstream project is trivial so making the
    tweak.

commit a72e22512e86653d86ec2015bbf3416a3de52aae
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 13 08:45:51 2013 -0700

    Using SocksiPy for client usage tutorial
    
    Exemplifying using tor with both SocksiPy and PycURL. It's quite a bit simpler
    with SocksiPy so I'll probably use this for most other examples. Change is
    thanks to Ashish.

commit c4b47fc89582dc96b75c90d8eb7629d88304b06d
Merge: 76f82d7 e96440e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 11 19:37:48 2013 -0700

    Controller cache thread safety
    
    Making our caching operations thread safe. Patch thanks to Akshit...
    
    https://trac.torproject.org/8607

commit e96440eda20e1bbd89ed542654b4f080b8ecb064
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 11 18:36:44 2013 -0700

    Revisions for cache thread safety
    
    Handful of changes for the prior couple commits. Most are stylistic, but
    there's a couple funcitonal ones:
    
    * _set_cache() could accidently remove 'exit_policy'
    * _confchanged_listener didn't use the new thread safe operations

commit f208da6d4ca07ee4d241bd67085ba331bcc07d4b
Author: Akshit Khurana <axitkhurana@gmail.com>
Date:   Sun May 12 03:03:50 2013 +0530

    Add function for single param cache query
    
    Separate functions for single param cache query and multiple params.
     * get_conf : takes single param, returns value
     * get_conf_map: takes list of params, return dict

commit 2c51c6a10cef7b68d9acc7703014bea2e4bd3101
Author: Akshit Khurana <axitkhurana@gmail.com>
Date:   Thu Apr 4 18:09:06 2013 +0530

    Make controller cache thread safe
    
    reads and writes take place under cache lock
    https://trac.torproject.org/8607

commit 76f82d74960df9bccececff6792aa7dab88cd77b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 11 17:30:29 2013 -0700

    Allowing for instantanious closure in test_status_notifications
    
    Well, that's a little odd. Our jenkins instance reported a failure of
    test_status_notifications...
    
    ======================================================================
    FAIL: test_status_notifications
    ----------------------------------------------------------------------
    Traceback:
      File "/srv/jenkins-workspace/workspace/stem-tor-ci/test/integ/control/base_controller.py", line 224, in test_status_notifications
        self.assertTrue(state_observer.timestamp < time.time())
    AssertionError
    
    ----------------------------------------------------------------------
    
    In the past we've seen test failures due to dixie being far, far faster than my
    netbook. I'm a little dubious that state_observer.timestamp truely equaled
    time.time() (that would imply that the above statements took zero nanoseconds)
    but oh well. I've only seen this happen once and there's little reason not to
    pass when they're equal.

commit 2f69e4de1e5d9982dfe94a90c62da3c71edf96a2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 11 17:21:13 2013 -0700

    Disabling test_event_handling
    
    Well damn. Even with expanded sleeps test_event_handling is occasionally
    failing on the jenkins host. No test is better than a flaky test so disabling
    this for the time being.
    
    Ticket for fixing this test...
    
    https://trac.torproject.org/8865

commit 7c0b3ca25596123b5cfbc9251bdfbe630d23ffed
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 11 17:09:52 2013 -0700

    Support for pulling multiple pids with get_pid_by_name()
    
    A leading cause of test failures in our jenkins environment is
    test_get_pid_by_name_lsof. This is because the test assumes that there is only
    a single tor process, but that's often not the case on the dixie host.
    
    This not only fixes the test, but expands get_pid_by_name() to include a
    'multiple' flag to pull all pids for a given process name. This includes unit
    test coverage.

commit bbd736d71e25db09af6b6be05a4a33ef2ff17465
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu May 9 08:47:24 2013 -0700

    Download page entry for Fedora
    
    Juan's package is now live so dropping our Red Hat entry in favor of Fedora
    (where the package is actually available).

commit 76909729f3169af14d194a2f57e78e1604af7132
Author: Tomasz Kunikowski <tomasz.kunikowski@gmail.com>
Date:   Mon May 6 19:36:34 2013 +0200

    Fixing DescriptorReader links in 'Mirror Mirror...' tutorial
    
    Links to DescriptorReader class were pointing to a wrong location.

commit de576cad4bbdd4901ee1857334a7420593ef2458
Author: Sreenatha Bhatlapenumarthi <sreenatha.dev@gmail.com>
Date:   Thu Apr 25 06:36:27 2013 +0530

    Updating GETCONF cache after CONF_CHANGED events
    
    https://trac.torproject.org/7713

commit f7d3a232a40ec93fec8c64f46fefba71a7fe9648
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 5 18:22:56 2013 -0700

    Handle relative targets in the DescriptorReader
    
    The _handle_file() funciton expects to receive absolute paths so expanding
    relative inputs. This is a slight modification of wfn's patch from...
    
    https://trac.torproject.org/8815

commit a5f81683e1c29ba37cc4e97f0581d20c3949eb8b
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Apr 29 08:11:52 2013 -0700

    Bumping sleep for test_reattaching_listeners
    
    On our jenkins instance we're seeing this test fail roughly 1/24 of the time
    (breaking one target in 1/3 of the test runs). We *should* be doing some kind
    of activity guarenteed to trigger an event rather than rely on a BW event being
    emitted each second. For now trying to simply bump its sleeps though to see if
    that relieve the symptoms.
    
    This adds around eight seconds to our integ runtime so definitely something we
    should look into doing right later.

commit 3d047cb483f0692ef94b5b0cdac478da448957b0
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri May 3 17:28:48 2013 -0700

    Use the more idiomatic staticmethod decorator
    
    This is just a coding style change and not a functional change.  Funnily
    enough, there was already one of these decorators added in
    commit 5da6b9790da266f9.  So, this standardizes on the more Pythonic
    form.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit f5d690e5e241c8b53f6c4b0efa9f2519819a024e
Merge: 1773eba 411f8cc
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 5 18:05:43 2013 -0700

    Using new syntax for exception variables
    
    Python has two styles for exception variables...
    
    * Python 2.7 and earlier (discontinued on 3.x)
    
      except ValueError, exc:
    
    * Python 2.6 and later (started in 2.5)
    
      except ValueError as exc:
    
    Change is thanks to Sean on...
    
    https://trac.torproject.org/8823

commit 411f8ccb508ae5eb941f5b8008909e4f1254374c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 5 18:00:11 2013 -0700

    Expanding comments for exception style check
    
    Just adding a bit more of an explanation to the new style check. I'm a little
    tempted to make the second conditional "',' in content" so it'll catch
    exceptions by other names, but that might be too loose.

commit 113f8c14723db3df5dacc02c5efaf54aeca6e688
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Thu May 2 16:29:18 2013 -0700

    Convert all except clauses to new grammar
    
    Use the try..except...as grammar available in Python 2.6/2.7 and required
    in Python 3.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit e65930b402cf4edee8ae18a812af4b3ea1aeeb21
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Thu May 2 16:15:26 2013 -0700

    Check that except clauses are not using old grammar
    
    This tests that code matches PEP3110 and is more Python 3 compatible
    out of the box.  This is possible because we no longer support
    Python 2.5.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 1773ebaab470206653ce6d84c3ef1276f81c5d0a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 28 18:13:12 2013 -0700

    Updating download page entry for debian
    
    Debian was having some issues getting new packages to be reflected on
    'packages.debian.org'...
    
    https://lists.debian.org/debian-infrastructure-announce/2013/04/msg00000.html
    
    Now that it's been resolved we can link to the package Derek made for us. Yay!

commit f13c439c6b5e3702dce29d90b3daf24a4232487d
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Apr 25 09:28:28 2013 -0700

    Typo in FAQ entry for 'how to run tests'
    
    Oops, s/static/style in "but static checking is run separately...".

commit a1149f5489065cf7445f83137f3747a4866351e1
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Apr 24 21:01:06 2013 -0700

    Standardizing on digest() providing unicode
    
    Our digest() method provided bytes with python 2 and unicode with python 3. I'm
    trying to avoid these sort of version-dependent differences since they're a
    headache when we're support both serieses. The comparison for digest validation
    was also still failing in python 3 due to types. Caught thanks to aj00200.

commit 92af5b8d50e3dac52e2380cd44ff706c88d9332e
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Apr 24 09:35:52 2013 -0700

    Server descriptor digest validation failed for python 3
    
    When validating the digest for server descriptors via python 3 we forgot to
    explicitly use byte values (rather than unicode). This is an adaptation of a
    patch from aj00200...
    
    https://trac.torproject.org/8755

commit c7ed4c84b30c8555ba8e2e63c158b74876d08db2
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Apr 24 09:27:07 2013 -0700

    Helper for getting torrc of a target
    
    Breaking a chunk of run_tests.py into a get_torrc_entries() helper function to
    improve readability. The runner could use an overhaul, and might later take on
    this functionality.

commit 2a205415ed9bb1601a1ac65ec7c48f056561b3c8
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Apr 22 07:40:35 2013 -0700

    Checking pycrypto version when running tests
    
    Adding a check similar to python, pyflakes, and pep8 for the version when we
    run our tests. Bundling in various other small testing tweaks.

commit 8adc55c0139ff9f0925541be1254bea4767716ed
Author: Tomasz Kunikowski <tomasz.kunikowski@gmail.com>
Date:   Thu Apr 18 23:54:25 2013 +0200

    Extra spaces were inserted into process names via set_process_name()
    
    After running set_process_name() ps would report spaces within our process
    name. For instance, when running our integ tests our name would become...
    
    p   y   t   h   o   n       .   /   r   u   n   _   t   e   s
    
    Fixed thanks to ragwater, who figured out that ctypes.memmove() wanted unicode
    rather than byte content...
    
    https://trac.torproject.org/8631

commit 3d2f54327d64937016542381816a3529dcbfaa84
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Apr 17 10:27:57 2013 -0700

    Client usage tutorial can be disrupted if tor is running
    
    If you already have a tor instance running then the 'To Russia With Love'
    tutorial will fail to bootstrap due to a conflicting data directory. Thanks to
    Sreenatha for reporting this!

commit 3ba0e6571bd0593c547bf27dd702679e12f802ef
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Apr 17 09:26:32 2013 -0700

    Adding a :spec: role
    
    Generalising our roles to also support a :spec: alias for torspec commits.

commit 1ad1f71be638fa21b8268d2f9cebec2c82ff17dc
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Apr 16 09:02:27 2013 -0700

    Dropping '-dev' suffix for stem version on the site
    
    Like arm we're appending a '-dev' suffix for our git repository (to
    differentiate it from releases). However, for our site we just want the version
    itself.

commit 5b58537814dcc4c2c2e474adb49b899add2fd0c4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Apr 15 20:40:41 2013 -0700

    Shrinking FAQ icon
    
    Reducing the size of the FAQ icon a little bit to more closely match the
    download icon. Ain't perfect, but closer.

commit f5aacc8bff0056748e8bc16e15f98d13e4dad024
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Apr 15 20:30:14 2013 -0700

    Raising FAQ category types
    
    Changing the category types so...
    
    'subcategories' => 'categories'
    'subsubcategories' => 'subcategories'
    
    This better emphasizes the questions and gives them a slight horizontal ruler.
    I'd like to get a more emphasized divider to better split up the questions but
    the only way I know of to do that is a table, which would be gross.

commit 6b7b50c5c7834f76ddba0a78f337a5f1e4998599
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Apr 15 09:29:32 2013 -0700

    Adding a FAQ page
    
    Introducing a page with FAQ items for the most common questions that I've seen
    so far. Suggested by gsathya on...
    
    https://trac.torproject.org/8577

commit f17e370bb75c28c36e5aae1fe942b5bc97ca5438
Merge: fae5b1d 4bdcba0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 14 16:07:18 2013 -0700

    Custom sphinx role for :trac:
    
    Thanks to ragwater we now have a :trac: role for our docs. With this we no
    longer need to hardcode ticket urls, and gives a nice template we can use for
    other customizations.
    
    https://trac.torproject.org/8671
    http://doughellmann.com/2010/05/defining-custom-roles-in-sphinx.html

commit 4bdcba090b5a29e2a64c229a5ba9246ddbb59a69
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 14 16:05:03 2013 -0700

    Style revisions
    
    Our trac.py is based on an external tutorial...
    
    http://doughellmann.com/2010/05/defining-custom-roles-in-sphinx.html
    
    As such it naturally had different stylistic conventions from stem. There were
    also some minor points that had some room for simplification.

commit 5ab172051665101ce3c8797f3aac24a6d79459e7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 14 15:37:17 2013 -0700

    Using the :trac: role
    
    Replacing our hardcoded trac links with our fancy new :trac: role.

commit d5ae78b127d1103927a6febb44587f62d1ed03d1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 14 15:26:24 2013 -0700

    Truncating generated trac urls
    
    Trac expands short urls. For instance, when visited...
    
    https://trac.torproject.org/8596
    
    ... is expanded to...
    
    https://trac.torproject.org/projects/tor/ticket/8596
    
    I kinda prefer using the shortened version in our docs so copied links are more
    readable.

commit 02fa064867d186a3c9bbcc984f93ac0c20669ba9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 14 15:17:39 2013 -0700

    Including docs in our pyflakes and pep8 checks
    
    Including source from our docs directory in static checks.

commit 0567974e7c3e0610dfa62e54b0a96ef54fd59866
Author: Tomasz Kunikowski <tomasz.kunikowski@gmail.com>
Date:   Sun Apr 14 22:29:06 2013 +0200

    Implemented custom sphinx role :trac: for trac tickets
    
    Modified:
    	docs/conf.py
    
    Added:
    	docs/trac.py
    
    In accodrance to ticket #8671
    Using :trac:`1234` will display 'ticket #1234'

commit fae5b1db09ff925ac41d8bfab36621e29dbf0e69
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 14 14:47:35 2013 -0700

    Checking for unused tests
    
    Well... shame on me. We've always had a hardcoded list of our tests, but
    somehow when I moved it to our settings.cfg I missed a few. I'm not so much
    troubled about that mistake, but more that this mistake went undetected.
    
    Adding a check at the start of our tests for this kind of misconfiguration.

commit b36a9e3dc44adfda40390527fbb667a174a4736d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 14 14:20:10 2013 -0700

    Revising ATTACHSTREAM 555 response handling
    
    Thanks to Roger we now know that a 555 response means that a stream's in a
    state where it can't be attached. Swapping the exception we raise to
    UnsatisfiableRequest and adding a little test for this.

commit 5268d28c885509de174e5b7c1a657bc0278fd797
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 14 13:22:54 2013 -0700

    Catching 555 response from ATTACHSTREAM
    
    ATTACHSTREAM should only respond with 250, 552, or 551 according to the spec
    but someone came across a 555 response in the wild...
    
    https://trac.torproject.org/8701
    
    Tentatively handling this with OperationFailed until the spec is updated with
    more information.

commit ae5193f93ec7f3c9db74e1b0452939a7cb1196c4
Merge: 314876a 62413a2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 13 21:27:29 2013 -0700

    Testing framework overhaul
    
    Over time we've accumulated quite a bit of cruft. Rewriting run_tests.py and
    related modules.

commit 62413a29dbc73377ef3dd2231da1e9f35e4f30a9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 13 21:22:40 2013 -0700

    Rewriting run_tests.py
    
    Now that the building blocks are in place giving run_tests.py a long overdue
    rewrite. This pushes a great deal of the work to the test utils in the form of
    Tasks, units of work we can do in groups.

commit 8a09c4951afb3b806a75fed00e70f280287a76f2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 13 12:54:51 2013 -0700

    Dropping direct term usage in run_tests.py
    
    Our use of the output module is trending more toward printing message
    categories rather than specific terminal attributes. This will let us customize
    how things are rendered through the output module rather than chasing down all
    of the individual println() calls.

commit 24e4f0fca0fefa865aee1be527e5bb491f32aab8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 13 12:47:35 2013 -0700

    Dropping direct use of print
    
    Replacing our print calls with println(). This is both to standardize how we
    print our output as well as make later customization easier (for instance, if
    we want a --quiet option).

commit e7513978c0a39fb815eecaff2f8d9a8c958236df
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 13 12:41:22 2013 -0700

    Less verbose usage of the test.output module
    
    We use the print_line() function throughout the runner and run_tests.py.
    Shortening it to 'println()' (pity we can't call it print()) and making its
    usage a little nicer by flattening the input attributes.

commit b1f3504ac7fac93b5a26fd02525cc962228aaea2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 13 10:05:57 2013 -0700

    Dropping stem.prereq.is_python_26()
    
    The is_python_26() check was used to support python 2.5 hacks. Dropping this
    and the hacks that it was supporting.

commit 00edd58211c9f67a0499aa6c04d5c4fa97580708
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 13 10:00:37 2013 -0700

    Making check_requirements() require python 2.6
    
    Oops, the prereq module was checking for python 2.5. We're now only compatable
    with 2.6 and above.

commit 3d7bcee1aeba0f2ec2b8d808a1b57ba45ef731ae
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 12 23:14:06 2013 -0700

    Merging static_checks.py with the test utils
    
    Revising the static check functions and merging them with the test util module.

commit 6076f4cf8375565acdde8dfbfffe2be4a2327480
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 12 09:13:30 2013 -0700

    Moving clean_orphaned_pyc() to test utils
    
    Moving the functional bit of clean_orphaned_pyc() to the test utilities (the
    module specifically will *not* be used for generating output).

commit 49c55af78b8adf699f934bcd48248539acc83d6b
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 12 08:57:50 2013 -0700

    Moving funcions for getting tests into util
    
    Now that we've freed up the util namespace I'm gonna gobble it up again for
    testing framework helpers. Hopefully this will be able to replace quite a few
    of our functions.

commit e28971e61ec41ec76be74d9b703cd728d7ec0d0f
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 12 08:40:16 2013 -0700

    Merging test.network and test.util
    
    Both modules are to support tests for the ONLINE target. The main difference is
    that one module was made by Ravi and the other by Sean. At some point we should
    truely unify their functionality, but for now sticking them in the same place.

commit 02fb92e92892f58c18c4f2de1bcf831ee43f4ff6
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 12 08:33:34 2013 -0700

    Dropping runability of static_checks.py
    
    Once upon a time static_checks.py had a single function that did a fair bit
    more validation. At that time it made sense for this to be runable as a
    standalone script. However, nowadays it's mostly just helpers for pyflakes and
    pep8.

commit f141a007655ae4d667942745b1acafa77391e0d0
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 12 08:31:49 2013 -0700

    Supporting list of paths in static checker functions
    
    Reudcing a bit of redundancy by having the static_checks module support list of
    paths.

commit 90cb6e347fc02dd60e19011a13e9947e0eed7381
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 12 08:10:56 2013 -0700

    Moving ERROR_ATTR to output module
    
    The ERROR_ATTR was defined in both run_tests.py and the runner. Making both
    modules a little cleaner by moving it to the output module.

commit 10d0bd80985c6dbdf8423890d0cd5f7f346ae559
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 12 07:56:32 2013 -0700

    Untangling integration targets from configuration
    
    Even I was starting to get confused by 'em. We've always implicitely had two
    kinds of targets: run targets which constitute an integraion test run and
    attribute targets which have a non-torrc change on all test runs (such as being
    in a chroot).
    
    Making this division more explicit and dropping the confusing configuration
    dance we were doing.

commit d46ecdfafd6883ac39bf24841af988d45d2fd64e
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Apr 11 09:28:18 2013 -0700

    Dropping the argument.* config options
    
    Global state is icky, and in essence all stem.util.conf usage is global state.
    As mentioned in earlier commits the argument.* options were never used. Happily
    they were only consumed within run_tests.py so narrowing their scope is
    trivial.

commit a8d1eb06768335ac2d95af78195608ba0f091f7b
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Apr 11 08:58:05 2013 -0700

    Dropping load_user_configuration() helper
    
    One step closer to simplifying our arg parsing. The load_user_configuration()
    was mostly there to unify our configuration and commandline arguments. Now that
    the new-world-order will only have the later we should drop the helper to
    simplify the code.

commit 5ccfcfcc1827e38901754dd3fab20c3f9dcba42b
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Apr 11 08:50:20 2013 -0700

    Dropping support for a --config argument
    
    Lets face it, no one ever uses this. I doubt anyone besides me even knows
    *what* the hell it even does. Dropping this will let us simplify the test
    runner quite a bit.

commit 930e3608cfef3678357f9e4ad4168f24da19ddf6
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Apr 11 08:44:23 2013 -0700

    Dropping argument.no_color
    
    Disabling output colorization stopped being a runner argument a long time ago.
    Like git, stem's test runner makes a pretty good guess about if colored output
    is supported or not. If it's wrong then we can fix the check.

commit 347d2859073672e9c0354727901769245fb3b758
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Apr 11 08:35:31 2013 -0700

    Moving test filter to helper function
    
    Shifting support for the '--test' argument to the helper. Oh, and the integ
    helper was running the unit tests instead. ;)

commit 76e751f9c8fd338c73752a8547cedc92a95363de
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Apr 10 10:16:35 2013 -0700

    Moving test imports to the runner module
    
    Ideally the run_tests.py module should be as simple as possible, simply parsing
    the user input and calling helper functions to run the tests. To this end I'm
    moving the code for enumerating unit and integration tests into the runner
    module.

commit 314876a6a7ad2067a5e630faae2b88de180cfd62
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Apr 8 09:10:10 2013 -0700

    Adding a link to our jenkins setup
    
    Link at the bottom of our index for our automated testing.

commit 8774d00808a33dc6265d994d05d61a8823d24c75
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Apr 8 08:13:39 2013 -0700

    Missing runner import for stem.connection
    
    One thing I dislike about python imports is that they can sometimes happen
    transtively. If modoule A imports module B which imports module C then module A
    seems to have a reference to module C. This gets confusing as hell because when
    you see that the 'import B' is unused and remove it you get completely
    unrelated errors.
    
    Anywho, I got a slew of stacktraces while running the integ tests about a
    missing import for stem.connection in the runner module...
    
    ======================================================================
    ERROR: test_get_pid_by_port_netstat
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/integ/util/system.py", line 250, in test_get_pid_by_port_netstat
        elif not runner.is_ptraceable():
      File "/home/atagar/Desktop/stem/test/runner.py", line 415, in is_ptraceable
        tor_version = self.get_tor_version()
      File "/home/atagar/Desktop/stem/test/runner.py", line 569, in get_tor_version
        control_socket = self.get_tor_socket()
      File "/home/atagar/Desktop/stem/test/runner.py", line 536, in get_tor_socket
        stem.connection.authenticate(control_socket, CONTROL_PASSWORD, self.get_chroot())
    AttributeError: 'module' object has no attribute 'connection'
    
    I'm not really sure when this was introduced, but fixing it.

commit 744941e781c0597b05166001bae00b81ec8aaab1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 6 13:25:38 2013 -0700

    Moving list of test modules to settings.cfg
    
    Further cleaning up our overly-verbose run_tests.py by moving the list of test
    modules to our settings.

commit 76a7a12ae041cf9bab120264acd52a08a318d8d7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 6 12:57:01 2013 -0700

    Dynamically importing test modules
    
    Whenever we add a new test we need to add it to both the imports and UNIT_TESTS
    or INTEG_TESTS tuple. This is both a little clunky and, now that we have a crap
    ton of tests, it makes run_tests.py really long.
    
    Using __import__() to dynamically import the tests as we run them. This might
    have a slight runtime advantage too since we're just importing what we need,
    though it's not noticeable. This trick comes from...
    
    http://effbot.org/zone/import-string.htm

commit 29e7169d9a9947b86fe7af557a7230932fcc49f8
Merge: b7d8cad c719b1e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 7 17:25:34 2013 -0700

    Site navbar menu
    
    Replacing the normal sphinx 'previous/next' navigation with a navbar menu.
    Hopefully this will make site navigation far, far easier...
    
    https://trac.torproject.org/8431

commit c719b1e5680898e9d1f6ede953b3169d12580f64
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 7 17:22:17 2013 -0700

    Simplifying site navigation via a navbar menu
    
    Using a CSS style navbar to make site navigation easier. This is based on the
    one from...
    
    http://aext.net/2009/09/elegant-drop-menu-with-css-only/

commit 174d39e9e5b6825da3ebb0490b3ebffb7d858fff
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 7 11:33:49 2013 -0700

    Replacing paging nav bar with categories
    
    Sphinx wants the layout of sites to be like a book, with a 'previous' and
    'next' page. I'm at a loss for how that could possibly be a good idea. Sites
    like stem's have categories (tutorial, api, download, etc) which our navbar
    should provide a quick method to access.

commit bc8ec1da82bd45e2ec7b694ea824fedc49ae6395
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 7 11:32:17 2013 -0700

    Forking the haiku theme
    
    The haiku theme has limited customizability. In particular we can't...
    
    * Overwrite the normal nav elements (previous, contents, next) with something
      else since they aren't within a block. The only way to overwrite them it to
      clobber the contents block, which kinda defeats the point of using their
      layout.
    
    * The css uses text layout right but we want text layout left.
    
    Hence I'm copying in the haiku layout and css so I can customize them. This is
    the first step toward a proper navbar for our site...

commit b7d8cadd12ff630bfe204e3f575bd624e9fae224
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 6 22:14:45 2013 -0700

    Favicon for our site
    
    Mmmm, 32x32 favicon. I tried to get an ico as the sphinx config suggests but
    that didn't work. Oh well, just about every browser except maybe IE can handle
    pngs so I'd prefer to go with that anyway.

commit ad6973d05a0b25dc80768264a41cb717326c75f3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 6 21:47:44 2013 -0700

    Adding a change log page
    
    As changes accumulate we need a page to convey them to our users. Adding a
    change log that both has this, and information about what our versioning means.

commit 64fef2d9ee6ab5aefc597eb4f68c910df8f0a197
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 6 18:13:35 2013 -0700

    Settling on a logo for stem
    
    Using Jakub Jankiewicz's WPClipart image as the logo for stem. This image is
    under the public domain, and a very nice fit for our library. :)

commit 9c8208474758862b7ae451c5c7eb93795df1015a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 6 17:43:17 2013 -0700

    Noting alternate onion images
    
    Citing some other more artistic (but less iconic) images that I came across
    while looking for a logo for stem. Might come in handy in the future and are
    under Creative Commons (thanks Andrea!).

commit 46d39df76cc51132c8e38c4fcbaabd45844e0c1c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 6 16:52:09 2013 -0700

    Support for the ADDRMAP event's CACHED flag
    
    Adding support for the new CACHED flag added by Desoxy...
    
    https://trac.torproject.org/8596
    https://gitweb.torproject.org/torspec.git/commitdiff/25b0d43a9ebe89541625e9f1a090e4a701748c76

commit 5879995bf9118792a9c27376a93361e779ec4fc2
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 5 09:19:48 2013 -0700

    Skipping test_get_pid_by_port when connection resolvers are unavailable
    
    Our jenkins setup revealed that when netstat, sockstat, and lsof are all
    unavailable test_get_pid_by_port fails. Skipping instead since those connection
    resolvers are dependencies for the test.

commit 95e9d663796a49fdc2427c7b70fe5f557b0e2cb3
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 5 09:02:59 2013 -0700

    Improving error message when get_system_tor_version() can't find tor
    
    When tor is unavailable the error output from get_system_tor_version() is
    pretty sucky...
    
    Traceback (most recent call last):
      File "./run_tests.py", line 525, in <module>
        our_version = stem.version.get_system_tor_version(CONFIG["argument.tor"])
      File "/home/atagar/stem/stem/version.py", line 84, in get_system_tor_version
        raise IOError(exc)
    IOError: [Errno 2] No such file or directory
    
    Making it more developer friendly.

commit f5971e1da626425062d8922a35b7b87447faa238
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 5 08:47:34 2013 -0700

    Still passing when tor's stime is zero
    
    The stime proc field is an integer, rounded down to the number of seconds that
    a process has been running. On my piddly little netbook tor takes long enough
    to start that this is always non-zero, but on our jenkins systems it's so quick
    that the test fails. Thanks to weasel for all of the help!

commit a920e5cfc2e3d0283bdb467a9a78bd281b8ee2a2
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Apr 3 08:33:40 2013 -0700

    Python 2.6 doesn't support assertIsNotNone()
    
    The controller integ tests included a couple assertIsNotNone() calls, which
    wasn't added until python 2.7. Caught and fixed by Desoxy - thanks!
    
    https://trac.torproject.org/8629

commit 5c9efa028b3e61d038d6fbf617e3f7807a75a1f3
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Apr 2 10:15:20 2013 -0700

    Adding resource for event tutorial logo
    
    Including the full sized image used for the 'Tortoise and the Hare' tutorial.

commit ec40da3a8a5959b92897c17d284d5883082567bd
Merge: d3f1d80 c713715
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 30 18:50:19 2013 -0700

    Revising the download page
    
    Expanding our download page to include packaging done for our 1.0.1 release.
    Many thanks to everyone involved!

commit c71371521ef0b8a294755f00a7689a935ab67f32
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 30 18:43:19 2013 -0700

    Emphasising installation commands
    
    Using a preformatted block for installation commands since this is what
    visitors are likely most interested in.

commit c9122fbd24f32b288eebfa60dc7c77e802e58935
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 30 18:24:15 2013 -0700

    Adding Red Hat to the download page
    
    Linking to the ticket Juan filed to add our stem release to redhat/fedora.

commit 170e44d3380b4cac140dc0ee52c280692d6aec0f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 30 18:07:49 2013 -0700

    Resizing FreeBSD logo
    
    Shrinking the logo a little bit to match the other entries.

commit 302ad96b7eaacc40784dfef43d4a3cdbcba4986e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 30 18:02:59 2013 -0700

    Using the official git logo
    
    I'm a bit torn between our custom git logo and the official one. Opting for the
    later in large part because it's now next to FreeBSD which has a red logo, so
    this looks a little better.

commit ee0089ba76d3c4357113fe11d3302682afca50de
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 30 17:10:19 2013 -0700

    Instructions links for ArchLinux and FreeBSD
    
    Adding the instruction links for installing via AUR and ports.

commit c3b37286675617930f7bd3aad2d8581bab2752d3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 30 16:20:15 2013 -0700

    Adding FreeBSD to our download page
    
    Adding the stem port Carlo made for our release.

commit 74dc0846a647b4a7c55b51e39fd9b3c6d8a24e0d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 30 16:06:50 2013 -0700

    Citing ITP request in debian download page entry
    
    Pointing the request Dererk filed for adding our release to debian.

commit 3db5fd85b275a9413c518760e8e78dd35272974e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 30 16:01:59 2013 -0700

    Updating ArchLinux download url
    
    Sjon has uploaded a new AUR package for our 1.0.1 release. Pointing our
    download page at that instead.

commit 626f8247b74e8500a4557b4da55f94f533f3ed3c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 30 15:58:15 2013 -0700

    Revising download page entry for the source repository
    
    Placing the source repository at the bottom of the download page and renaming
    it.

commit 9e4d1b9adde0414efaec23bb1547f055b8db0f92
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 30 15:48:45 2013 -0700

    Adding PyPI to download page
    
    Stem is now available on the python package index...
    
    https://pypi.python.org/pypi/stem/
    
    Thanks to refnode for providing us with the namespace!

commit d3f1d80483d88deed04871aac4403377458fb406
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 27 09:41:58 2013 -0700

    Stem release 1.0.1

commit 4ee02d05f266a0fd60f470d28992026f6a6d49d6
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 27 09:37:58 2013 -0700

    Avoiding stem imports in setup.py
    
    Our setup.py may be ran under python 2.x or 3.x. Because of this the file and
    anything it imports needs to run under both python serieses *without* a 2to3
    conversion.
    
    We were importing basic module information from __init__ which in turn brought
    in the enum and str_tools utils...
    
    atagar@morrigan:~/Desktop/stem$ python3 setup.py install
    Traceback (most recent call last):
      File "setup.py", line 5, in <module>
        from stem import __version__, \
      File "/home/atagar/Desktop/stem/stem/__init__.py", line 504, in <module>
        "ERR",
      File "/home/atagar/Desktop/stem/stem/util/enum.py", line 64, in UppercaseEnum
        return Enum(*[(v, v) for v in args])
      File "/home/atagar/Desktop/stem/stem/util/enum.py", line 77, in __init__
        if isinstance(entry, (bytes, unicode)):
    NameError: global name 'unicode' is not defined
    
    Bug caught by Dererk.

commit 07553f7b4cb3e21b74b5582f035a2f22477183c4
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 26 09:21:06 2013 -0700

    Stem release 1.0.0

commit 8bc11004558a1362f61f75d967a221183777672d
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 26 08:29:13 2013 -0700

    Another python3 authentication issue with mixed str types
    
    Fixing another failure in our python3 integ tests with cookie authentication.
    The hmac function provides us with bytes but we need to be careful to use
    unicode in the formatted string to avoid a b'' wrapper.

commit c75ab2812b90a730e135b5970ff35d431fa2c4a4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 25 08:33:57 2013 -0700

    Safecookie authentication broken when using python 3
    
    The os.urandom() provides bytes so there's no need to convert that, but we
    needed to be more careful about using the converted hex in the formatted
    string. It too was bytes, causing an extra embedded b''...
    
    >>> "hello %s" % b"damian"
    "hello b'damian'"
    
    Tor in turn rejected the nonce as being invalid. I don't get why python3 does
    it. With python2 bytes and unicode could both be used in a reasonable fashion
    for formatted strings - these extra characters are almost always undesirable.
    Oh well...

commit 83b99dbe6579af04f18553740b175d6de44fa27d
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 25 08:12:30 2013 -0700

    Integ tests failed to create a control socket
    
    The RUN_SOCKET test was failing while trying to make the control socket because
    it tried to change the permissions of /tmp to 700 (which, of course, normal
    users can't).
    
    Tor rejects to start if the control socket is in a globally readable directory,
    so making a temporary directory we can set the permissions on.

commit 89974d0f67f09f44c0fb8b3f2344a78b27280dc4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 25 08:10:36 2013 -0700

    Only running RUN_OPEN by default or if requested
    
    We were configured to always run the RUN_OPEN target, even if something else
    had been requested. This bug lasted quite a while because we usually either run
    the default or RUN_ALL (which includes RUN_OPEN).

commit 5e6b5407f087b6b31d28ed5047e44b07f082d197
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 24 21:05:14 2013 -0700

    Controller provided unicode microdescriptors with python 3
    
    When using python 3 the get_info() method provides unicode content. We were
    passing this content to descriptor constructors, causing its raw_content to be
    unicode rather than bytes. This in turn caused a stacktrace when calling
    methods like hashlib.sha256().
    
    This doesn't effect python 2.x, where str is bytes anyway.

commit fa0fbd195d0531980a6e250ee2e32c99a0708259
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 24 20:52:07 2013 -0700

    Allowing stem.descriptor.parse_file() to handle paths
    
    Several times now I've passed a string path to parse_file() rather than a file
    object and been confused for a sec by the stacktrace. On reflection there's no
    reason *not* to accept paths - that's commonly what we want it for after all
    and eliminating the extra open() lets our callers have much nicer code.

commit e10b6124e22fd9febd7bd23abcc4b820353fbf8b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 24 20:35:22 2013 -0700

    Simplifying use of digest values for microdescriptors
    
    To be useful a microdescriptor often needs to be matched against its
    corresponding router status entry. The design of microdescriptors makes this
    suck - the only way to do so is to iterate over all router status entries
    looking for one with the corresponding digest.
    
    Adding examples for doing this and making the following changes to make the
    matching easier...
    
    * changing the digest value of RouterStatusEntryMicroV3 from base64 to hex
    * adding an attribute with the hex digest to Microdescriptors

commit b05bf7a7eeb9807bf003518b39677f8c6abc39c8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 24 19:40:23 2013 -0700

    Having ExitPolicyRules report /0 as a wildcard
    
    ExitPolicyRules can apply to both IPv4 and IPv6. Because of this I treated * as
    a wildcard, but *not* /0 (since 0.0.0.0/0, though it applies to all IPv4
    addresses, doesn't cover IPv6). However, in most cases (everywhere except the
    torrc?) Tor's exit policies are limited to IPv4. As such, /0 should usually be
    treated as a wildcard. Change suggested by Aaron Johnson.

commit caee7d6c968fd6cd9912dc4a554d66e303316b61
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 23 19:50:52 2013 -0700

    Switching exit policy's can_exit_to() and is_match() be permissive
    
    Switching how the ExitPolicy's can_exit_to() and ExitPolicyRule's is_match()
    handle an undefined address or port. Previously we reported if we matched or
    allowed exiting to *all* instance of the defined destination. Now, however,
    rather if we match against *any*.
    
    Change was requested by Aaron Johnson. You can still get the old behavior by
    setting the 'strict' argument.

commit cefe584cd57c45a77379428a4c15fb225c0afdcc
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 23 19:17:31 2013 -0700

    Testing errors when descriptors are unavailable
    
    Fixing a couple testing errors spotted by Dererk. These arise when we're either
    unable to fetch descriptors (disconnected with a fresh test data directory) or
    possably if descriptors are still in the process of being fetched.
    
    ======================================================================
    ERROR: test_get_microdescriptors
    ----------------------------------------------------------------------
    Traceback:
      File "/home/dererk/stem/stem/test/integ/control/controller.py", line 843, in test_get_microdescriptors
        for desc in controller.get_microdescriptors():
      File "/home/dererk/stem/stem/stem/control.py", line 1050, in get_microdescriptors
        raise exc
    OperationFailed: Data directory doens't contain cached microescriptors
    
    ======================================================================
    FAIL: test_get_microdescriptor
    ----------------------------------------------------------------------
    Traceback:
      File "/home/dererk/stem/stem/test/integ/control/controller.py", line 823, in test_get_microdescriptor
        test_relay = self._get_router_status_entry(controller)
      File "/home/dererk/stem/stem/test/integ/control/controller.py", line 1018, in _get_router_status_entry
        self.fail("Unable to find any relays without a nickname of 'Unnamed'")
    AssertionError: Unable to find any relays without a nickname of 'Unnamed'
    
    He also got a third error that mystifies me a bit. Fingers crossed that earlier
    changes have addressed it, if not then I'll need repro steps to dig in
    further...
    
    ======================================================================
    ERROR: test_get_network_status
    ----------------------------------------------------------------------
    Traceback:
      File "/home/dererk/stem/stem/test/integ/control/controller.py", line 922, in test_get_network_status
        self.assertRaises(stem.ControllerError, controller.get_network_status, "blargg")
      File "/usr/lib/python2.7/unittest/case.py", line 476, in assertRaises
        callableObj(*args, **kwargs)
      File "/home/dererk/stem/stem/stem/control.py", line 1166, in get_network_status
        raise exc
    ValueError: Router status entries must have a 'r' line:

commit 73ba530be8d7df7600e1b9a29df904de29a94831
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 23 19:03:46 2013 -0700

    Switching network status document routers to a dict
    
    Changing the 'routers' attribute of NetworkStatusDocuments to a dict mapping
    fingerprints to the router status entry. This is generally far more convenient
    to work with - thanks to gsathya for the idea!

commit ea060119690650b4997c5f4c25f036b3ddeda3ac
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 23 16:24:47 2013 -0700

    Presently no quoted GETCONF responses exist
    
    As per ticket #6172 the GETCONF response currently returns each result on a
    separate line, so quotes are unnecessary. If tor does add quoted values then
    we'll need to see at that time how it should be handled.

commit b3c00c4a6d0565939745e6eba2064b17aaff491a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 23 15:55:13 2013 -0700

    Dropping python 2.5 compatibility hacks
    
    As discussed on tor-dev@ there's a lot of arguments against maintaining support
    for python 2.5 and precious few for...
    
    https://lists.torproject.org/pipermail/tor-dev/2013-March/004551.html
    
    Dropping all the hacks we adopted to support it.

commit 88c3b1a6ae4152f1724fcc7267b7a5b69aae5a74
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 23 15:06:08 2013 -0700

    Changing isinstance() checks to look for both bytes and unicode
    
    Commit b12e9c0 broke a couple of our integration tests because it caused us to
    call controller methods with unicode rather than str arguments...
    
    https://trac.torproject.org/8576
    
    This should be fine - python 2.x can usually use bytes and unicode inputs
    interchangeably. The problem was that we routinely use isinstance() checks to
    see if an input is a string and, if it isn't, assume it's a collection. As a
    result we were calling...
    
    GETINFO m d / n a m e / f o o
    
    ... rather than...
    
    GETINFO md/name/foo
    
    Tor in turn errored saying that those single character getinfo options didn't
    exist. To fix this I'm changing most of our isinstance() checks to look for
    both bytes and unicode.

commit cba36249020b0f322ce0ca6438d31dd5d4281905
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 23 12:33:56 2013 -0700

    Issuing exit status 1 if there's any testing failures
    
    We explicitely set an exit status at several points but we don't set one for
    the most important situation: testing success/failure at the very end. Caught
    by weasel.

commit bee3c2b53653976a36b16a7cefb6e11bb56e4045
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 22 20:32:07 2013 -0700

    Running python 3 tests errors when done with python 2.6
    
    When running tests with '--python3' we were using the named tuple
    sys.version_info. This was just a normal tuple prior to python 2.7, causing an
    error when run with python 2.6. Caught and fixed by Desoxy...
    
    https://trac.torproject.org/8565

commit c7f34c929a4f091b0527bec30d340472b55eacab
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 22 19:49:06 2013 -0700

    Swapping contact and platform attributes to bytes
    
    The 'contact' and 'platform' lines in a server descriptor can have any
    arbitrary byte content. This isn't necessarily unicode, so we can't provide
    that to our callers without potentially mangling the data.
    
    Non-unicode contact lines are surprisingly common (fourteen instance in one
    cached-descriptors, around 0.4%).

commit b12e9c01c318863f4e4cddacf341abfd197ff412
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 20 18:28:55 2013 -0700

    Storing raw descriptor content as bytes
    
    Tor descriptors are byte strings. While most content is ASCII (and hence maps
    to UTF8) there are a couple exceptions: the server descriptor's contact and
    platform lines which are arbitrary byte content. Usually this is unicode but on
    occasion it isn't (14 of 3044 in my old consensus cache).
    
    In those cases we converted the content to unicode and replaced the
    un-decodeable characters. This in turn caused our _verify_digest() check to
    report the descriptor content as being malformed if the user had pycrypto
    installed.
    
    Changing the raw contents the Descriptor class tracks to be bytes instead of
    unicode. This actually leads to far more elegant code, letting us drop the
    UnicodeReader which was always a pretty gross hack.
    
    This breaks python 2.5 compatability by using the io.BytesIO class. Presently
    I'm checking with tor-dev@ to see if anyone objects with dropping python 2.5
    support entirely...
    
    https://lists.torproject.org/pipermail/tor-dev/2013-March/004551.html
    
    If the answer is...
    
    * yes, drop support: we'll replace test.mocking.BytesBuffer with io.BytesIO
      (they both do the exact same thing)
    
    * no, keep support: we'll need to move test.mocking.BytesBuffer to a util and
      use it instead of io.BytesIO

commit 98cb69522254ddfd7bdd3c8df3b0dc7020c8e0a3
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 20 17:34:07 2013 -0700

    Avoiding use of sign_descriptor_content()
    
    The test.mocking.sign_descriptor_content() function is slow. When the user has
    pycrypto the runtime of our unit tests jump from 17 to 83 seconds.
    
    This is a no-go since it discourages the use of our tests (our unit tests were
    already taking too long imho before this jump). By dropping
    sign_descriptor_content() the digest validation checks fail, so mocking those
    out.

commit 617661308bbc70c08b2d45767965969c690ac9f9
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 20 09:50:12 2013 -0700

    Test still used old arg name for socket connections
    
    Oops. When running other targets we were sometimes hitting an instance that
    still used 'socket_path' rather than 'path'. Caught by gsathya.

commit dd220882e6c73863cc574ec4ba3833ee478314cb
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 19 13:40:10 2013 -0700

    Dropping support for port lists in descriptors
    
    The 'or-address' and 'a' descriptors supported comma separated lists of ports.
    However, as per spec commit 99a9587 tor has never supported this, and the spec
    no longer includes this capability.

commit cbd83abf6e91955d84a87d3dada8548ad13ab99f
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 19 08:25:06 2013 -0700

    Commenting tutorial use of curses.wrapper()
    
    Suggestion from Lunar for a rough spot in the tutorials.

commit 32244cdc3e8fbfff0cb9ca17f848ce8ede8ade3b
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 19 08:15:19 2013 -0700

    Tutorial wording revisions
    
    Addressing various suggestions from Sean for improving the tutorials.

commit 796ea199b44fbd0fe4d2b55faad1a1bdcd64bef0
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 19 07:09:50 2013 -0700

    Switching the server descriptor's family to a set
    
    Suggestion from gsathya. Duplicate family entries are moot and order doesn't
    matter, so might as well make this a set.

commit decc35249d2605644aec5a3ef5bb337cd0d41175
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 18 18:40:13 2013 -0700

    Test accidently left around a file in /tmp
    
    The test_save_processed_files_malformed checks that save_processed_files()
    fails when presented with malformed data. However, this function still writes
    out a blank file. I noticed this because I just tried running the tests as a
    different user from prior test runs, causing...
    
    ======================================================================
    ERROR: test_save_processed_files_malformed
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/integ/descriptor/reader.py", line 172, in test_save_processed_files_malformed
        self.assertRaises(TypeError, stem.descriptor.reader.save_processed_files, "/tmp/foo", listing)
      File "/usr/lib/python2.7/unittest/case.py", line 465, in assertRaises
        callableObj(*args, **kwargs)
      File "/home/atagar/Desktop/stem/stem/descriptor/reader.py", line 229, in save_processed_files
        with open(path, "w") as output_file:
    IOError: [Errno 13] Permission denied: '/tmp/foo'
    
    ----------------------------------------------------------------------
    Ran 21 tests in 1.558s

commit 8099efdecf3d64abef787d1b19b537c4b9c30088
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 18 18:32:28 2013 -0700

    Only query by name for relays with the Named flag
    
    Our integ tests pick a relay at random from its present consensus to test the
    'GETINFO desc/name/*', 'md/name/*', and 'ns/name/*' functions. However, this
    could fail if we're querying a relay without the Named flag.
    
    ======================================================================
    ERROR: test_get_microdescriptor
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/integ/control/controller.py", line 826, in test_get_microdescriptor
        md_by_nickname = controller.get_microdescriptor(test_relay.nickname)
      File "/home/atagar/Desktop/stem/stem/control.py", line 1001, in get_microdescriptor
        raise exc
    InvalidArguments: GETINFO request contained unrecognized keywords: md/name/MukiMukiAmaguri
    
    ======================================================================
    ERROR: test_get_network_status
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/integ/control/controller.py", line 928, in test_get_network_status
        desc_by_nickname = controller.get_network_status(test_relay.nickname)
      File "/home/atagar/Desktop/stem/stem/control.py", line 1165, in get_network_status
        raise exc
    InvalidArguments: GETINFO request contained unrecognized keywords: ns/name/MukiMukiAmaguri
    
    ----------------------------------------------------------------------
    Ran 31 tests in 11.307s

commit 6862dc4994996e78ebf4516094c29aad66ceb524
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 18 18:20:35 2013 -0700

    Skipping tests that fail when run as root
    
    Some of our integ tests rely on permission failures for files we lack access
    to, or path expansions. Running as root causes these to fail. Caught by
    Dererk...
    
    https://trac.torproject.org/7925

commit 98fd0a7ba4787340d05d0baf341594555da87d00
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 18 16:36:11 2013 -0700

    The 'name' attribute of TarInfo classes can't be set in python 3
    
    Workaround for the 'name' attribute being unwriteable in python 3. On
    reflection this is a little more elegant than our prior hack. :)
    
    Exception in thread Descriptor Reader:
    Traceback (most recent call last):
      File "/home/atagar/Python-3.3.0/Lib/threading.py", line 639, in _bootstrap_inner
        self.run()
      File "/home/atagar/Python-3.3.0/Lib/threading.py", line 596, in run
        self._target(*self._args, **self._kwargs)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/reader.py", line 435, in _read_descriptor_files
        self._handle_file(target, new_processed_files)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/reader.py", line 514, in _handle_file
        self._handle_archive(target)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/reader.py", line 557, in _handle_archive
        entry.name = target
    AttributeError: attribute 'name' of '_io.BufferedReader' objects is not writable

commit c40ea0fb2f2188d540655cfc4c6975bffc9585b2
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 18 16:19:39 2013 -0700

    Fixing open() mock for python 3
    
    Our unit tests were broken under python 3 due to needing a slightly different
    mock for the open() function.
    
    ======================================================================
    ERROR: test_mirror_mirror_on_the_wall_2
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/tutorial.py", line 107, in test_mirror_mirror_on_the_wall_2
        tutorial_example()
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/tutorial.py", line 95, in tutorial_example
        for desc in parse_file(open("/home/atagar/.tor/cached-consensus")):
    FileNotFoundError: [Errno 2] No such file or directory: '/home/atagar/.tor/cached-consensus'

commit bc155d6ee800c1741fb7b1b7780cf696217115f4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 18 14:00:45 2013 -0700

    Dropping the memory usage assertion when reading descriptors
    
    Our integ test for reading descriptors tries to check that we aren't consuming
    gobs of memory. Unfortunately this assertion seems to be platform dependent (it
    fails on OSX) so dropping the assertion. Yea, yea, I should be fixing it but we
    haven't had this issue for a very long while and it isn't likely to reoccure.

commit c614080d911cec936a90b1f106e777f962ad8249
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 18 13:53:17 2013 -0700

    Tor switched from server to microdescriptors in 0.2.3.3

commit bf5bac34abf95f1e12f0ce4030c305c364939c1f
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 18 13:34:41 2013 -0700

    Digest validation fails for 'non-ascii_descriptor'
    
    The 'non-ascii_descriptor' test descriptor fails digest validation due to being
    slightly modified. Finding that descriptor was a PITA so simply mocking out the
    digest validation rather than trying to dig up the original.

commit b8d79e24b4599b59df1d68bd99273cebb11a8038
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 16 11:30:36 2013 -0700

    Unescaped characters shouldn't be considered in further unescaping
    
    When we unescape responses we should process the string from start to end, and
    when something is unescaped it needs to be removed from further consideration.
    For instance, '\\t' should be unescaped to '\t', not a tab. Caught on...
    
    https://trac.torproject.org/8471

commit a124b62f26c8ae1cee01f780a0383389275ee873
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 16 10:12:57 2013 -0700

    Supporting 'Unmeasured' in router status entries
    
    Spec commit c3860e3 added a new 'Unmeasured' key/value entry to the 'w' lines
    of v3 router status entries.

commit cb7915bb4734742f487519dad6ce0ec2a84b5bb7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 16 09:42:41 2013 -0700

    Including param constraints for UseOptimisticData
    
    We were missing the bounds for the UseOptimisticData parameter.

commit 2f90285dd88db3fdac5dc14441b7ffe19ce8da38
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 16 09:05:22 2013 -0700

    Chaning GuardEvent to match spec revisions
    
    Nick has clarified GUARD events in...
    
    https://trac.torproject.org/7619
    
    Updating our enum docs to reflect the revisions and changing our 'name'
    attribute to reflect that it could contain either a fingerprint or
    fingerprint/nickname pair. This is a similar pattern to what we use for
    ORConnEvent.

commit 0ce65a43d68291388a5a0f61be2427d78ce7167f
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Mar 14 09:23:21 2013 -0700

    Providing default ipv6-policy entries
    
    The server descriptor's newish 'ipv6-policy' lines are now documented as being
    reject-all if it's missing...
    
      A missing "ipv6-policy" line is equivalent to "ipv6-policy reject 1-65535".
    
    https://gitweb.torproject.org/user/nickm/torspec.git/commitdiff/ef5513b1ad3671153f20b3c3929e1acba1b873bc

commit 8474971d99da327169e5827db660f8f1f91bbd59
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 13 09:38:19 2013 -0700

    Dropping the V3Dir flag
    
    Nick just pushed a change noting that the V3Dir flag has never existed, so
    updating our enum too.
    
    https://gitweb.torproject.org/torspec.git/commitdiff/67768b16c247f937a4345af94beb58a4238b0730

commit 89e8fb8f37ebaad8d2009fedba766d1d42831dd8
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 13 09:14:33 2013 -0700

    More uniformly renaming to 'tutorials'
    
    Several of our resources, including individual tutorial pages, were still under
    'tutorial'. I'm also dropping the old 'tutorial.rst' (I'm not getting any
    replies to the tor-dev@ email and having a stale copy is confusing).

commit c1abfa73fc24528713be233347cbd0f8796c537e
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 13 09:09:54 2013 -0700

    'Tortoise and the Hare' tutorial
    
    Adding a tutorial for event listening. This has the most substantial example
    yet, making a curses graph for tor's bandwidth usage.

commit 92d395c8515e72e9c9b9b89d9e84512419f51aa4
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 12 06:55:22 2013 -0700

    Tutorials still referenced a control_port arg
    
    Oops. When updating the tutorial tests I guess that I forgot to update the
    examples. :(

commit fb35fc7a9dd9d0217bedbc38bda51f58579d5f77
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 11 19:01:16 2013 -0700

    Filling in a description for MEASURE_TIMEOUT
    
    Nick added a description to address...
    
    https://trac.torproject.org/7626
    
    I'm not quite sure what it means (does "see how long it takes" mean measure the
    time to make the circuit? or the time it takes before being timed out like the
    name suggests?), so just copying what he put in the spec.

commit 98f878087dfb5eeac3fd2720b47b4ffe1fe1bc75
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 11 18:28:24 2013 -0700

    Dropping 'aka' from tutorial labels
    
    The labels look a little nicer this way. Suggested by Josh Barr.

commit 2c9710060e6194e30d6971377ddc80fa681eeaa9
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 11 18:07:54 2013 -0700

    More changes for tutorial rename
    
    Drats, I pushed the prior commit before I had intended to. Fixing the buttons
    too to reflect the rename.

commit 48938b1d95d25ef255c19cfdd7a8a30714a80b76
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 11 00:22:43 2013 -0700

    Pluralizing the tutorials page
    
    Renaming the 'tutorial' page to 'tutorials'. I can't remove the old page yet since I just sent it to tor-dev@, but now that its only link is swapped over we should be able to move it in a few weeks.

commit 74b80d61fe3279ca006ecae86271557523e33967
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 10 23:07:48 2013 -0700

    Making all labels links
    
    Among the tutorials the labels (ie. the text summary for a section) was a link
    but on the download and example pages they weren't. Standardizing on them being
    links.

commit 7d13f767c08e35d55324186832deb7c6df6de62d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 10 22:59:23 2013 -0700

    More clearly stating tutorial topics
    
    While our cutesy titles are fun, they shouldn't get in the way of our tutorial
    page making sense. Dropping them to be a subheading, in favour of a clearer
    summary of what the tutorial's about.

commit 4d89032ba6f7d1c2babd54c593668c8e74b23f98
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 10 22:40:11 2013 -0700

    Adding an examples section to our site
    
    Example sections containing projects and scripts using stem. Presently this is
    pretty small - hopefully it'll grow soon!

commit 99dbc604fb6781163d2cdf5262b759cbdc62c71a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 10 17:41:58 2013 -0700

    Making stem.util.term public
    
    Now that we're using it for a tutorial we should vend this module. I'm a little
    unsure about the name of the format() function since there's a builtin by that
    name, but I'm having a hard time thinking of something better.

commit ecc5f7bcd7d9669521e3b3b9275b18dff98c4270
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 10 11:28:58 2013 -0700

    'To Russia With Love' tutorial
    
    Soviet themed tutorial for programmatically starting tor and reading a site
    through it. I've been wanting to write this for quite a long time...
    
    https://trac.torproject.org/7505

commit 860e89bed7a4d6769d9d4376a770e9af19de684b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 9 23:13:25 2013 -0800

    Dropping 'last updated' timestamp from site
    
    While I'd kinda like to have this on our site, it's stupidly broken. This is
    probably due to sphinx caching (it only makes static files for things that have
    changed). Presently it says that our site was last updated on December 31st
    which is... very wrong. No date is better than the wrong date so dropping it.
    
    https://trac.torproject.org/7780

commit a7fae56f9dab6cf766699c229edb2fd0531b7bd9
Merge: d45f89d aab3bf5
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 9 23:10:26 2013 -0800

    Marking stem.util modules and functions as being private
    
    The stem.util space has several things that I'd rather not vend. Things that
    are either likely to change in the future or simply don't make sense outside of
    stem's internal use. Marking functions and some entire modules as being
    private.

commit aab3bf57a87bb09d4d572195e568e73da489f38a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 9 23:07:23 2013 -0800

    Marking stem.util.str_tools functions as being private
    
    Swapping several of the functions that we don't want to vend as being private.
    The to_bytes() and to_unicode() functions in particular are simply there for
    python 3.x support, so they'll disappear if/when we drop python 2.x support.

commit 7f4b7a93d2ff55388a8566919fc19baf2f1a48df
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 9 22:47:05 2013 -0800

    Marking the log, proc, term, and tor_tools utils as being private
    
    More modules that I'd rather not yet vend to stem users. They're more than
    welcome to use them, but I can't yet promise a stable API (and in many of these
    cases there's little reason they ever should be used outside of stem).

commit ec0b5538a95a39f7e2b3a3fbcf47a03739f16a6b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 9 22:33:07 2013 -0800

    Notice to stem users not to use stem.util.ordereddict
    
    The ordereddict module is to allow us to support python 2.5 and 2.6. Adding a
    notice that it may be dropped in the future.

commit f8b538b00bd16fece4d368c892a7977dd0f8260b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 9 22:29:37 2013 -0800

    Making several stem.util.connection functions private
    
    The connection module has several functions that I'd rather not vend. Marking
    them as private to discourage their use.

commit 26c854e7c032d2f7df822316f56b0d62075d29d2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 9 22:18:22 2013 -0800

    Renaming stem.util.connection.get_mask()
    
    Renaming the function to get_mask_ipv4() to match the module conventions now
    that we have ipv6 support.

commit 08c5f3d6020c860077125a976a336f076f9b9e7c
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 8 09:37:35 2013 -0800

    Renaming stem.util.connection.is_valid_ip_address()
    
    Changing is_valid_ip_address() to is_valid_ipv4_address() so it'll be in better
    keeping with the modules conventions now that it also supports IPv6. We might
    add is_valid_ip_address() later to check if it's an IPv4 *or* IPv6 address.

commit d45f89dbce487aec2a3988138966387da55adaf5
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 8 07:57:28 2013 -0800

    Clarifying requirements for timing out launch_tor()
    
    The signal timeout we were doing isn't just *nix specific, but relies on global
    state. It turns out that no languages really have a good method for doing this
    sort of arbitrary timeout - pity...
    
    http://eli.thegreenplace.net/2011/08/22/how-not-to-set-a-timeout-on-a-computation-in-python/

commit a564ae2a6c24aebb822e777e7dea8bda3cffcf68
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Mar 7 09:33:37 2013 -0800

    stem.process.launch_tor() doesn't work without stdout logging
    
    The launch_tor() function uses tor's stdout logs to figure out when it has
    finished bootstrapping. This, of course, doesn't work if tor isn't logging to
    stdout...
    
    https://trac.torproject.org/7677
    
    Changing launch_tor_with_config() to slip in a 'Log NOTICE stdout' option so it
    always works, and adding a note to launch_tor() warning users of the
    requirement.

commit 32d39439b3a1cb2a72b2efc9d3010d2c90b174ee
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 6 20:12:06 2013 -0800

    Making the arrived_at arg of events optional
    
    The kwargs of stem.response.convert() are documented as being optional, however
    the arrived_at argument for events was mandatory. Caught by meejah.

commit f40ab20110a1979efbffaa7854eca5ff77a5e9c3
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 6 19:56:57 2013 -0800

    Dropping the _get_pid() helper from stem.process
    
    Huh, that's odd. There's a comment saying that I tried to use the mocking
    module for os.getpid() but it was a no-go, necessitating an ugly hack. I just
    tried again and it worked - maybe a bug we have since fixed with the mocking?

commit de9b6220b21f0c09239181ab17a9f7554ea601eb
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 6 09:39:34 2013 -0800

    Mocking didn't properly revert static methods
    
    When working on our tutorials I expanded our mocking module to handle static
    methods. This worked well for our unit tests, but when we reverted the mocks we
    did so as standard rather than static methods.
    
    This in turn broke all of our integ tests that relied upon
    Controller.from_port(). This went undetected because it only manifests when
    *both* unit and integ tests are run together (and I've only been running one or
    the other recently).

commit 27630c4d7f17052975238a0f7622ec3418bdbd5e
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 5 11:01:56 2013 -0800

    Replacing Controller's enable_caching constructor arg with setter
    
    It's clunky for cache toggling to be a constructor arg since most of our users
    will be using from_port() or from_socket_file() instead.

commit 0c4cb0d9c0178a353df4894bb87ccb835bca2ace
Merge: 954bced 7d4efa2
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 4 00:11:59 2013 -0800

    Revising the tutorials
    
    Changing our tutorial page to be a friendly listing of the tutorials rather
    than a wall of text. Also rewrote our present tutorials and their corresponding
    tests.

commit 7d4efa2102727a1135bd19204e7d35f7b8900ec4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 4 00:10:07 2013 -0800

    Tests for descriptor tutorial examples
    
    Unit tests for our new 'mirror, mirror' tutorial examples, and writing our
    prior tests to conform with the way-less-sucky style that separates the mocking
    from the example.

commit a86d250b6702f7dae9ed74778a6e0a5cb2f9358a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 3 23:11:30 2013 -0800

    Rewriting the descriptor tutorial
    
    Replacing the 'Mirror, Mirror' tutorial with a new one that gives a better
    overview of the various descriptors and how to get/use them. This keeps the old
    example (listing the fastest exits), but otherwise is a full rewrite.

commit fce9b2e92f4d6bf9a01431f993fd0e42dcd6f6ad
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 3 16:12:24 2013 -0800

    Adding microdescriptors to our api docs

commit 478ba3a44f3291367eeaea9ba9ce0f4cab24f517
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 23 11:49:02 2013 -0800

    Cleaning up 'The Little Engine that Could' tutorial test
    
    Of all our tests the tutorial example tests stick out as being... just
    aweful. They work I guess but their mix of heavy mocking and the example being
    tested is both damn confusing and unmaintainable.
    
    Breaking it up so the mocking is entirely separate from the tutorial code.

commit 1749855559044f2547920fa206f72c6fdf7c86f4
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Feb 22 08:53:21 2013 -0800

    Mocking stdout for tutorial tests
    
    Mocking stdout so the tests can be a little more faithful to the tutorial
    examples.

commit 63bad222dc4de6b7e3c8510e4704ac7af0cabcdb
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Feb 22 08:55:14 2013 -0800

    Revising 'The Little Relay that Could' tutorial
    
    Changing small bits if the tutorial page and 'hello world' tutorial, adding a
    little detail here and cleaning up the wording there.

commit a3e486856810e99cfa065df9ebf24961ec7a3bff
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Feb 19 09:23:11 2013 -0800

    Rearranging contents
    
    The contents determine the next/previous links in the page's upper right. In
    general these don't make sense (there's not really a logical order to our API),
    but for the tutorial it's nice if they link to each other.

commit f4d2b20b2d954852faa59d39ce9f38f7cb11dcd2
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Feb 18 22:12:01 2013 -0800

    Moving tutorials to sub-pages
    
    Making pages for each of the tutorials and making tutorial.rst a pretty listing
    similar to the download page. This should hopefully make this page far more
    approachable by newcomers.

commit 0fde6ad67f7250b3de573726343844eb7ea573fe
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Feb 18 08:58:10 2013 -0800

    Moving download images into subdirectory
    
    I'm about to use '_static/section' for tutorial images, so moving the download
    page's resources to keep them separate.

commit 954bced4f16b05bd6a4923cfe3a8297f0e3347a6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 3 15:08:01 2013 -0800

    Only first 'transport' line was being used
    
    The extrainfo descriptor parser only utilized the first 'transport' line,
    causing further lines to be ignored. Caught by asn.

commit 7f3f8722d99b2747cafd2247b14e362c743d5290
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 2 22:16:20 2013 -0800

    Adding get_microdescriptors() method to the Controller
    
    Controller method to fetch all microdescriptors. This is modeled after its
    counterparts for server descriptors and network status documents. However, as
    mentioned in 'https://trac.torproject.org/8323', the controller interface
    presently lacks a method to get them.
    
    In the meantime we're reading them from disk.

commit db19cafb398e9626502cf9d1226803a5fb4cb4b1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 2 19:57:57 2013 -0800

    Replacing get_server_descriptor() and get_network_status() tests
    
    Our integration tests for get_server_descriptor() and get_network_status() have
    been skipped for quite some time due to...
    
    https://trac.torproject.org/7163
    
    Replacing them with a simpler (and hopefully more reliable) test like what
    we're now doing for get_microdescriptor().

commit f5c7c226321f801e448f759e2ba3617c0ef5814b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 2 19:28:06 2013 -0800

    Adding get_microdescriptor() method to the Controller
    
    Adding a method to query individual microdescriptors. This is very similar to
    its server descriptor and network status counterparts.

commit 23d2332b2514fd9194e64cb1859614014087394b
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Feb 28 09:14:01 2013 -0800

    Tor microdescriptor support
    
    Adding parsing support for tor microdesriptors...
    
    https://trac.torproject.org/8253
    
    These have replaced server descriptors as the self-published descriptor content
    tor fetches by default. They're a bit clunckier to use compared to server
    descriptors, and lack much of the information controllers might be interested
    in, but the lighter weight of microdescriptors make them better for the overall
    network.
    
    Next up is to add support for these to our Controller. Unfortunately the tor
    control protocol only supports querying microdescriptors individually...
    
    https://trac.torproject.org/8323

commit d25f946d06a26e2ee0947eca5b1944f190e6299f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 23 12:16:32 2013 -0800

    Notice that server descriptors are no longer available by default
    
    As of Tor 0.2.3.25 relays stopped fetching server descriptors by default. This
    is an important thing to know since it makes get_server_descriptor() and
    get_server_descriptors() useless unless you set 'UseMicrodescriptors 0'.
    
    velope has been emphatically pointing out that we want to discourage stem
    developers from disabling microdescriptors so advising them against it.

commit ec3d0c01c9ba4aa30541ceb19e499fbe732453cb
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Feb 27 07:35:00 2013 -0800

    Don't pass unsupplied optional args to wrapped file
    
    Karsten reports that the descriptor reader is failing for him with...
    
      ParsingFailure: readlines() takes exactly 1 argument (2 given)
    
    This is because to add python 3.x support we had to wrap the parse_file()'s
    file argument with a wrapper. In most cases this wrapper is a passthrough, but
    it converts the read() and readlines() output to unicode.
    
    I based the methods of the wrapper on a StringIO file object, but evidently
    other files (most likely the tarfile) deviate a bit. Addressing the readlines()
    instance and one other that might cause troubles.

commit 3872befa5c62a91eb83d0d1376741b51a91acdb6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 23 16:07:19 2013 -0800

    Renaming controller and socket keyword arguments
    
    Less is more. I'm not sure why I called these keyword arguments 'control_addr',
    'control_port', and 'socket_path' but these will be highly used constructors.
    We should aim to have them be as user friendly a possible, and that means less
    verbose.

commit be9a5323a37ea0f1b7d497d7fc33e101453eb2cf
Author: Karsten Loesing <karsten.loesing@gmx.net>
Date:   Wed Feb 20 12:26:29 2013 +0100

    Change fingerprints and digests to upper-case hex.

commit 13de78a989b4e9ba68ffcecdd6022335baae4b39
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Feb 18 07:42:46 2013 -0800

    Catching unexpected exception from libc.setproctitle()
    
    Sathyanarayanan reports that on OSX libc.setproctitle() can raise an
    AttributeError...
    
    ERROR: test_set_process_name
    ----------------------------------------------------------------------
    Traceback:
      File "/Users/sathya/Documents/stem/test/integ/util/system.py", line 412, in test_set_process_name
        stem.util.system.set_process_name(initial_name)
      File "/Users/sathya/Documents/stem/stem/util/system.py", line 774, in set_process_name
        _set_proc_title(process_name)
      File "/Users/sathya/Documents/stem/stem/util/system.py", line 831, in _set_proc_title
        libc.setproctitle(ctypes.byref(name_buffer))
      File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/ctypes/__init__.py", line 366, in __getattr__
        func = self.__getitem__(name)
      File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/ctypes/__init__.py", line 371, in __getitem__
        func = self._FuncPtr((name_or_ordinal, self))
    AttributeError: dlsym(0x7fff6a41d1e0, setproctitle): symbol not found
    
    https://trac.torproject.org/8266

commit bdbb508ae2ffcaa5480992d33c28ec52a6e61b7e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 17 20:34:11 2013 -0800

    Making get_process_name()'s ps lookup work on OSX
    
    The get_process_name() function uses a couple approaches to determine our
    process name, the first being to query ps. This tends to work pretty well
    across various platforms (barring Windows, of course), but there's a slight
    difference in the header on OSX/BSD. Accounting for this...
    
    Example ps output:
    
    stem git:(master) : ps -p 60437 -o args
    ARGS
    tor --socksport 9050
    
    Caught by Sathyanarayanan on...
    
    https://trac.torproject.org/8266

commit 08529723708ed514f6f5e2489d3c1be8c0e9ae36
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 17 18:09:58 2013 -0800

    Accounting for NULL access by ctypes
    
    Evidently accessing argc can raise a ValueError...
    
    https://trac.torproject.org/8266

commit 1a099106c43e37bf41f5c1f803f27cd2aad270f5
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 17 13:57:15 2013 -0800

    Providing a string when str() is called on descriptors
    
    Python 2.x gets pretty confused when an object's __str__ method provides a
    unicode string. Calling...
    
    >>> str(desc)
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
    UnicodeEncodeError: 'ascii' codec can't encode character u'\xab' in position 28: ordinal not in range(128)
    
    Providing an ascii str in python 2.x and unicode str in python 3.x. Thanks to
    Sathyanarayanan for the catch!

commit d70a2d394f24e471379906e561d80fb47ecd2230
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 17 13:22:54 2013 -0800

    Adding copyright headers
    
    Uggg I hate IP law. As pointed out by Juan on...
    
    https://trac.torproject.org/7954
    
    ... we need copyright headers to properly comply with the requirements of being
    under the LGPL. I'm not looking forward to keeping this up to date (and likely
    won't), but oh well.
    
    I've added the header to all python source files except the unit and integ
    tests (patches welcome if someone wants to spend the time adding those too).
    These headers...
    
    * Declare a copyright from the year of the file's creation to now (2013).
    
    * Include Sean if he worked on it (he is the only contributor that hasn't made
      his contributions public domain to avoid copyright headaches).

commit 4a2e7e74652c2c93654035bede41f203397ed16b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 17 12:25:43 2013 -0800

    Cleaning up our TODO comments
    
    Several of our TODO comments were no longer relevant or could be expanded.

commit 6927e68d80bf78d898afe321a599bf174e116e40
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 16 21:53:24 2013 -0800

    Treat descriptor archive contents as individual files
    
    When the descriptor reader encountered an archive and read non-descriptor
    content it stopped reading. This has caused me almost two weeks of headaches in
    troubleshooting...
    
    https://trac.torproject.org/8049
    
    Changing the reader's behaviour to instead handle each file within the archive
    separately. Thanks to Karsten for catching this!

commit f83c7efce3380c60c81028802e198791aa30dfbd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 16 10:53:39 2013 -0800

    Adding get_archive_path() method to descriptors
    
    We can't use a TarInfo's 'name' attribute for get_path() since that corresponds
    to its location within the archive. That said, I've often wanted both paths so
    both fixing get_path() for tarballs and adding a get_archive_path().

commit 8cdcb0883f0da8bb492da31cfea7155df5e9efed
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Feb 15 09:38:02 2013 -0800

    Dropping the 'path' argument from stem.descriptor.parse_file()
    
    File objects have a 'name' attribute that we can use to guess the path. This
    isn't entirely reliable, but nothing is...
    
    http://stackoverflow.com/questions/2458676/absolute-path-of-a-file-object
    
    The path argument was only there to support the descriptor reader. Now that
    parse_file() is something for our users it's nice to get rid of arguments they
    can't use.

commit 2a952ec98054cf656efd5d9b2bffa1960cdf5b42
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 9 15:29:00 2013 -0800

    Accepting "NEVER" expiration in ADDRMAP events
    
    The expiry value in ADDRMAP events can be 'NEVER'. This is a little troublesome
    since it means that the field might or might not be quoted (making this unique
    among all tor events).
    
    Caught by Desoxy on 'https://trac.torproject.org/8162'.

commit dba169f5cb3d28da29c7c2011ef5de7cc5825d37
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 9 13:07:23 2013 -0800

    Fixing SingleLineResponse interlinking
    
    The SingleLineResponse class wasn't in the module's __all__, causing it to not
    appear in the sphinx output.

commit 74b60f7552b587ea576db0be01e31e0e5531bb7d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 9 12:53:31 2013 -0800

    Improving stem.response.convert() pydocs
    
    The convert() pydocs were pretty clunky. Replacing the listing with a nice
    table mapping the response_type to classes, like what we do elsewhere.

commit 5f8b7b4231bcda7f69296821aa6417fd3e2ce5f3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 9 12:35:11 2013 -0800

    Adding a ControlMessage.from_str() function
    
    In discussions with Mike about using stem for txtorcon a major use has been
    response parsing. Using stem for this is dead easy, but requires a hack. Adding
    a function to negate the need for hackery.

commit f96d5f64a94ecfa9d05b80e2b3a6c7c3ba78bcea
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Feb 8 08:32:44 2013 -0800

    Noting '(Tor_internal)' addresses in the pydocs
    
    Noting that StreamEvents can have '(Tor_internal)' as a target address.
    
    Spec change:
    https://gitweb.torproject.org/torspec.git/commitdiff/3ad9d19e03bd816e1e0f0b9eeb839ee1eedcaedf

commit ea5212863a97efa4f29e47280faf3c9566ec5349
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Feb 8 08:29:22 2013 -0800

    Using numeric 'flag-thresholds' values
    
    Now that the spec has been revised to specify numeric values we can provide
    'flag => int/float' mappings (which are much nicer for our users).
    
    Spec change:
    https://gitweb.torproject.org/torspec.git/commitdiff/52d0eb4858ad3eb191df3afe324f43683467ae22

commit e8784466401a8120669128682b75eaa947e91c7a
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Feb 6 07:46:22 2013 -0800

    Minor stylistic corrections
    
    Couple PEP8 bugs that slipped in concerning spacing between code and inline
    comments.

commit 3687dde657dc8df40ffecadd45c4035ee3d44110
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Feb 6 07:41:30 2013 -0800

    Avoiding static /tmp usage
    
    Our tests had static /tmp paths at a couple places. Issue caught by Dererk and
    patch by Abhishek...
    
    https://trac.torproject.org/7926

commit 4e8aaa4daadb02db1686dde57c42c15aef6821e9
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Feb 5 09:48:12 2013 -0800

    Support for 'flag-thresholds' lines in network status votes
    
    Parsing the new 'flag-thresholds' in network status votes - thanks to Karsten
    for pointing this out.
    
    metrics-lib change:
    https://gitweb.torproject.org/metrics-lib.git/commitdiff/c2a0dbf8bf100a19660ad512b88d93f3d7c18a1e
    
    dir-spec addition:
    https://trac.torproject.org/8165

commit c6a9cde0f4eb5c627f5ba41f9d263be0c5854ae8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 3 22:14:20 2013 -0800

    Allowing for IPv4 'a' lines in router status entries
    
    Karsten reports on ticket #8036 that IPv4 addresses are indeed allowed on a
    router status entry's 'a' line. This is a little unfortunate since it means a
    less friendly attribute but not a big whoop.

commit b8baf77c13154ccb6c43299a36979d812ca7b8be
Author: Karsten Loesing <karsten.loesing@gmx.net>
Date:   Mon Feb 4 06:17:28 2013 +0100

    Try to make minor descriptor versions clearer.

commit 6c99a28e83490537615de9388484c574aa1b85dd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 3 20:43:28 2013 -0800

    Using port lists for addresses_v6 rather than ranges
    
    Huh, I wonder where I got the idea that 'a' lines had port ranges. Dropping
    that. According to the spec the 'a' lines should be parsed in the same way as
    'or-address'. However, I suspect that the spec is a little off here - checking
    if it can contain IPv4 addresses...
    
    Caught by Karsten on...
    
    https://trac.torproject.org/8036

commit d44018a5566fe5c92a326a413768bdd8462c425e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 3 20:01:35 2013 -0800

    Renaming check_whitespace.py to static_checks.py
    
    The check_whitespace.py module no longer... well, checks whitespace. Rather, it
    has become a dumping ground for all of the static checks that we do. Renaming
    it to something more appropriate.

commit ea0b73a5aa221fadafc2ba718a0ef42e151e5ad6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 3 13:00:56 2013 -0800

    Providing alternative methods for parsing a NetworkStatusDocument
    
    Adding support in both the DescriptorReader and parse_file() function for three
    ways of parsing network status documents...
    
    a. Provide the router status entries (ie. the current behavior).
    
    b. Provide the document itself with the router status entries that it contains.
       This has the biggest cost in terms of upfront parsing time and memory usage,
       but provides the caller with everything they might want.
    
    c. Provide the document but skip reading the router status entries. A handy
       option of you just care about the document's header/footer.
    
    Now that we have these capability I'm further simplifying the descriptor API a
    bit. The network status docs encouraged users to use the NetworkStatusDocument
    constructors to achieve option 'b' above, but now that it's in the reader and
    parse_file() there's no reason for them to do that.
    
    Users should now *always* use either the DescriptorReader or parse_file(). If
    they don't then they're off the reservation.

commit 479f535614081f097ee2e39ecde70ffaf658ba25
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 3 11:29:53 2013 -0800

    Dropping Version.meets_requirements() in favour of comparisons
    
    Once upon a time you checked your requirements via simple comparisons...
    
      if my_version >= requirement:
        ... do stuff...
    
    I reluctantly changed this to a meets_requirements() method when we added the
    VersionRequirements class since it was no longer simple comparisons the __cmp__
    method could handle. However, now that we're using rich comparison operators we
    can go back to the nicer style of comparisons. Apologies for any confusion
    this back-and-forth has caused.

commit 30146d77243f7a25d54019d9d2641be258c444f8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 3 11:12:57 2013 -0800

    More succinct python 3 warning for parse_file()
    
    Read speeds and universal newline translation are both addressed by reading in
    binary mode. There's no need to have a separate warning for each.

commit 9cd4c9feb479fe00987526c272a1b2a1853bd295
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 2 12:23:07 2013 -0800

    Using binary mode for the controller socket file
    
    Yay! Now that I have a version of python 3 that doesn't segfault I can finish
    making our integ tests work.
    
    The socket file used for controller connections should be normalized to use
    binary mode. This is its behavior in python 2.x, and in 3.x having it in text
    mode can cause sadness.
    
    Exception in thread Tor Listener:
    Traceback (most recent call last):
      File "/home/atagar/Desktop/Python-3.3.0/Lib/threading.py", line 639, in _bootstrap_inner
        self.run()
      File "/home/atagar/Desktop/Python-3.3.0/Lib/threading.py", line 596, in run
        self._target(*self._args, **self._kwargs)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/control.py", line 573, in _reader_loop
        control_message = self._socket.recv()
      File "/home/atagar/Desktop/stem/test/data/python3/stem/socket.py", line 115, in recv
        return recv_message(socket_file)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/socket.py", line 539, in recv_message
        line = control_file.readline()
      File "/home/atagar/Desktop/Python-3.3.0/Lib/codecs.py", line 300, in decode
        (result, consumed) = self._buffer_decode(data, self.errors, final)
    UnicodeDecodeError: 'utf-8' codec can't decode byte 0xc3 in position 2005: invalid continuation byte
    
    After addressing this and a few encoding issues the controller integ tests now
    pass, but after we're done testing python spews out a dump following...
    
    *** glibc detected *** python3: munmap_chunk(): invalid pointer: 0x097f1620 ***
    
    At this point I'm pretty well persuaded that the python 3.x series leaves
    something to be desired in terms of stability.

commit 3930f1f1eda99bcf22fe574b95665f8ef47ffdc4
Merge: 7921a46 2b2a645
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 2 10:12:02 2013 -0800

    Python 3.x support
    
    Adding support for the python 3.x series. You can install the python 3 version
    of stem by running...
    
    python3 setup.py install
    
    The 2to3 conversion can be tested trough run_tests.py with the '--python3'
    argument. It passes all of the unit tests and the integ tests... er, don't
    technically fail. However, python 3.2 has a bug causing a segfault when it gets
    to the BaseController integ tests. Filed a ticket about it...
    
    http://bugs.python.org/issue17105
    
    However, stem's descriptor functionality checks out and this issue has likely
    been addressed in later python releases so there's little point to hold off on
    merging.
    
    Ticket for python 3 support...
    
    https://trac.torproject.org/7843

commit 2b2a645ac778570d99961f9b9592769fdc9caa10
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Feb 1 07:48:21 2013 -0800

    Converting cookie auth token to unicode
    
    Well, this is dumb. Making a formatted string with ascii bytes includes the b''
    wrapper in python 3. This broke our authentication calls, making calls like...
    
    AUTHENTICATE b'd55e81eb9c3a1e22a2db919ec2efd22df4aeb88ee0ab3d10e64dbb2450d06921'
    
    Converting the token to unicode to avoid this.
    
    ======================================================================
    ERROR: test_authenticate_cookie
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/connection/authentication.py", line 400, in _check_auth
        stem.connection.authenticate_cookie(control_socket, auth_arg)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/connection.py", line 604, in authenticate_cookie
        raise CookieAuthRejected(str(auth_response), cookie_path, False, auth_response)
    stem.connection.CookieAuthRejected: Invalid hexadecimal encoding.  Maybe you tried a plain text password?  If so, the standard requires that you put it in double quotes.
    
    During handling of the above exception, another exception occurred:
    
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/connection/authentication.py", line 303, in test_authenticate_cookie
        self._check_auth(auth_type, auth_value)
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/connection/authentication.py", line 411, in _check_auth
        failure_msg = _get_auth_failure_message(auth_type)
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/connection/authentication.py", line 99, in _get_auth_failure_message
        raise ValueError("No methods of authentication. If this is an open socket then auth shouldn't fail.")
    ValueError: No methods of authentication. If this is an open socket then auth shouldn't fail.

commit 384411b26dd140e3d4b181f3c2b1a48057238096
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Feb 1 07:16:25 2013 -0800

    Replacing file() with open()
    
    I'm not sure why we were using file() at one point rather than open(), but it
    makes python 3 sad...
    
    ======================================================================
    ERROR: test_authenticate_cookie
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/connection/authentication.py", line 303, in test_authenticate_cookie
        self._check_auth(auth_type, auth_value)
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/connection/authentication.py", line 400, in _check_auth
        stem.connection.authenticate_cookie(control_socket, auth_arg)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/connection.py", line 583, in authenticate_cookie
        cookie_data = _read_cookie(cookie_path, False)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/connection.py", line 877, in _read_cookie
        with file(cookie_path, 'rb', 0) as f:
    NameError: global name 'file' is not defined

commit a047a74abef0dc9fedf463c093c14defe958d81b
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Feb 1 07:12:04 2013 -0800

    Accounting for ascii/unicode for network status documents
    
    Woohoo! Last descriptor type. Unlike the other descriptor types callers are
    encouraged to sometimes use our NetworkStatusDocument classes directly so
    swapping the input to unicode if we get ascii.
    
    With this all of the descriptor integ tests now pass with python 3!

commit 284797ace4cf0b8e4469454b65809fc66c7ae1d2
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Feb 1 06:57:54 2013 -0800

    Pydoc missing version from descriptor type listing
    
    The 'network-status-microdesc-consensus-3' listing was missing the '1.0'.

commit 4815afe52c4af2d4cb3d687b35b7012239d7471b
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 31 09:54:27 2013 -0800

    Using stem.descriptor.parse_file() for extrainfo integ tests
    
    Going through parse_file() so we do the proper unicode conversion.
    
    ======================================================================
    ERROR: test_cached_descriptor
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/descriptor/extrainfo_descriptor.py", line 150, in test_cached_descriptor
        for desc in stem.descriptor.extrainfo_descriptor._parse_file(descriptor_file):
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/extrainfo_descriptor.py", line 155, in _parse_file
        extrainfo_content = stem.descriptor._read_until_keywords("router-signature", descriptor_file)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/__init__.py", line 350, in _read_until_keywords
        line_match = KEYWORD_LINE.match(line)
    TypeError: can't use a string pattern on a bytes-like object

commit 3afa4346d102d99eab93929be738cdb594105d9b
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 31 09:44:18 2013 -0800

    Using binary mode when reading descriptors
    
    Now Damian, repleat after me: text mode is bad.
    
    In python 2.x text mode and binary mode seem to be indistinguishable, but in
    python 3 there's one tiny little difference: text mode is around 33x slower.
    The integ test that read the cached-consensus took over five minutes (by
    comparison to ten seconds with python 2.7), and in one case simply hung for
    twenty minutes before I killed it.
    
    I'm not aware of any disadvantage to using binary mode, so opting for that.

commit d9d46cb5b35f586bcf8025cb7682c4ad79e57e15
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 31 08:31:53 2013 -0800

    Skipping deletion of pyc in __pycache__
    
    I disabled the deletion of orphaned pyc files when testing python 3 but on
    reflection that wasn't enough. Python 2.x test runs still delete the python 3
    bytecode. Changing the orphaned check to skip those files.

commit bec2e9721e228f73c8b1438f1d398182893df27f
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 31 08:22:38 2013 -0800

    Fixing server descriptor test expecting unicode
    
    One of the server descriptor integ tests had a failing assertion because the
    expected text was ASCII bytes and the descriptor content was unicode. Fixing
    the test and moving the to_unicode helper to str_tools where it belongs.
    
    ======================================================================
    FAIL: test_non_ascii_descriptor
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/descriptor/server_descriptor.py", line 221, in test_non_ascii_descriptor
        self.assertEquals(expected_contact, desc.contact)
    AssertionError: '2048R/F171EC1F Johan BlÃ¥bÃ¤ck ã\x81\x93ã\x82\x93ã\x81«ã\x81¡ã\x81¯' != '2048R/F171EC1F Johan Blåbäck こんにちは'
    - 2048R/F171EC1F Johan BlÃ¥bÃ¤ck ããã«ã¡ã¯
    + 2048R/F171EC1F Johan Blåbäck こんにちは

commit f4ee5ab36ee512f7cb904bb371f593f7e2f1f8a6
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 30 09:39:45 2013 -0800

    Checking that to_bytes has unicode before converting
    
    Adding a check to the to_bytes() helper so we don't attempt to convert ASCII
    bytes to ASCII bytes (which doesn't work so well).
    
    ======================================================================
    ERROR: test_old_descriptor
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/descriptor/server_descriptor.py", line 120, in test_old_descriptor
        desc = stem.descriptor.server_descriptor.RelayDescriptor(descriptor_contents)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/server_descriptor.py", line 642, in __init__
        self._validate_content()
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/server_descriptor.py", line 687, in _validate_content
        key_der_as_hash = hashlib.sha1(stem.util.str_tools.to_bytes(key_as_bytes)).hexdigest()
      File "/home/atagar/Desktop/stem/test/data/python3/stem/util/str_tools.py", line 73, in to_bytes
        return _to_bytes(msg)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/util/str_tools.py", line 54, in _to_bytes
        return codecs.latin_1_encode(msg)[0]
    TypeError: Can't convert 'bytes' object to str implicitly

commit 46f1bfb98cee79f9a00b76513d2c70df1824bb33
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 30 09:36:09 2013 -0800

    Changing is_python_2* prereq checks to include python 3
    
    The is_python_26 and is_python_27 were checking if we were 2.6-2.x or 2.7-2.x.
    On reflection it makes more sense for these to be '2.y and above' checks rather
    than '2.y and above in the 2.x series'.

commit 6c855996c0daed3b6db222996dd651534c134c0d
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 30 09:32:03 2013 -0800

    Providing ASCII bytes to hashlib.sha1()
    
    Another unicode/ASCII bytes conversion issue...
    
    ======================================================================
    ERROR: test_metrics_descriptor
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/descriptor/server_descriptor.py", line 89, in test_metrics_descriptor
        self.assertEquals("2C7B27BEAB04B4E2459D89CA6D5CD1CC5F95A689", desc.digest())
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/server_descriptor.py", line 666, in digest
        digest_hash = hashlib.sha1(for_digest)
    TypeError: Unicode-objects must be encoded before hashing

commit 9ff618eebed92e5646090ce3faa802763a5f7720
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 30 09:10:18 2013 -0800

    Skipping newline translation for descriptor integ tests
    
    Using a custom open() call for python 3's integ tests to prevent newline
    translation (and the resulting test failures).

commit 8a3cfb57e1982d26bafe6289f8ed9dea47192604
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 30 08:53:06 2013 -0800

    Checking for 2to3 and python3 when needed
    
    Warning the user if 2to3 or python3 aren't in our PATH when the user provides
    the '--python3' testing argument.

commit 0288267be5a325aadd078a487751cf4ce83a2f73
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 30 08:24:59 2013 -0800

    Skip universal newline translation in descriptor reader
    
    Python 3 introduces universal newline translation, converting '\n', '\r', and
    '\r\n' into the local system's newline style. This is a really neat feature and
    will solve many-a-headaches... but not for us. We conform to the tor spec which
    specifies when CRLF appears verses other newline types.
    
    Universal newline translation broke our ability to read the
    'cr_in_contact_line' example which has multiple '\r' within a contact line
    (https://trac.torproject.org/5637). Fixing the reader to disable newline
    translation and adding a warning to our parse_file() pydocs.

commit d0f4a0c429b1e3a2bead48403605bb361ec66832
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 30 08:07:50 2013 -0800

    Adding --python3 to the run_tests.py help output

commit b4c4835d957463d6a453d9d7f7ad6007627abc96
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 29 09:03:30 2013 -0800

    Normalizing descriptor handling as unicode
    
    Our python 3 descriptor integration tests were having troubles due to the
    ASCII/unicode switch. Adding a file interceptor so we always parse descriptors
    as unicode.
    
    Yes, yes, I know. Ewwww. I definitely don't like the _UnicodeReader helper so
    suggestions welcome for an alternative.

commit 9d53a2854178d12222480958795876e45ec0f1d6
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 29 08:54:03 2013 -0800

    Improving error output for test_skip_nondescriptor_contents
    
    The descriptor reader's test_skip_nondescriptor_contents integ test is
    especially tickle since it can fail whenever there's new non-descriptor content
    in the test/integ/descriptor/data directory.
    
    Improving its error output to say what it expected and got rather than simply
    saying that the counts mismatch.

commit e99dc2b618cb33d5e23166510d224429b8f67046
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 19:53:32 2013 -0800

    Accounting for tarfile.is_tarfile() raising AttributeErrors
    
    When we lack read permissions the tarfile.is_tarfile() funcion raises an
    AttributeError with python 3 (rather than an IOError as it does with python 2).
    This is most likely a bug...
    
    http://bugs.python.org/issue17059
    
    Working around it for now.

commit 313e5251da4a390fa283c440ffbe0f0ef048ed38
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 19:31:42 2013 -0800

    Skipping orphaned file check if testing with python 3
    
    The orphaned *.pyc check is both broken with python 3 and pointless since it's
    an exported copy of the python 2 codebase (we're not acually making changes to
    it).

commit 2850199fc513fc4a276e16eaa59ccbc2dc69ea23
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 19:21:20 2013 -0800

    Process renaming requires ASCII bytes
    
    Yet another unicode/ASCII compatability issue...
    
    ======================================================================
    ERROR: test_set_process_name
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/util/system.py", line 409, in test_set_process_name
        stem.util.system.set_process_name("stem_integ")
      File "/home/atagar/Desktop/stem/test/data/python3/stem/util/system.py", line 764, in set_process_name
        _set_prctl_name(process_name)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/util/system.py", line 809, in _set_prctl_name
        name_buffer.value = process_name
    TypeError: str/bytes expected instead of str instance
    
    During handling of the above exception, another exception occurred:
    
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/util/system.py", line 412, in test_set_process_name
        stem.util.system.set_process_name(initial_name)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/util/system.py", line 764, in set_process_name
        _set_prctl_name(process_name)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/util/system.py", line 809, in _set_prctl_name
        name_buffer.value = process_name
    TypeError: str/bytes expected instead of str instance

commit 7d1e8a4ffc08ca3b5e7fb34d3b025ec3c8755497
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 19:08:40 2013 -0800

    Returning unicode from stem.util.system.call()
    
    The call() method is the basic call for shelling out and, like tor's stdout
    earlier, its output was ASCII bytes. This is fine in python 2.x, but causes
    problems for python 3.x. Both are fine with this being unicode output so going
    with that.
    
    ======================================================================
    ERROR: test_get_cwd
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/util/system.py", line 334, in test_get_cwd
        self.assertEquals(None, stem.util.system.get_cwd(99999))
      File "/home/atagar/Desktop/stem/test/data/python3/stem/util/system.py", line 528, in get_cwd
        elif results[0].endswith("No such process"):
    TypeError: expected an object with the buffer interface

commit 6a85a34c753f34c7cc79f4ad4e9f67d92bce1046
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 19:01:32 2013 -0800

    Converting str stats for proc tests
    
    The proc tests were comparing strings with floats. It coincidentally passed,
    but wasn't really testing what we wanted. In python 3.x the comparisons fail
    due to the type mismatch.
    
    ======================================================================
    ERROR: test_get_stats
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/util/proc.py", line 74, in test_get_stats
        self.assertTrue(utime > 0)
    TypeError: unorderable types: str() > int()

commit c2b92450016ccfbe46edec151136c8a8a9f694cb
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 18:18:45 2013 -0800

    Disabling newline translation in our socket
    
    In python 3 the socket file object tries to be helpful by translating newline
    characters. By default '\n', '\r', and '\r\n' are all translated to '\n'. Tor
    uses '\r\n' newlines and we check for this so disabling universal newline
    translation.
    
    ======================================================================
    ERROR: test_get_pid_by_port_netstat
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/stem/connection.py", line 321, in authenticate
        protocolinfo_response = get_protocolinfo(controller)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/connection.py", line 800, in get_protocolinfo
        protocolinfo_response = _msg(controller, "PROTOCOLINFO 1")
      File "/home/atagar/Desktop/stem/test/data/python3/stem/connection.py", line 837, in _msg
        return controller.recv()
      File "/home/atagar/Desktop/stem/test/data/python3/stem/socket.py", line 114, in recv
        return recv_message(socket_file)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/socket.py", line 511, in recv_message
        raise stem.ProtocolError("All lines should end with CRLF")
    stem.ProtocolError: All lines should end with CRLF
    
    During handling of the above exception, another exception occurred:
    
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/util/system.py", line 245, in test_get_pid_by_port_netstat
        elif not runner.is_ptraceable():
      File "/home/atagar/Desktop/stem/test/data/python3/test/runner.py", line 413, in is_ptraceable
        tor_version = self.get_tor_version()
      File "/home/atagar/Desktop/stem/test/data/python3/test/runner.py", line 567, in get_tor_version
        control_socket = self.get_tor_socket()
      File "/home/atagar/Desktop/stem/test/data/python3/test/runner.py", line 534, in get_tor_socket
        stem.connection.authenticate(control_socket, CONTROL_PASSWORD, self.get_chroot())
      File "/home/atagar/Desktop/stem/test/data/python3/stem/connection.py", line 323, in authenticate
        raise IncorrectSocketType("unable to use the control socket")
    stem.connection.IncorrectSocketType: unable to use the control socket

commit 6219ed6e6c517e0ce7d25b5d54355ed85329615e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 18:02:49 2013 -0800

    Making sure that the socket's file is writeable
    
    In python 2.x the socket.makefile() method *should* only be readable
    (according to the default argument) but I've been using it for write operations
    too all this time. In python 3.x doing this raises "io.UnsupportedOperation:
    not writable" so guess we gotta fix it now. :)
    
    ======================================================================
    ERROR: test_get_connections
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/integ/util/proc.py", line 92, in test_get_connections
        elif not test.runner.get_runner().is_ptraceable():
      File "/home/atagar/Desktop/stem/test/data/python3/test/runner.py", line 413, in is_ptraceable
        tor_version = self.get_tor_version()
      File "/home/atagar/Desktop/stem/test/data/python3/test/runner.py", line 567, in get_tor_version
        control_socket = self.get_tor_socket()
      File "/home/atagar/Desktop/stem/test/data/python3/test/runner.py", line 534, in get_tor_socket
        stem.connection.authenticate(control_socket, CONTROL_PASSWORD, self.get_chroot())
      File "/home/atagar/Desktop/stem/test/data/python3/stem/connection.py", line 321, in authenticate
        protocolinfo_response = get_protocolinfo(controller)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/connection.py", line 800, in get_protocolinfo
        protocolinfo_response = _msg(controller, "PROTOCOLINFO 1")
      File "/home/atagar/Desktop/stem/test/data/python3/stem/connection.py", line 836, in _msg
        controller.send(message)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/socket.py", line 82, in send
        send_message(self._socket_file, message, raw)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/socket.py", line 422, in send_message
        control_file.write(message)
    io.UnsupportedOperation: not writable

commit 1e2e6204eb8ba618e79c7bb1459e53e5154147fa
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 17:43:22 2013 -0800

    Socket closures raise a ValueError in python 3
    
    Evidently reading from a closed socket file raises a ValueError rather than
    socket.error in python 3...
    
    Exception in thread Tor Listener:
    Traceback (most recent call last):
      File "/usr/lib/python3.2/threading.py", line 736, in _bootstrap_inner
        self.run()
      File "/usr/lib/python3.2/threading.py", line 689, in run
        self._target(*self._args, **self._kwargs)
      File "stem/control.py", line 573, in _reader_loop
        control_message = self._socket.recv()
      File "stem/socket.py", line 114, in recv
        return recv_message(socket_file)
      File "stem/socket.py", line 467, in recv_message
        line = control_file.readline()
      File "/usr/lib/python3.2/socket.py", line 269, in readinto
        self._checkClosed()
    ValueError: I/O operation on closed file.

commit 7507edec828603a860d9ebf0cbb049060d32fc3e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 17:41:08 2013 -0800

    Converting tor initialization lines to unicode
    
    The stdout of the tor process is an ASCII byte stream, causing our following
    regex search() call to error.

commit 8b95332960e5343dd9b716ef25682548f0de9d7e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 16:32:06 2013 -0800

    Mocking for open() broken in python 3
    
    In python 2.x there's both __builtin__.open() and io.open(), but in python 3.x
    there's just io.open() (which is aliased by builtin). Accounting for this when
    we mock it for the descriptor reader unit tests.
    
    I'm a little puzzled why the mocking wasn't failing more spectacularly before
    (the __builtin__ import should always fail on python 3), but oh well - fixed
    now.
    
    With this the unit tests finally pass with python 3!!!
    
    ======================================================================
    ERROR: test_load_processed_files
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/descriptor/reader.py", line 52, in test_load_processed_files
        self.assertEquals(expected_value, stem.descriptor.reader.load_processed_files(""))
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/reader.py", line 179, in load_processed_files
        with open(path) as input_file:
    IOError: [Errno 2] No such file or directory: ''
    
    ======================================================================
    FAIL: test_get_connections
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/util/proc.py", line 171, in test_get_connections
        self.assertEquals(expected_results, proc.get_connections(pid))
    AssertionError: Lists differ: [('17.17.17.17', 4369, '34.34.... != []
    
    First list contains 2 additional elements.
    First extra element 0:
    ('17.17.17.17', 4369, '34.34.34.34', 8738)
    
    + []
    - [('17.17.17.17', 4369, '34.34.34.34', 8738),
    - ]

commit 8d5d2f14edf92c7240aae0b7894fea6e27b41699
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 16:17:34 2013 -0800

    Converted _required_fields() encounter TypeErrors
    
    The server and extrainfo descriptors' _required_fields() functions get
    converted by 2to3 into a line that tries to concatenate a list with a tuple
    (raising a TypeError). On reflection the converted version which uses list
    comprehension is a little nicer than filter(), so using a variant of that which
    makes everyone happy.

commit 264c9e673f71dde707a03ec6b24931f1bd0f3fa4
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 16:02:06 2013 -0800

    Accounting for different exception type from base64.b64decode()
    
    When the base64.b64decode() function encounters a value with malformed padding
    it raises a TypeError in python 2.x and a binascii.Error in python 3.x.
    
    ======================================================================
    ERROR: test_malformed_fingerprint
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/descriptor/router_status_entry.py", line 231, in test_malformed_fingerprint
        self._expect_invalid_attr(content, "fingerprint")
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/descriptor/router_status_entry.py", line 525, in _expect_invalid_attr
        entry = RouterStatusEntryV3(content, False)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/router_status_entry.py", line 330, in __init__
        super(RouterStatusEntryV3, self).__init__(content, validate, document)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/router_status_entry.py", line 150, in __init__
        self._parse(entries, validate)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/router_status_entry.py", line 337, in _parse
        _parse_r_line(self, value, validate, True)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/router_status_entry.py", line 496, in _parse_r_line
        desc.fingerprint = _decode_fingerprint(r_comp[1], validate)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/router_status_entry.py", line 703, in _decode_fingerprint
        identity_decoded = base64.b64decode(stem.util.str_tools.to_bytes(identity))
      File "/usr/lib/python3.2/base64.py", line 92, in b64decode
        return binascii.a2b_base64(s)
    binascii.Error: Incorrect padding

commit cc3f1be711665ab203f8a657a8400e2828bb9aef
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 13:41:01 2013 -0800

    Dropping direct __dict__ usage
    
    Using an object's __dict__ attribute is generally thought to be a bad idea, and
    indeed it bites us from time to time. Functions like hasattr(), getattr(),
    setattr(), and vars() makes direct __dict__ usage unnecessary. This also fixes
    a python 3 mocking issue...
    
    ======================================================================
    ERROR: test_examples
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/descriptor/networkstatus/document_v3.py", line 129, in test_examples
        with support_with(io.StringIO(content)) as consensus_file:
      File "/home/atagar/Desktop/stem/test/data/python3/test/mocking.py", line 303, in support_with
        obj.__dict__["__enter__"] = return_value(obj)
    AttributeError: '_io.StringIO' object has no attribute '__dict__'

commit de9ac90b6db59f29a42453b829a66dff7b3b1767
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 13:17:48 2013 -0800

    NetworkStatus' meets_consensus_method() comparisons didn't account for None
    
    Another spot where comparison with None fails with python 3...
    
    ======================================================================
    ERROR: test_with_directory_authorities
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/descriptor/networkstatus/document_v3.py", line 800, in test_with_directory_authorities
        document = NetworkStatusDocumentV3(content)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/networkstatus.py", line 509, in __init__
        self._footer = _DocumentFooter(document_file, validate, self._header)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/networkstatus.py", line 776, in __init__
        if header.meets_consensus_method(9):
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/networkstatus.py", line 581, in meets_consensus_method
        return bool(self.consensus_method >= method or [x for x in self.consensus_methods if x >= method])
    TypeError: unorderable types: NoneType() >= int()

commit b96f434623cd829dc8e143e1e685167facad4e22
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 12:51:57 2013 -0800

    Python 3 can't do comparison with other types
    
    Another change in python 2 => 3 is that comparison ('<' and '>') can no longer
    work with None, ints, and other types...
    
    In python 2:
    
    >>> 5 > None
    True
    
    In python 3:
    
    >>> 5 > None
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
    TypeError: unorderable types: int() > NoneType()
    
    I expected the 'unorderable types' error would cause it to use the Version's
    comparison function but evidently that's not the case. Rather, when using
    python 3 callers need to check the values' type first (there's nothing we can
    do to make the comparison work). Personally I think that this is pretty stupid,
    but that's the workaround done by python devs on...
    
    http://bugs.python.org/issue13545
    
    Oh well. Dropping the test that comparison with None and ints work.
    
    ======================================================================
    ERROR: test_nonversion_comparison
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/version.py", line 125, in test_nonversion_comparison
        self.assertTrue(test_version > None)
    TypeError: unorderable types: Version() > NoneType()
    
    ----------------------------------------------------------------------
    Ran 9 tests in 0.043s

commit 4ccbe411ff3320a09e21baa44280c357fd44df6e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 12:41:50 2013 -0800

    The min/max() functions can't accept None in python 3
    
    In python 2:
    
    >>> max(5, None)
    5
    
    In python 3:
    
    >>> max(5, None)
    Traceback (most recent call last):
    File "<stdin>", line 1, in <module>
    TypeError: unorderable types: NoneType() > int()
    
    ======================================================================
    ERROR: test_comparison
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/version.py", line 107, in test_comparison
        self.assert_version_is_equal("0.1.2", "0.1.2.0")
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/version.py", line 252, in assert_version_is_equal
        self.assertEqual(version1, version2)
      File "/usr/lib/python3.2/unittest/case.py", line 643, in assertEqual
        assertion_func(first, second, msg=msg)
      File "/usr/lib/python3.2/unittest/case.py", line 633, in _baseAssertEqual
        if not first == second:
      File "/home/atagar/Desktop/stem/test/data/python3/stem/version.py", line 214, in __eq__
        return self._compare(other, lambda s, o: s == o)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/version.py", line 197, in _compare
        my_version = max(0, self.__dict__[attr])
    TypeError: unorderable types: NoneType() > int()

commit e6e5005b8afd9b7e0a1239a92d7c7a3f39e3ef42
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 12:28:59 2013 -0800

    Using byte string for binascii.a2b_hex in python 3
    
    Like base64.b64decode(), binascii.a2b_hex() expects a byte string.
    
    ======================================================================
    ERROR: test_valid_response
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/response/authchallenge.py", line 31, in test_valid_response
        stem.response.convert("AUTHCHALLENGE", control_message)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/response/__init__.py", line 117, in convert
        message._parse_message(**kwargs)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/response/authchallenge.py", line 40, in _parse_message
        self.server_hash = binascii.a2b_hex(value)
    TypeError: 'str' does not support the buffer interface
    
    ----------------------------------------------------------------------
    Ran 2 tests in 0.009s

commit 2e7913aa6c59d147ef5ae926d456c54ec459dd52
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 27 12:17:43 2013 -0800

    Using byte string for base64.b64decode in python 3
    
    First correction for the ASCII -> Unicode switch over in python 3.
    
    ======================================================================
    ERROR: test_get_network_status
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/router_status_entry.py", line 702, in _decode_fingerprint
        identity_decoded = base64.b64decode(identity)
      File "/usr/lib/python3.2/base64.py", line 83, in b64decode
        raise TypeError("expected bytes, not %s" % s.__class__.__name__)
    TypeError: expected bytes, not str
    
    During handling of the above exception, another exception occurred:
    
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/control/controller.py", line 247, in test_get_network_status
        router = stem.descriptor.router_status_entry.RouterStatusEntryV2(desc)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/router_status_entry.py", line 254, in __init__
        super(RouterStatusEntryV2, self).__init__(content, validate, document)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/router_status_entry.py", line 149, in __init__
        self._parse(entries, validate)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/router_status_entry.py", line 261, in _parse
        _parse_r_line(self, value, validate, True)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/router_status_entry.py", line 495, in _parse_r_line
        desc.fingerprint = _decode_fingerprint(r_comp[1], validate)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/descriptor/router_status_entry.py", line 707, in _decode_fingerprint
        raise ValueError("Unable to decode identity string '%s'" % identity)
    ValueError: Unable to decode identity string '/96bKo4soysolMgKn5Hex2nyFSY='
    
    ----------------------------------------------------------------------
    Ran 9 tests in 0.063s

commit 4f108d1dffd2e23db8ced6d4cb755a336a0767f3
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 25 09:57:54 2013 -0800

    Including signature method in DocumentSignature's comparison
    
    The DocumentSignature's method argument is new-ish, and its comparison function
    wasn't taking it into account. Both fixing this and using the default method
    (sha1) in the mock instance so our tests still pass.

commit 5914ae523eb393383a120cd24c12c508e0ffaeac
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 22 09:27:45 2013 -0800

    Using rich comparison functions
    
    Python 3.x drops support for __cmp__, requiring that users opt for rich
    comparison methods instead...
    
    http://python3porting.com/problems.html#unorderable-types-cmp-and-cmp
    
    ======================================================================
    ERROR: test_event_listening
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/control/controller.py", line 285, in test_event_listening
        self.assertRaises(InvalidRequest, self.controller.add_event_listener, mocking.no_op(), EventType.BW)
      File "/usr/lib/python3.2/unittest/case.py", line 574, in assertRaises
        callableObj(*args, **kwargs)
      File "/home/atagar/Desktop/stem/test/data/python3/stem/control.py", line 1438, in add_event_listener
        if not self.get_version().meets_requirements(event_version):
      File "/home/atagar/Desktop/stem/test/data/python3/stem/version.py", line 175, in meets_requirements
        return self >= requirements
    TypeError: unorderable types: Version() >= Version()
    
    This change uncovered a couple bugs where events failed to strip off the
    trailing 'OK'. I'm not entirely sure why the previous comparisons didn't catch
    this, but oh well - fixed now.

commit 137aed3eadb184fd3e96828cd207220d3e09a043
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 21 09:02:18 2013 -0800

    Disabling static checks when using python 3.x
    
    Not only are our static checks useless with the python 3.x codebase (pyflakes
    probably ins't python 3.x compatable and who cares about PEP8 issues with the
    output?), the pyflake check errors...
    
    Traceback (most recent call last):
      File "./test/data/python3/run_tests.py", line 554, in <module>
        style_issues.update(test.check_whitespace.pyflakes_issues(os.path.join(base_path, "stem")))
      File "/home/atagar/Desktop/stem/test/data/python3/test/check_whitespace.py", line 132, in pyflakes_issues
        line_match = re.match("^(.*):(\d+): (.*)$", line)
      File "/usr/lib/python3.2/re.py", line 153, in match
        return _compile(pattern, flags).match(string)
    TypeError: can't use a string pattern on a bytes-like object

commit 62086346044c9eab1ce68d830669559773ef22b9
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 21 08:57:02 2013 -0800

    Accounting for __builtin__ renaming
    
    Python 3.x renames __builtin__ to builtins...
    
    ======================================================================
    ERROR: test_parse_circ_path
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/data/python3/test/unit/control/controller.py", line 24, in setUp
        mocking.revert_mocking()
      File "/home/atagar/Desktop/stem/test/data/python3/test/mocking.py", line 404, in revert_mocking
        if module == __builtin__:
    NameError: global name '__builtin__' is not defined

commit 84df4318abce565fd2f682ee6e12ea2fe9acf072
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 21 08:56:35 2013 -0800

    Adding is_python_3() to prereq module
    
    Something tells me that we'll need this...

commit 54eac73e62739cb40bced307ff6967548eb3af42
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 21 00:42:46 2013 -0800

    Dropping hardcoded python 2.x check
    
    Our prereq module asserted that stem isn't compatable beyond the 2.x series.
    Clearly I'm trying to 'fix' that with this branch.
    
    Ideally we'd keep this check for our python 2.x codebase, and strip it when
    we do a 2to3 conversion. However, that's more error prone than I'd like so
    simply dropping the check entirely.

commit 77b6fa86df045298564b1c0b9f66bd3d71d66d2e
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 21 00:36:24 2013 -0800

    Making the Enum's keys() method provide a list
    
    The 2to3 converter is evidently confused because we have a keys() method that
    provides a tuple rather than a list. Dictionaries provide lists via their
    keys() method so there's probably not a terribly compelling reason for us to do
    differently.

commit f9a2f147b6fe6187e3ba8d1b5bffa5efce14fac1
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 21 00:18:20 2013 -0800

    Making static checks skip the data directory
    
    Our static checkers (most noticeably pep8 and pyflakes) run over the test data
    directory. This was fine until I started using it for the python3 export. Now
    that it has python code these checkers have a lot to say about it, none of it
    helpful.
    
    Simply ignoring the output if it concerns the data directory. This is a sucky
    solution since pyflakes/pep8 are still running over an extra copy of our
    codebase, taking a noticeable bit of extra time. However, pyflakes at least
    lacks any options for ignoring directories.
    
    I'll probably simply move our python3 export to another location later if this
    annoys me.

commit 919502c53baff9f823ed537dab86ec930cf2eaf7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 20 23:42:29 2013 -0800

    Python 3 testing support
    
    Adding python 3 support to the tests. This introduces two new testing
    arguments...
    
    --python3     Does a 2to3 export of our codebase and tests against that. If a
                  python3 export already exists then reuses that (... it takes an
                  annoyingly long time to run).
    
    --clean       Wipes our prior python3 export.
    
    Next to start sorting out the 2to3 conversion issues...

commit 7921a46ba8655baa7c4b81d900a9854444675564
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 24 09:21:35 2013 -0800

    Populating BridgeNetworkStatusDocuments with v2 entries
    
    Karsten reports that bridge network status documents have v2 router status
    entries, not v3.
    
    https://trac.torproject.org/8036

commit 0b1d47d92d447614095cf43fdb35214650db8d82
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 24 09:20:47 2013 -0800

    NetworkStatusDocumentV3's params misdocumented as being a list
    
    The params is a dict, not a list. Caught by amj703 on...
    
    https://trac.torproject.org/8036

commit ce8fcf13521567c439696d7e87894eb09e238347
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 23 09:25:03 2013 -0800

    Broken link for bad relays wiki
    
    Wrong format for external links.
    
    /home/atagar/Desktop/stem/stem/__init__.py:
      docstring of stem:56:
      ERROR: Unknown target name: "https://trac.torproject.org/projects/tor/wiki/doc/badrelays".

commit ab992cd1716b4bc9f61660b746281c61dd878ab5
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 23 09:21:53 2013 -0800

    Fixing ExitPolicyRule's get_address_type() pydocs
    
    We had an ending colon rather than a backtick for the AddressType references.
    
    /home/atagar/Desktop/stem/stem/exit_policy.py:
      docstring of stem.exit_policy.ExitPolicyRule.get_address_type:1:
      WARNING: Inline interpreted text or phrase reference start-string without end-string.
    
    /home/atagar/Desktop/stem/stem/exit_policy.py:
      docstring of stem.exit_policy.ExitPolicyRule.get_address_type:3:
      WARNING: Inline interpreted text or phrase reference start-string without end-string.

commit 2fe1e2a9ee709f745fb65d2716f6b42c7ba24b28
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 23 09:16:24 2013 -0800

    Fixing extend_circuit exception pydocs
    
    Correcting a couple mistakes in extend_circuit's pydocs...
    
    /home/atagar/Desktop/stem/stem/control.py:
      docstring of stem.control.Controller.extend_circuit:30:
      ERROR: Unexpected indentation.
    
    /home/atagar/Desktop/stem/stem/control.py:
      docstring of stem.control.Controller.extend_circuit:31:
      WARNING: Block quote ends without a blank line; unexpected unindent.

commit 5bcb53923214330d3d6b62eea48c930dde65adf8
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 23 09:09:03 2013 -0800

    Linking the RouterStatusEntry's document attr
    
    Fixing the pydocs for the RouterStatusEntry's document attribute. Interlinking
    requires a fully qualified class.

commit ea4964267e519c33b039f88d9d857a8f6452b3a3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 20 23:55:05 2013 -0800

    Empty bridge-ip-versions value broke parser
    
    A 'bridge-ip-versions' line without a value broke our parser...
    
      File "/home/karsten/tasks/task-7828/stem/stem/descriptor/extrainfo_descriptor.py", line 858, in __init__
        super(BridgeExtraInfoDescriptor, self).__init__(raw_contents, validate)
      File "/home/karsten/tasks/task-7828/stem/stem/descriptor/extrainfo_descriptor.py", line 413, in __init__
        self._parse(entries, validate)
      File "/home/karsten/tasks/task-7828/stem/stem/descriptor/extrainfo_descriptor.py", line 882, in _parse
        raise stem.ProtocolError("The bridge-ip-versions should be a comma separated listing of '<protocol>=<count>' mappings: '%s' (entry = '%s')" % (line, entry, ))
    ProtocolError: The bridge-ip-versions should be a comma separated listing of '<protocol>=<count>' mappings: 'bridge-ip-versions ' (entry = '')
    
    Reported by Karsten on...
    
    https://trac.torproject.org/7967

commit d9553f57fedb4692b44c497942ad1a952fb38a02
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 20 16:37:32 2013 -0800

    Adding validate argument to the DescriptorReader
    
    Request by Aaron for the validate attribute to be available on the reader.

commit 998abb262eb4ed6a3d88a7be865412685ab26d3a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 20 13:59:05 2013 -0800

    Exposing the validate arg through parse_file()
    
    Users may want to opt out of validation, either for performance or to accept
    quasy-malformed content. Giving them that option.

commit 9183f64f2085d0f22d66b6cc381b6e66eef227f7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 20 13:44:28 2013 -0800

    Parsing multiple descriptors from non-cached files
    
    We were parsing single descriptors from metrics archives because... well,
    that's what they contained. This is unintuitive to users making then parsing
    their own descriptors, so always using the _parse_file() functions when able.
    
    An extra advantage is that all server descriptors parsed through parse_file()
    will support server annotations (so things like "@downloaded-at 2012-09-17
    17:44:50" will no longer trip it up).
    
    https://trac.torproject.org/8005

commit 7bd5b49ea27158b44f97e3f5f400aa7df84c2e4b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 19 21:24:54 2013 -0800

    Adding Arch Linux to the download page
    
    Sjon has made a stem package so adding it to our download page.

commit 80f11181966c5390a0aabbb531c20bbcd06a3860
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 19 20:51:12 2013 -0800

    Py_GetArgcArgv unavailable with pypy
    
    Peer reports that stem fails with an AttributeError on pypy. Unsurprisingly
    this is in the process renaming voodoo - adding a try/catch to make this more
    relient.
    
    Traceback (most recent call last):
      File "app_main.py", line 51, in run_toplevel
      File "first_pass.py", line 62, in <module>
        for router in parse_file(consensus_file):
      File "stem/descriptor/__init__.py", line 94, in parse_file
        import stem.descriptor.server_descriptor
      File "stem/descriptor/server_descriptor.py", line 40, in <module>
        import stem.version
      File "stem/version.py", line 65, in <module>
        import stem.util.system
      File "stem/util/system.py", line 74, in <module>
        Py_GetArgcArgv = ctypes.pythonapi.Py_GetArgcArgv
      File "/usr/lib/pypy/lib-python/2.7/ctypes/__init__.py", line 371, in
    __getattr__
        func = self.__getitem__(name)
      File "/usr/lib/pypy/lib-python/2.7/ctypes/__init__.py", line 376, in
    __getitem__
        func = self._FuncPtr((name_or_ordinal, self))
      File "/usr/lib/pypy/lib_pypy/_ctypes/function.py", line 250, in __init__
        ptr = self._getfuncptr([], ctypes.c_int)
      File "/usr/lib/pypy/lib_pypy/_ctypes/function.py", line 421, in _getfuncptr
        self._ptr = cdll.getfunc(self.name, ffi_argtypes, ffi_restype)
    AttributeError: No symbol Py_GetArgcArgv found in library <None>

commit 3d3e69417975499fbe5d2a3eb591d808c6800874
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 19 15:53:06 2013 -0800

    Always processing @type headers
    
    The parse_file() function only consumed the @type annotation when we didn't
    have a descriptor_type nor recognized filename. Changing it so we always
    consume the @type header.
    
    I'm also changing the priority order of parse_file() so if the file could be
    both a metrics archive or cached descriptor (for instance a file named
    'cached-consensus' that starts with a @type annotation) then we process it as a
    metrics archive.

commit ca0375e0c47b8162c94745ec4ae6a648c9c0021c
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 18 22:41:09 2013 -0800

    Clearing request cache on sighup
    
    Sighups can render our cache as being stale. Clearing it.

commit fa6ef6bca6e703e9b69a140ec5abe4c972122872
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 18 09:24:11 2013 -0800

    Adding support for '@type dir-key-certificate-3 1.0'
    
    We already parsed key certificates for network status documents. I didn't know
    though that they had their own @type annotation.
    
    Test data is from...
    
    https://metrics.torproject.org/data/certs.tar.bz2
    https://trac.torproject.org/7987

commit 42ddd84ce3106d350890db064c631eaafe9b8d17
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 18 08:48:57 2013 -0800

    Updating test for the network status example
    
    I change one of the pydoc examples in the networkstatus module to use
    stem.descriptor.parse_file() rather than the module's (now private) parse_file
    method. Updating the unit test for this example.
    
    Updating this test has the side benefit that it now gives us test coverage of
    the descriptor_type argument (I had missed that previously).

commit 8b3e9ed828070b2da9010220e989017668cb994d
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 18 00:04:30 2013 -0800

    Table providing desctriptor_type to class mappings

commit d34b30903a071cc0fd8601c9af8a35da3f57f3d2
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 17 23:28:53 2013 -0800

    Replacing our tuple descriptor_type arg with a str
    
    Changing the type of the descriptor_type arg I added earlier today. A string
    would be more intuitive to use than a tuple.

commit b060be224ef31342f813b8224fcac7f17fdcb3a4
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 17 09:37:37 2013 -0800

    Malformed url in docs
    
    Missing the proper reStructuredText formatting for a url.

commit 5818640e54542ff079b0a9328f82bb0c5897bc9a
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 17 09:28:56 2013 -0800

    Only providing a single parse_file() function
    
    Ok, our interface for parsing descriptors was damn confusing. Users had three
    options...
    
    * stem.descriptor.reader
    
      This was the method I had intended for users to always read descriptor files
      with, but folks keep finding parse_file() first...
    
    * stem.descriptor.parse_file()
    
      I didn't intend for this to be used by our users (I made it to support the
      DescriptorReader). However, it's pretty convenient for reading individual
      descriptor files.
    
    * stem.descriptor.*.parse_file()
    
      The parse_file() in individual descriptor classes definitely weren't intended
      for external users. They were made to support stem.descriptor.parse_file(),
      but turned out to be used most of all...
    
    I'm making the last group private to discourage their usage and changing
    stem.descriptor.parse_file() to be both more capable and user friendly...
    
    * users can now explicitly give it a descriptor type, as per the metrics types
      (https://metrics.torproject.org/formats.html#descriptortypes)
    
    * the path is now optional (it was only required because I expected this to be
      used by the DescriptorReader)
    
    * ... and better pydocs
    
    This breaks backward compatibility, but should greatly help our usability going
    forward. Thanks to Aaron for pointing out the pain points.

commit a85aa8cfcdabfaf2176497abeaf0c90cb80a9624
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 17 08:43:17 2013 -0800

    Moving Flag enum to stem's base module
    
    I'm making an effort to put most general purpose enums in the base module.
    Oddly I never referenced the Flag enum in the descriptor pydocs...

commit 7d013d66db31d9bef398b00e12e2fb78c7222a32
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 17 08:37:58 2013 -0800

    Adding documentation for Flag enums
    
    We were missing docs for the Flag enums along with a couple values
    (BADDIRECTORY and V3DIR).

commit 1fb3a96c75f5eb6f1706ab7d34d582df3ee2cce7
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 16 09:27:27 2013 -0800

    Notifying status listeners of SIGHUPs
    
    Our status listeners should be notified when three things happen...
    
    * we newly connect to a controller
    * we disconnect from a controller
    * tor's state is reset by a SIGHUP signal
    
    We had implemented the first two, but not the third. Correcting this oversight.

commit 5d9c25300e6783b092a64fc66877f39566fc4a4f
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 16 09:21:18 2013 -0800

    Letting add_event_listener() work when unauthenticated
    
    If we called add_event_listener() prior to authentication then we'd throw an
    error when calling get_version() to figure out if we meet the requirements.
    
    Changing the behavior so that we check requirements and issue SETEVENTS if
    able, but if not then simply enquing the listener. It'll then be attached
    during our post-authentication hook.
    
    The change that I'm more interested in this though is that our
    post-authentication hook is less likely to go belly up. If we, say, added a
    listener then attached the Controller to an older tor instance then our
    SETEVENTS call could fail. This was an all-or-nothing call. :(
    
    Changed it so that the hook will re-attach the events that it can, then warn
    about the rest.

commit a0e13fbcc8fd8e0678b7906329d11e7f567839d1
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 16 09:09:06 2013 -0800

    test_get_exit_policy failed if we get a public address
    
    I forgot to account for the exit policy including a 'reject [public_addr]:*'
    entry. This is a bit of a pain both because it's dynamic, and because it may or
    may not be included at all. Accounting for this by just checking the policy
    prefix and suffix instead.

commit 32092205f1be4ccc01ac2cd4bd78a63fd732ae45
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 16 07:38:29 2013 -0800

    Adding a BaseController is_authenticated() method
    
    I've been running into some situations where I need to know if the controller
    is not only connected, but authenticated (since that determines if it's...
    well, usable).
    
    A use case where this could get confused is...
    
    * User gets a ControlSocket instance.
    * Manually authenticates via the send() and recv() methods.
    * Makes a new Controller object from this control socket.
    
    Unfortunately this is *exactly* what our test runner did, so tweaking it so
    that it hits our post_authentication method (I'm surprised that it hadn't
    bitten us before).

commit 1f498f1ce24e149e166ce280872283a61fa76ef4
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 16 07:33:22 2013 -0800

    Moving event version requrements to the version module
    
    Moving the location to the Requirement enum so users can more easily check if
    an event is available or not.

commit f09e5efc22264c8c83bd39b9ae679f3f6beeb752
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 14 08:34:37 2013 -0800

    Dropping expect_alive arg from _notify_status_listeners
    
    On reflection the expect_alive attribute can be implied by the state.
    Simplifying things.

commit 32b9de7c8e2690d4642ac72019c7ed684e3b3576
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 13 21:12:08 2013 -0800

    Caching ExitPolicy's can_exit_to() results
    
    That's odd. I wonder why I didn't add a cache here - can_exit_to() is a method
    that's highly likely to be called repeatedly so if we can get contant time
    lookups then great.

commit b03613cc62d7ae19faa1bbf63e95cffb3547eeae
Merge: 427fce8 716f8a6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 13 17:38:12 2013 -0800

    Controller get_exit_policy() method

commit 716f8a693e9b814da5e2c9df551dbe6768f4f324
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 13 17:19:33 2013 -0800

    Controller method to get the exit policy
    
    What started out with 'we should have a method to get our exit policy' turned
    out to not be as simple as I thought. A relay's exit policy is really the
    combination of three things:
    
    * 'reject private:*, reject [public address]:*' if ExitPolicyRejectPrivate is
      set.
    
    * Our ExitPolicy. Tor provides the torrc value rather than a proper policy
      spec, so I added a get_config_policy() function to account for the
      differences.
    
    * The default exit policy (as per 'GETINFO exit-policy/default').

commit f0ae1eaeec229e51ce8ce223dd1b862d769c1a0c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 13 17:14:55 2013 -0800

    Dropping redundant policy rules
    
    While working with 'reject private:*' entries I realized that
    ExitPolicyRejectPrivate makes for really long, ugly policies. In general this
    is just life - exit policies are more complicated than just the ExitPolicy
    torrc option. Hoever, in the case of 'reject *:*' we can safely boil things
    down.

commit f264a626d682f6b8a4cdea8b58a260b1c1069524
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 13 16:33:03 2013 -0800

    Providing ExitPolicy via get_config_policy()
    
    On reflection it desn't make sense to provide a list of ExitPolicyRules.
    Usually we want a policy itself, and if we do want a rule list then that's
    easily done via 'list(get_config_policy())'.

commit 67c00e3c89b0cebd00ce5f6f19ce194b200eb8f4
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 13 16:19:17 2013 -0800

    Truncating exit policies after a catch-all
    
    It's pointless for an exit policy to contain rules after an 'accept *:*' or
    'reject *:*'. Cropping policies if they go on after that.

commit 54d2c103940b31195b3c91ac440d7f07cb6b9489
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 13 12:57:02 2013 -0800

    Function to parse configuration exit policies
    
    Tor exit policies found in the torrc (and 'GETCONF ExitPolicy') differ slightly
    from the exitpattern definition found in the spec. Adding a get_config_policy()
    function that converts these into proper exit policies.

commit 427fce8bd6777542f234e1738b35ee553853a366
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 12 23:42:11 2013 -0800

    Support process renaming
    
    Renaming our python process from "python our_script.py" to something more
    sensible is a very handy capability to have. Thanks to Jake arm has had this
    for years, which makes it easier to identify in top and ps listings.
    
    We'll want this capability for other controllers so snagging arm's utility and
    cleaning it up.

commit 84664f1d6f388c93ff686dd049dfe48804ead41f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 12 21:07:47 2013 -0800

    Missing files for prior commit
    
    Oops. Did a 'git commit -a' while forgetting that we were adding a couple new
    files.

commit b236ac4e0ba830352c447537be6cf59d85650ae0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 12 21:04:01 2013 -0800

    Support for bridge network status documents
    
    Tor metrics has network status documents for bridges. These are not part of the
    dir-spec, and presently not even in the metrics spec. However, they're trivial
    to parse, consisting of just a 'published' line followed by v3 router status
    entries.
    
    This resolves...
    
    https://trac.torproject.org/7938

commit 98a8842c9ab800f9891a01caf35d02f2dfc60809
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 12 10:46:39 2013 -0800

    Supporting footers in pre-version 9 method documents
    
    All version 3 network status documents can have a footer. The caveat is that
    prior to consensus method 9 they just contained signatures, and after that they
    were a properly marked 'directory-footer' section.
    
    Reproed the issue that Karsten discovered in 'https://trac.torproject.org/7932'
    via our unit tests then fixed.

commit 04391b319fd63fb10c6bf925b86796d2d26ec526
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 11 09:07:35 2013 -0800

    String explanation for FileSkipped exceptions
    
    For #7828 I wrote a little consenses parser script. That script included a very
    simple skip listener...
    
    lambda path, exc: LOGGER.warning("  skipped due to %s" % exc)
    
    "How could it fail?" I asked myself. Yet when I just ran it the listener gave
    me the very helpful warning of "skipped due to ". It turns out that we weren't
    providing our exceptions with a string explanation - go me. :P

commit 153ada322653f95eb21c360d3cac4fe6ad436ed7
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 10 08:37:17 2013 -0800

    Supporting '-legacy' authority entries
    
    Directory authorities have an alternative format if their nickname has a
    '-legacy' suffix. Caught by Karsten on...
    
    https://trac.torproject.org/7866

commit 29af06cd5854ad70290b8066bdb1f3219831c458
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 9 21:27:41 2013 -0800

    Using https link for cloning from git

commit 4267b6c42a0de9546f664b7498bb3f2ca9b065b2
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 9 20:29:19 2013 -0800

    Fixing issues caught by pyflakes
    
    Mostly extra imports, though in one case it caught an error (default arguments
    for get_server_descriptors() and get_network_statuses() would cause a
    NameError).

commit c88e3da8c481dfc4e726a5e3d69bc19ffffddb57
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 9 20:20:28 2013 -0800

    Adding a '--all' argument to the tests
    
    It'll be a whole lot easier to ask patch submitters to run 'run_tests.py --all'
    rather than 'run_tests.py --unit --integ --style'. :)

commit 73e69ac2a612d3eef7b4a6f8efd99063ecb7babb
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 9 20:15:38 2013 -0800

    Integrating pyflakes with our tests
    
    Inspired by txtorcon, using pyflakes for static error checking of our codebase.
    Unlike pylint it has precious few false positives for our codebase. For the
    false positives we do have I've added a 'pylint.ignore' config mapping so they
    can be whitelisted.

commit d29ec4d12567a6af6f483400658d2d2d17769a5a
Merge: d580192 6cd814d
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 7 00:57:28 2013 -0800

    Changing our style to mostly match PEP8
    
    Stem now conforms with PEP8 with the following exceptions...
    
    * four space indentations
    * strict 79 character line widths
    * no spaces around '=' in keyword args
    
    If you're curious about why then see 'test/check_whitespace.py' for a short
    explanation, but I'm not interested in discussing this bike shed's color
    further. If both Sean and Ravi both want one of these to change then I'll
    accept a patch.
    
    This also adds the pep8 style checker to our tests...
    http://pypi.python.org/pypi/pep8
    
    You can run it via the '--style' argument.

commit 6cd814dcb3c24a301f5e4c6396388a424629fd3f
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 7 00:56:47 2013 -0800

    Adding '--style' to run_tests.py's help output

commit 44c018886e0ff43e7eb42f7ec34402a301d40cbf
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 7 00:48:21 2013 -0800

    Skipping E111 and E121 (four space indentations)
    
    I strongly prefer two space indentations. It reads perfectly well and leads to
    shorter lines. I've never heard a coherent argument why four spaces, eight
    spaces, tabs, or any of the other measurement is actually superior.
    
    When it comes to this bike shed I'm keeping the color as it is for now.

commit 48776aee58767f0b8fef3b49d6e77ec65c09681f
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 7 00:42:05 2013 -0800

    Conforming to W293 (blank line contains whitespace)
    
    Finally we get to the main thing that started this whole odyssey. Sed was my
    friend with this one...
    
    find . -name "*.py" | xargs sed -i 's/[ \t]*$//'

commit 26b27b6eff21214b17b4018a16ac97f3bd674734
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 7 00:31:22 2013 -0800

    Conforming to E302 (two blank lines above methods and functions)
    
    Another long one. I hope this is almost done...

commit 1eed64be89d0cf0382234786c020ab4760a4e8e4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 7 00:00:20 2013 -0800

    Conforming to E701 (multiple statements on one line)
    
    Well, that took a while.

commit ed070ee78b2a3d0baed01158b669611646957d97
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 22:48:59 2013 -0800

    Skipping E251 (no spaces around keyword / parameter equals)
    
    I got a little ways through switching us over to this one then stopped. It
    looks aweful and makes the code less readable. If Sean and Ravi both decide
    that they want this then fine - one of them can be the person to swap it over.

commit 23b06d380c4e2b3e0bfb0348ff03c5685e59754b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 22:37:33 2013 -0800

    Conforming to E261 (at least two spaces before inline comment)

commit 82a2994d21120b7505c7ebc9f78fbf6774843f35
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 22:15:58 2013 -0800

    Conforming to W391 (blank line at end of file)
    
    I've been including an empty line at the end of files since it makes it a bit
    easier for me to work with via vim. I don't care overly much though so
    appeasing pip8.

commit ff92770fba42f0daa40204795b9568fb6389d7cc
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 21:57:07 2013 -0800

    Skipping E127 (continuation line over-indented for visual indent)
    
    Bug with pep8 where detection only works for this if we have four space indents
    (which we aren't swapping to at present).

commit 1af5594eda89aa0008f20d3c63f6447b0a4e44bf
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 21:50:32 2013 -0800

    Conforming to E241 (multiple spaces after ':')

commit 9e45887a491b6b1938303bc3077d00673fca119a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 20:40:51 2013 -0800

    Conforming to E128 (continuation line under-indented for visual indent)
    
    Meh.

commit b8b32e0aa751d8f2c2fbe43502b1f400871f0919
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 20:29:16 2013 -0800

    Conforming to E226 (missing optional whitespace around operator)
    
    Another nice one.

commit b2a087ad2c84eaaa89d147fa92a5ef3c5f2f1b9a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 20:26:17 2013 -0800

    Conforming to E231 (missing whitespace after ':')
    
    Another nice catch.

commit f3fada29235b9d504c4daa34f451d31249058fa9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 20:23:09 2013 -0800

    Conforming to E201 (whitespace after '{')
    
    Agreed on this one.

commit 6a32f1a257cc1f7bf7d8499dea7d2a5df5e9d9c0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 20:21:25 2013 -0800

    Conforming to E203 (whitespace before ':')
    
    Meh. I'm having trouble thinking of this as either any better or worse.

commit 708bcad9746e4b757fd14555ed8bf13e50990475
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 20:18:55 2013 -0800

    Conforming to E124 (closing bracket does not match visual indentation)
    
    This one seems to be another false alarm, but can't expect it to always be
    right.

commit e08c0f6604a80509c001899e39553b17a21ebc0f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 20:16:12 2013 -0800

    Conforming to E211 (whitespace before '[')
    
    Another nice catch by pep8.

commit 64113e5a505d8c7b57933ab2d1e3c0a4d6feac6d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 20:14:40 2013 -0800

    Conforming to E222 (multiple spaces after operator)
    
    Extra space around an assignment. Oops.

commit 28a39f1127c7b455076e86364eb4880ae19e9c18
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 20:12:50 2013 -0800

    Conforming to E225 (missing whitespace around operator)
    
    Ok, PEP8 is trying to make me like it again. These I definitely agree with.
    Shame on me for letting 'em slip in.

commit f735fbc11973fd03435df0bc306d800433bbbeae
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 20:06:55 2013 -0800

    Conforming to E221 (multiple spaces before operator)
    
    I'd forgotten that PEP8 forbade aligned assignments. This is another spot that
    I think PEP8 is in the wrong in terms of readability but I don't care overly
    much.

commit 23b6ffd899370f3076a845c7de4bc92220a49eaa
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 20:01:28 2013 -0800

    Conforming to E126 (continuation line over-indented for hanging indent)
    
    Ok, I haven't a clue what the hell this one was about. In one spot pep8
    complained about the hanging indent. However, it looks right to me. For some
    odd reason it wants the hanging indent to match the
    'stem.descriptor.networkstatus' import above - maybe this is a bug with pep8.
    
    Oh well. Changing our code to shut it up.

commit 366f77882261031f86ac3d9a6217ae0379dc48cd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 19:52:58 2013 -0800

    Conforming to E262 (inline comment should have a space after the '#')
    
    Another nice catch.

commit 6bc6c231e16122e66205e8e04b849f316fa6ada7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 19:51:13 2013 -0800

    Conforming to E271 (multiple spaces after keyword)
    
    Caught an extra space in the ordereddict we brought in. Nice catch.

commit b25d5332a8b10c70211c83df9f48ea7b97710fde
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 19:47:55 2013 -0800

    Conforming to E502 (backslash to split lines)
    
    Ok, this one I think is stupid. "Thou shalt not have long lines" says PEP8,
    "... and thou shalt not spit up thy long lines."
    
    In two of the instances I agree, it wasn't buying us anything. But in the third
    it was actually making things more readable.
    
    Oh well, I'm not gonna put up a fuss over just one line of code.

commit 026dbce5bf191c38a48605f041b10f10629d22b2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 19:44:13 2013 -0800

    Conforming to E303 (too many blank lines)
    
    Had an extra newline at one point. I'm starting to like this style checker
    thing...

commit cca1b3c9854fc04ad3c4ff13a22d1ff9c632b5fe
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 19:40:43 2013 -0800

    Conforming to E711 (None checks by identity)
    
    E711 is actually something that we *were* trying to comply with (as are several
    other PEP8 warnings we're getting, actually). Fixing the one occurance.

commit 4c7af528d2ab80737ff6ec09d88d194ba12cce2f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 19:38:42 2013 -0800

    Still run tests if style checking is the only thing
    
    Revising the check where we abort if there's no tests to run.

commit 3ef0964914f670c05252ad139f88611b636aab37
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 19:34:58 2013 -0800

    Prompting testers to install pep8 if unavailable
    
    Style checks will require pep8 so prompting the user to install it if it's
    unavailable when they run with '--style'.

commit 662a1e00746ecbfdece8136b92f9da9494007ad5
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 19:26:00 2013 -0800

    Running PEP8 checks under a '--style' arg
    
    Well... damn. I had hoped that we would be able to wrap the PEP8 style checks
    into our normal testing runs (the reason check_whitespace.py works so well
    is that it's trivial to do with every test run).
    
    However, pep8 takes on the order of ten seconds to run over our codebase. This
    is twice as long as all of our unit tests and hence too long to include in
    tests without opting in.
    
    I'm keeping my fingers crossed that they're doing something stupid with ignored
    issues so we'll get better performance when we've made it happier. We'll see.
    
    On the up side PEP8 is wonderfully configurable with respect to what it accepts
    and ignores so we can start using it right away. Also, the output is well
    formed so it was trivial to wrap it into our present style checker (which means
    consolidated and colorized output - yay!).

commit d580192a95d0b323165cf22cfb8bbaca11af9600
Merge: 04546f8 37fd201
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 15:07:44 2013 -0800

    Expansion of Controller unit tests
    
    Several nice additions from Sean...
    
    https://trac.torproject.org/7874

commit 37fd201683180377ae7dfe0e908c199433166b12
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Jan 5 14:23:45 2013 -0700

    Add unit tests for Controller.get_network_status()
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 043fe6aede76ece7c1a2d44f2e738f471892f3ff
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Jan 5 14:22:52 2013 -0700

    Add unit tests for Controller.get_protocolinfo()
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit e70af604c1716d6baff3e0e908ecc94e861b2fd3
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Jan 5 14:21:36 2013 -0700

    Move test_get_socks_listeners_*() methods up in module
    
    Controller.get_socks_listeners() is next after get_version().
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 87b56850d18a39dcb020d3c3e3d45dce2a374d0b
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Jan 5 14:16:04 2013 -0700

    Add unit tests for Controller.get_version()
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit e69530c8af668f7b5f20c350c460e21c0b0cf081
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Jan 5 14:15:21 2013 -0700

    Explicitly enable Controller value caching
    
    I'll be playing with turning caching on and off later, so I'll set a
    known state.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 6005da73c3e667352b7bcf78b2f0b41da5ce70ae
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Jan 5 14:14:40 2013 -0700

    Move _parse_circ_path() test to bottom of module
    
    Moves the test method in its module to an analogous position as the
    tested function in its module.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 04546f858b1c7c7fe737acd0360d3a4c25e83d06
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Jan 4 15:41:16 2013 -0700

    Use tuple-of-types with isinstance
    
    Move to more Pythonic idiom for multiple type check.  This form of
    isinstance was added in Python 2.2, so it's safe for us to use.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 1bb38dd2460c0674d97d88504763b2c63ba6e72b
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Jan 4 15:32:34 2013 -0700

    Use isinstance for type checks
    
    Change the four uses of type() == x to the more Pythonic isinstance idiom.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit c482e68ebff140bdde174e15566b6be05f5b5dbd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 6 14:48:31 2013 -0800

    Fixing test_reattaching_listeners to work with RUN_ALL
    
    Checking the runner's torrc to determine what kind of authentication it's
    using. This way we can skip this part of the test (rather than fail) for the
    RUN_COOKIE and RUN_SCOOKIE targets.

commit 01cfab7d61cb167a786c509afb505ac9546e728b
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sun Jan 6 08:48:51 2013 -0700

    Test different authentication styles with passwords
    
    This restores testing various authentication paths and keeps password
    use, if available.  RUN_COOKIE and RUN_SCOOKIE targets still do not
    work.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 19d9a1ccde704ea05bd8e4b4231820a0b28df974
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 5 20:35:31 2013 -0800

    Fixing test_reattaching_listeners when there's authentication
    
    The test_reattaching_listeners wasn't making any effort to authenticate
    connections, causing the test to break for every integ target except RUN_NONE
    and RUN_OPEN (the later being the default which is why this went undiscovered
    so long). Oops.

commit 69a2d6d8aba1e538bdaf7f0ca8d2c66c57bb99de
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 5 19:48:07 2013 -0800

    Missing a requirement from test_get_latest_heartbeat
    
    All controller tests (including test_get_latest_heartbeat) require an open
    control socket. Missing this requirement broke the RUN_NONE target.

commit c78d9cda3ab5fc307264bce774187b6f4188f96d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 5 18:01:49 2013 -0800

    Parsing server descriptor's ntor-onion-key line
    
    Parsing for the newly introduced 'ntor-onion-key' line. I'm not entirely clear
    of the format that our users will want so checking with Nick and just parsing
    the raw string.

commit ec306804fc325b8670758df36ec76a1e10fe3604
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 5 16:14:12 2013 -0800

    Renaming MicrodescriptorExitPolicy to MicroExitPolicy
    
    Suggestion by Sean on "Stem code review 2013-01-03". The new name both fits
    better (since the rule class is called MicroExitPolicyRule) and a friendlier
    length to type.

commit 9cd9b4004995015c0a0a04a32bb9c68a1addf728
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 5 16:04:50 2013 -0800

    Replacing 'assert' keyword in extend_circuit()
    
    Issue spotted by Sean in "Stem code review 2013-01-03". The check itself is
    fine but I'm trying to avoid using the 'assert' keyword.

commit 33cd87351b3bfa945a5f741078d6795c18dd3972
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 22 03:27:01 2012 -0700

    Corrects pydocs for mocking.get_object
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 9dbb7fc438192af25c93f08d43686563df18e8ac
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sun Dec 30 15:08:51 2012 -0700

    Fix variable name is close_circuit() test
    
    This fixes a problem I created in commit 9119f9f0d0091a03.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 80f03bdd427813eedf88d847e9229415ca42541f
Merge: ec1078c 9808fbc
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 4 19:12:19 2013 -0800

    Controller get_streams() method and close_stream() test
    
    Merging patches by Sean for...
    
    * adding a get_streams() method to the controller
      https://trac.torproject.org/7859
    
    * integ test for the close_stream() method he added earlier
      https://trac.torproject.org/7687

commit 9808fbc0abf13ea0a6e5cef3249c286e36592494
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 4 09:59:44 2013 -0800

    Minor get_streams() testing revisions
    
    Minor tweaks for the get_streams() tests, for instance the 'stream' argument
    was unused (likely used to assert that the stream lists were equal, which
    didn't work due to mock objects leading it to be replaced when a len() check).

commit 7b2f6be091f4bb340277a49565bf37ec3a97e3c2
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 4 09:34:39 2013 -0800

    Adding a default arg to get_streams()
    
    Adding a default argument to the new get_streams() method (see ffdd61e).

commit 40a8d3da7505c3b62c8d1a2c353b17d3635bcab8
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 4 09:31:28 2013 -0800

    Raising get_streams to the top of the category
    
    This might be yet another "Damian's conventions are weird" thing that I'll come
    to regret, but I placed the getters above the modification funtions for
    circuits, so doing the same for streams.

commit 4f8efd38bbaccb573d19f8ce98f1d590bd8918f5
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Jan 4 07:37:33 2013 -0700

    Add integration tests for Controller.close_stream()
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 7e5377906b6b0e0bcc20fb442fb11554df929868
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Wed Jan 2 05:52:29 2013 -0700

    Add a Controller.get_streams method and its tests
    
    Re-use Ravi's idea for parsing get_info("circuit-status") responses with
    get_info("stream-status").  The stream status GETINFO reply contains just
    the mandatory parts of a STREAM event, so the event optional keyword
    arguments are not (yet) used.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit ec1078ccc39af8d154ef2f70781f1a3787be6789
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 3 09:42:39 2013 -0800

    Whitelist support for shell commands
    
    Shell commands have at times been the bane of my existance. They live outside
    the PATH (causing our is_available() to fail), and now it turns out that
    subprocess.Popen() fails to find them too. In retrospect both make sense but at
    first both are worthy of some head scratching.
    
    Adding a whitelist we can add shell commands to so they'll behave as we expect.

commit cde6de6bb3c132ba657ed2135ece83b9d3b60190
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 3 08:46:05 2013 -0800

    Config changes broke integ tests
    
    Shame on me for not running the integ tests (they're sluggish enough that I run
    the unit tests a lot more). I both missed a get_str_csv() usage (grepped the
    wrong thing) and an integ test with a synchronize() call.

commit 76205611acb95231226bf42c4c6b20388ef3cfe7
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 3 08:35:09 2013 -0800

    system.call() support for a default
    
    The 'suppress_exc' argument is a relic from long ago when I frequently made
    suppression the default in arm. Stem does things better, following "Errors
    should never pass silently. Unless explicitly silenced."
    
    A default argument lets us do the same thing as a 'suppress_exc' boolean, but
    with more control. It's also more uniform with our controller.

commit 5ee4b1b67876dc1ddba9aa21dbec6ef8871a8ba6
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 3 08:24:26 2013 -0800

    expand_path() pydoc inaccurate
    
    The pydocs for system.expand_path() said that it was unix specific. However,
    Beck expanded this to support Windows in 62e51e9 and others.

commit 0439570531bb909ba85b9b4ede4070e96994083d
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 2 20:21:33 2013 -0800

    Minor improvement to to_camel_case() function
    
    On reflection arm has a better function signature than us for to_camel_case().
    Adopting it.

commit 8a746ba1023a521cdb09a068ab68102a0ca75498
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 2 18:48:38 2013 -0800

    Error message printed wrong arg
    
    Drats. Just got an unexpected exception for this that I can't repro, and the
    stacktrace had the wrong attribute... >:(

commit 45f82fab334265d8ca6162192080a3e03853f323
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 1 15:15:31 2013 -0800

    Dropping the Config's synchronize() method
    
    Choice is highly overrated. Actually, in the world of libraries providing too
    many options of doing a single can make things damn confusing.
    
    The Config's synchronize() method was a great improvement in its day, but now
    that we have config_dict() it's redundant. Hopefully our module's docs are now
    more friendly for newcomers.

commit 93e1e8a1c4c5c52f09a9b974f868a223a1cda1ca
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 1 14:03:53 2013 -0800

    Dropping Config's get_int_csv() and get_str_csv()
    
    Many, many moons ago when arm first began it used integer enumerations (as is
    often the case for python projects). I added get_int_csv() to allow users to
    configure those enums, and later added get_str_csv() when we moved to our Enum
    class.
    
    Long story short, both are legacy methods for a bygone age before we had our
    new config_dict() pattern. We don't use it, and now neither does arm. Time to
    drop the cruft.

commit 2b43313d72c4a73ff5055e2bf5c5b162d8f1f4b6
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 1 13:04:56 2013 -0800

    Configuration parse_enum() function
    
    Convenience function to do the same thing as parse_enum_csv() but for a single
    enum value.

commit 3521953c1be9110f3c22cc20bf697990a5287e3a
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 31 23:26:31 2012 -0800

    Configuration parse_enum_csv() function
    
    While swapping over arm's config usage I realized that we likely don't need the
    int or str csv functions. In practice the csvs have been for enumerations.
    
    Adding a helper function that better handles configurations for enum values.

commit 661eee8dbe2f02cd714a363a48ed3ffd050a013c
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 31 16:53:15 2012 -0800

    Wrong class in docs for CircuitEvent
    
    We were missing the 'response' part of the path, breaking interlinking.

commit 7cbb066c214c227fb65dece7b282aa314a731d41
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 31 15:16:58 2012 -0800

    Updating TorCtl url
    
    TorCtl is no longer on the volunteer page so pointing to its git repo instead.

commit 683b1ba479f2aff3c277e0ff53bdc8b1f81664af
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 31 01:43:00 2012 -0800

    Support for consensus 'a' lines
    
    Adding support for the 'a' lines in version 3 router status entries. These
    contain IPv6 addresses and port lists. The port lists will make this unweildy
    for users but that's out of our hands.

commit 7df236509dca6d32d5cebb8aa0b634d3b3d962b1
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 31 00:47:58 2012 -0800

    Duplicate network status 'm' lines were ignored
    
    The 'm' lines in version 3 router status entries can appear multiple times, but
    we were only parsing the first instance.

commit cdf2c13bbe5db5a85670760572f425dffb6ca897
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 31 00:33:30 2012 -0800

    Server descriptor parsing for 'ipv6-policy' lines
    
    Adding support for IPv6 exit policies in server descriptors. Descriptors with
    these are already in the wild.
    
    https://trac.torproject.org/7826

commit 87b05342286546f603ea4bca2aee68e2b95d875e
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 31 00:12:40 2012 -0800

    Network status integ tests should check for unrecognized lines
    
    Probably the most useful integration test that we have is the check for
    unrecognized lines in the server descriptors. That one test has found dozens of
    upstream tor bugs (mostly new additions without spec updates).
    
    It turns out that we were missing this check for network statuses, and it's
    already tripping on an issue. Oops.

commit 974b71c65a3e67da25254418856c4a65c5b0993a
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 31 00:07:34 2012 -0800

    Descriptor reader callbacks for when reading a file
    
    While writing a script for 'https://trac.torproject.org/7828' I realized that
    it would be nice to be able to add hooks for when reading a file.

commit d35b9c737e320500fa6cfdfa874534cfbe6f1b3a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 21:26:34 2012 -0800

    Using a lightweight rule counterpart for microdescriptor policies
    
    The exit policies used in microdescriptors are wonderfully simple. There's no
    point in storing full ExitPolicyRule instances for them, so making an alternate
    lightweight subclass.
    
    Advantages are faster for parsing and less memory usage. Disadvantage is that
    this doesn't use the ExitPolicyRule constructor, so we'll need to take care
    that ExitPolicyRule changes are reflected in this subclass.
    
    This dropped the runtime of the network status integ tests on my system from
    9.4 to 8.3 seconds.

commit da65f282699d3e8e763e1f3ba3c7612cc970178a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 21:00:50 2012 -0800

    Lazily parsing ExitPolicy
    
    Why parse exit policies that we never use?
    
    ExitPolicy instances exist numerous places, note notably in the descriptors.
    Users may never use or care about the exit policy, so there's no point in
    putting effort into parsing them until they're needed.
    
    The advantages is that this means faster untime and less memory usage if they
    go unused. Down side is that we won't pick up on malformed content until
    they're used (handy for tests, but not somethign that users care about).
    
    This dropped the runtime for the server descriptor integ tests on my system
    from 10.1 to 7.7 seconds.

commit c34e90b55d744b0e19922d61b638c8987b089233
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 19:39:54 2012 -0800

    Tor 0.2.2.9 required for tests with new/extend_circuit()
    
    Our tests utilizing new_circuit() and extend_circuit() do not provide an
    explicit path. This breaks the tests if tor isn't version 0.2.2.9 or later.
    
    https://trac.torproject.org/7833

commit af6fd83f74f6f39981cb32e1f871421755edfdaa
Merge: dc003e8 f716fed
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 19:05:14 2012 -0800

    Circuit attachment and getter defaults
    
    Originally I set out to merge in Ravi's branch for attach_stream
    (https://trac.torproject.org/7152) but this turned into an odyssey of general
    Controller improvements. The most notable bits (besides attach_stream) are...
    
    * All of the Controller's getter methods now support a 'default' argument.
      https://trac.torproject.org/7832
    
    * Added methods to get circuits. This was part of Ravi's branch though I
      expanded it a bit.
    
    * The new_circuit() and extend_circuit() now have an 'await_build' option to
      block until the circuit is ready to be used.

commit f716fedc97ea3cc522571ad8bd505e847379c0d7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 18:59:11 2012 -0800

    Allow for blocking circuit build and extension
    
    Callers of new_circuit() and extend_circuit() will often (maybe usually?) want
    their call to block until the circuit is actually ready to be used.
    
    Unfortunately this is a little tricky for them to hack together on their own
    since it requires a listener (polling doesn't work reliably - after a couple of
    hours of hair pulling I now know that all too well). Hence doing this work for
    them.
    
    This includes a CircuitExtensionFailed exception so we can communicate the
    circuit event that reported the failure.

commit f87d9d75ef76b8755e5a3f78ae85002050ef8db0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 16:40:32 2012 -0800

    Using get_socks_listeners() for attachstream test
    
    Using our shiny new get_socks_listeners() method in test_attachstream. Besides
    being nicer, this will work with older versions of tor.

commit 11ef43c64479af8a9fe50b01e980d02234671977
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 16:32:16 2012 -0800

    Reverting __LeaveStreamsUnattached when test fails
    
    Putting the cleanup activity for test_attachstream() into a finally block so we
    still do it when the test fails. We also forgot to remove the handle_circ
    listener.

commit 9401596d2d07a2baadab5c4d144566f99d16df7d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 16:28:17 2012 -0800

    Renaming attach_stream()'s hop argument
    
    Renaming the method's 'hop' argument to 'exiting_hop' to be more descriptive.
    Also tweaking how we make the query a bit.

commit 86de12ecab643b809d2e75265512e0dc4267dd12
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 16:23:42 2012 -0800

    Reordering Controller methods
    
    Rearranging the Controller's methods to match the header. This'll both make it
    easier to find things and rearrange our documentation.

commit ffdd61ea41ca844047ac57725639a7cf0e22d41d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 16:03:30 2012 -0800

    Uniformly providing a default argument for getters
    
    As disussed on 'https://trac.torproject.org/7832' it's important from a
    usability standpoint for all of our getter methods to support a 'default'
    argument. It's also nice to have class-wide uniformity on this since it makes
    the Controller class more intuitive to use.
    
    This also reorganizes the method listing in the header to group related
    methods, and has a couple small bits of backward incompatability (trying to get
    these breakages in early before making our initial release announcement)...
    
    * renamed protocolinfo() to get_protocolinfo() to match the other getters
    * renamed one of the 'circuit' arguments to 'circuit_id' (we did that for all
      of the methods but one)

commit 22f167fa48bf51d0f66736624293050bb022a936
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 15:03:49 2012 -0800

    Adding a default arg to get_circuits()
    
    Getter methods should have a default argument (though admittedly I've been
    doing a bad job of requiring this, #7832).

commit cb9b4b605e8bdb639b22b2e58e5d2e80da9dcb56
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 14:52:41 2012 -0800

    Adding a get_circuit() Controller method
    
    While reading through the tests I realized that we frequently wanted a method
    for 'gimme the circuit with id X', so adding a method for that.

commit b01a2e9b1474f6ff858ff85f40bcdcb68b8c66ac
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sun Dec 30 23:17:13 2012 +0530

    use SocksListenAddress instead of SocksPort

commit 4d9a71295f5f4afcfb41f3f0f75ef4ecc45b2124
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sun Dec 30 23:03:48 2012 +0530

    Add missing import

commit d70e98ab0c58aa5a209b24394852ecb7ae39e563
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Dec 26 14:44:38 2012 +0530

    Make test_attach_stream more robust

commit 66af2a98e85db0442687b5a684013a41bdf50d9c
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sun Dec 23 22:55:32 2012 +0530

    Make test_extend_circuit and test_close_circuit use get_circuits

commit bcd03831cace7352a7cbdf4e9f4dd651d86a48af
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sat Dec 22 17:25:19 2012 +0530

    Make test_close_circuit use Controller.get_circuits()

commit 299a872c0b7ba26babec3febf8c0a8adad09f4c7
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sat Dec 22 17:17:47 2012 +0530

    Fix test_repurpose_circuits to work with string ids

commit 631b210c59fcbeadd3e75a1b1ac9e9f9a635d142
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sat Dec 22 14:33:26 2012 +0530

    Fix test_get_circuits to work with string ids

commit e0929d2cde0b57232dea966192dd4fe954667d93
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Dec 17 08:09:18 2012 +0530

    Variable naming consistency

commit 32a54a71f8391b5300e6cae5a3f558f190160995
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Jul 3 10:35:52 2012 +0530

    Implement Controller.attach_stream

commit 1b9ffa597408e9b7f0f8863e8c7a2f66d9922e1c
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Dec 5 08:08:18 2012 +0530

    Add a Controller.get_circuits method

commit dc003e8bb26bf9261718ee2ac6ba5a3109972c44
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 02:23:12 2012 -0800

    Moving the ExitPolicyRule's masked_bits behind a method
    
    In looking at TorCtl's ExitPolicyLine class I'm realizing that we may want to
    use our mask attribute as our main subnet representation rather than the
    masked_bits.
    
    Changing the class so the masked_bits are accessed via a method so we'll have
    the flexability to change this in the future.

commit d1d00b635c53c885ff930fe28771cef51f1edebb
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 30 01:18:59 2012 -0800

    Flagging the ExitPolicy's set_default_allowed() as private
    
    The ExitPolicy's set_default_allowed() method is there to support
    microdescriptor policies, not for our users. Acually, exit policies were
    entirely immutable with the exception of this attribute. Changing the method to
    be private.

commit bbb2d5a44a898e1a7202d876fe27cc87065ffd3f
Merge: bf28436 79fc8cd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 29 23:34:13 2012 -0800

    Adding get_socks_listeners() to Controller
    
    Adding a get_socks_listeners() to the controller to provide the (address, port)
    tuples where tor is listening for SOCKS connections. This also includes some
    mocking changes to better support the tests.
    
    Changes are from Sean...
    
    https://trac.torproject.org/7783
    https://trac.torproject.org/7784

commit 79fc8cd7514cda072abf855b6a1908f0dd10bb17
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 29 23:27:06 2012 -0800

    Revisions for prior mocking and controller changes
    
    Changes include...
    
    * Rewrite of get_socks_ports() with the following changes...
    
      * Renaming to get_socks_listeners() since it provides (address, port) tuples,
        not just ports.
    
      * Better error handling. Previously malformed content such as invalid ports
        would cause us to raise ValueErrors.
    
      * The method wasn't documented as raising any exceptions, but if the
        get_conf() calls fail then a ControllerError will be raised.
    
    * Minor corrections to the return_for_args() examples.
    
    * Renamed return_for_args()'s "method" arg to "is_method" (it's nice when the
      an arg name can imply the type).
    
    * I'm not sure what was meant by "Also, remove the the space in the join so
      that there is only one parameter passed to the string substitution.".
      Personally I'd find a message like...
    
      "Unrecognized argument sent for return_for_args(). Got 'foo, bar' but we only
      recognize 'foo, blarg'."
    
      ... to be a bit nicer than...
    
      "Unrecognized argument sent for return_for_args(). Got 'foo, bar' but we only
      recognize 'foo;blarg'."

commit 3a2875f22d1bf5dba6171e98807ef22f2a02c729
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 22 06:16:56 2012 -0700

    Add get_socks_ports method to Controller
    
    Adds a method to find the SOCKS listeners on a controlled tor.  This
    method is tested on tor versions 0.2.1.32 - 0.2.4.6-alpha and provides a
    similar response regardless of the underlying tor process.  Also includes
    unit and integration tests for the new method.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 667cafa0c5fbdcc74dfb781ef8f55ac84d727112
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 22 06:14:37 2012 -0700

    Replace SocksPort with SocksListenAddress in base torrc
    
    This is the most widely accepted format for specifying a SOCKS port,
    it works with tor 0.2.1.x - 0.2.4.x.  SocksPort is preferred for tor
    0.2.4.x and later, but this also acceptable.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit d3cd0841393b20b4b333be0b39fb4890609426cf
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 22 04:34:36 2012 -0700

    Add explicit declaration of method use for mocking.return_for_args
    
    The previous way of removing 'self' for MockClass objects only worked
    for fully mock objects.  By using an explicit declaration, mocked
    methods for real objects can be used.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 1dbb7befdbd10169a35e27977a0736a9dc9465d0
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 22 04:29:46 2012 -0700

    Handle keyword arguments with mocking.return_for_args()
    
    Allow a structured way to match keyword arguments to be included in the
    mapping keys for return_for_args().  This changes the API for
    return_for_args to require tuples as the mapping keys.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit ef5c552e59017936699048d680f5f377f1b60d66
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 22 04:17:38 2012 -0700

    Convert keys before reporting error
    
    Non-string argument keys caused the ", ".join() to fail with a TypeError.
    Also, remove the the space in the join so that there is only one parameter
    passed to the string substitution.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit bf28436368aea5a0cad3545de9406934434a9f1a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 23 11:58:38 2012 -0800

    Fixing eronious 'missing with' warning
    
    Running our tests gave a working that network.py was missing an import of the
    'with' statement. This was wrong - the with statement was in its pydocs. The
    checker tries to be smart about what is and isn't documentation by looking for
    block quotes but an unusual string confused it...
    
      """ Base error for proxy issues. """
    
    I avoid using block quotes unless things span multiple lines which is why I
    didn't account for this. We could fix the checker (... which isn't a bad idea)
    but for the moment I'm just dropping the odd block quotes since it'll probably
    introduce extra whitespace before and after the message in the sphinx docs.

commit 0c4cb6d69e1b47faad757402fb666d6310252c09
Merge: cfc745e 8aba54d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 22 16:29:51 2012 -0800

    Utility to test client usage in integ tests
    
    Utility class to make it easy to use the tor instance that we have for testing
    for client traffic...
    
    https://trac.torproject.org/7693

commit 8aba54d53bc2738ecea503ec09e71a32edb5cc0a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 22 16:27:32 2012 -0800

    Dropping test.network.Socks interlink
    
    The ":class: `~test.network.Socks`" had a couple issues...
    
    * there's a space between the :class: and back-tic (not sure if that's a
      problem for sphinx, but definitely different)
    
    * we can't use things like :class: within the preformatted text block

commit 31cbbbad01230b26722c7524617518f2228e55f1
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Tue Dec 18 06:21:13 2012 -0700

    Advertise network module inside test
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 270f173deddd2e3dded794d23306f1f32b78d400
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Mon Dec 17 20:03:29 2012 -0700

    Add SocksPatch to socksify network-using Python code
    
    A context manager which helps re-direct network socket traffic, from
    existing code, through the tor process.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 4485657e3840f0ea9d9fc96623dd7f50a476a79d
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Mon Dec 17 20:02:22 2012 -0700

    Add Socks class to handle a socket connection through a SOCKS5 proxy
    
    I attempted to keep this generic and open enough that IPV6 support can
    be easily added at a later date.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit bf6651aa0121e51e938e87608a4df69c0e80c917
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Mon Dec 17 19:20:23 2012 -0700

    Add exception classes to test.network module
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit ef0cee8123578ed8ace0e3304735eb683dc6ca5e
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Mon Dec 17 19:12:37 2012 -0700

    Add a new module to test to support network connections through tor
    
    This module aims to provide basic connectivity options on which tests
    may be built.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit cfc745e95010b4c86a8d794c118bfafafee4d59d
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 21 09:17:23 2012 -0700

    Documentation and output string fixes
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit a5b05edefbbcbc7261e40a5842a1ef901a60a25c
Merge: f60e327 5687229
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 22 14:30:29 2012 -0800

    Integ test fixes concerning string circ/stream ids
    
    Fixes from Sean to fix integration tests broken by commit 6711458...
    
    https://trac.torproject.org/7778

commit 568722919c499c1e409680fe38fb3471a97b425c
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sun Dec 16 20:02:49 2012 -0700

    Fix test_extendcircuit by removing int() conversion
    
    Traceback:
      File "./test/integ/control/controller.py", line 464, in test_extendcircuit
        self.assertTrue(filter(lambda x: int(x.split()[0]) == circ_id, controller.get_info('circuit-status').splitlines()))
    AssertionError: [] is not true
    
    Also, change to standard variable name "circuit_id".

commit 848d4c95b5c6f908755ac7db4019a3ac2bf2e9eb
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sun Dec 16 19:46:41 2012 -0700

    Fix test_repurpose_circuit crash by changing directive in format string
    
    Traceback:
      File "./test/integ/control/controller.py", line 486, in test_repurpose_circuit
        circ = filter(re.compile("^%i " % circ_id).match, circuit_output.splitlines())[0]
    TypeError: %d format: a number is required, not str
    
    Also, change to standard variable name "circuit_id.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 9119f9f0d0091a030654002a2e7ae6ccf91a789f
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sun Dec 16 19:32:22 2012 -0700

    Fix test_close_circuit crash by concatenating strings
    
    Traceback:
      File "./test/integ/control/controller.py", line 521, in test_close_circuit
        self.assertRaises(stem.InvalidArguments, controller.close_circuit, circ_id + 1024)
    TypeError: cannot concatenate 'str' and 'int' objects
    
    Also, change to standard variable name "circuit_id".
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit f60e3272f78acdc3c66c07f897688d1e3a381ca3
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 21 08:19:10 2012 -0800

    Dropping duplicate stop() call
    
    The test runner is ran within a finally block that calls() stop. Reguardless of
    if tor fails to start or not stop() will be called, so this call was a
    duplicate.
    
    Calling stop() multiple times is fine (the method is idempotent), but it logs
    each time so we were saying "Shutting down tor... done" twice. It's still
    possible for us to give double messages in some other use cases so we should
    address this more generally later, though I'm not sure offhand what that fix
    should be.

commit bf2f7f2a461b535ddd21728e29dc2091ef35a74d
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 21 08:06:15 2012 -0800

    Reporting testing as having failed if tor can't start
    
    As pointed out by Simon on...
    https://lists.torproject.org/pipermail/tor-dev/2012-December/004296.html
    
    ... we presently say that testing has passed if tor fails to start. This is
    wrong.

commit e22a753058a2b9c61301c62e2d2123ccea046f29
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 21 08:02:10 2012 -0800

    Sanity testing the that the tor command points to tor
    
    Adding some basic sanity tests to launch_tor() so that we provide a better
    error message if our tor command is wrong. This is to address the following,
    which was admittedly really confusing error output...
    
    https://lists.torproject.org/pipermail/tor-dev/2012-December/004296.html
    
    Caught thanks to Simon.

commit b20d7acc9cd6fad1fd17f8dddccf8473d0034a09
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 15 15:32:47 2012 -0700

    Add more test cases for re-attaching event listeners
    
    This covers the three anticipated cases when re-attaching event listeners
    is required: Controller.authenticate(), stem.connection.authenticate(),
    and BaseController.msg("AUTHENTICATE").
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit f5620f2dbd4bbd5c3bdcb2c19abad765f4e324c3
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 15 15:28:11 2012 -0700

    Use threading.Event in event listener tests
    
    wait() up to two seconds, for a threading.Event to be set() in the
    test event listener before continuing assertions.
    
    This should speed up success or failure in these tests, even if not by
    much.  test_event_handling() goes from a constant time of 5 seconds, to
    a variable 4-8 seconds, with 4 seconds most common in my testing.
    test_reattaching_listeners() goes from a constant 6 seconds, to a variable
    4-6 seconds.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 6bc92816dc4356a204fdde0f160ee344875143a3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 16 18:50:46 2012 -0800

    Adding get_latest_heartbeat()
    
    Adding a method to the BaseController to get the time that we last heard from
    tor. This is useful to figure out if our controller connection has stalled, but
    only works if we're continually using our controller (such as by listening for
    BW events).

commit 229a6319109372dc529106516a9fcb65f5c2eb73
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 16 18:35:41 2012 -0800

    Making the Controller's enabled_features private
    
    The Controller's 'enabled_features' attribute wasn't documented so it was
    probably meant to be private. Changing its name to reflect that.

commit 844073f3813d496fe4d5037283413d837d1ae325
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 16 13:42:10 2012 -0800

    Adding the requirement for 'PortForwarding'
    
    Adding in the version requirement for the 'PortForwarding' torrc option. This
    was hardcoded in arm.

commit 5b45d3125c811067d7c3ba0a11e324423c88d925
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 16 11:03:07 2012 -0800

    Revising error handling for get_conf()
    
    The Controller's get_conf() function was buggy if the configuration value was
    unset. It's documented as returning the default, but instead it returned None
    or [None] depending on if 'multiple' was set.
    
    On reflection the pydocs for these functions were pretty unfriendly, so giving
    them an overhaul. Functionally this changes two things...
    
    1. When you query for an unset value it will...
    
       * provide the default value if you gave it one
       * provide None if multiple was False, and [] if it was True
    
    2. The get_conf_map()'s 'default' argument is now the defaultly *mapped value*
       rather than the default *response*. For instance...
    
       # old behavior
       get_conf_map('NoOption', 'Blarg') => 'Blarg'
    
       # new behavior
       get_conf_map('NoOption', 'Blarg') => {'NoOption': 'Blarg'}
    
       This means that the get_conf_map() pydocs can now assure the caller that
       they'll always get a dict, and it also better matches the get_conf()
       handling for default.

commit fc4bdf7ab686b4bbd12e3aefc83d78d0a171df60
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 15 16:22:30 2012 -0800

    The str_tools functions didn't handle negative values
    
    Functions like str_tools.get_size_label() provided labels using the smallest
    unit type if the count was negative. For instance, '-2000000' would be
    translated into '-2000000 B' rather than '-1 MB'.
    
    An inline comment seems to indicate that this was on purpose, but I haven't a
    clue why. Fixing it.

commit a201c0c5f4b92b119f89fa39bd43680d67a64817
Merge: e30f2df 256676d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 15 13:46:08 2012 -0800

    Unit tests for event error conditions
    
    Testing expansion from Sean for malformed event content, plus a couple fixes.
    
    https://trac.torproject.org/7736

commit 256676d5e686fb544927a715335c567d693254cc
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 15 13:06:41 2012 -0800

    Event testing revisions
    
    Minor revisions for the prior event testing additions...
    
    * Adding comments saying how the 'BAD' events are malformed. Some of them took
      me a few seconds to figure out. :)
    
    * Renaming the 'ORCONN_CONNECTED_BAD_*' events to 'ORCONN_BAD_*' since it's not
      especially important that these are connection events (and in one case it
      isn't).
    
    * Putting ORCONN_BAD_* events on a single line to match the non-bad
      counterparts. The rough convention that I've been using is to add a newline
      if there's either a key/value argument or a really long positional arg.
    
    * Renaming ADDRMAP_BAD_3 to ADDRMAP_ERROR_EVENT. I like the ADDRMAP_BAD_1 and
      ADDRMAP_BAD_2 names since they make it clear that they're malformed input.
      However, ADDRMAP_BAD_3 is a valid response where tor is conveying that an
      error occurred.
    
    * Actually, I think that your STREAM_SENTCONNECT_BAD_1 event reveals a bug. The
      spec doesn't seem to indicate that the target is optional, so making it
      mandatory.

commit b3aca85301439fd2441d098775b9c56d7365c506
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 16:28:25 2012 -0700

    Add test cases for STREAM events
    
    These test events exercise all the exceptions in StreamEvent._parse()
    and the case of an empty target address:port.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 38190ab9cbedf9844022fe02f4385f1b129c5f52
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 16:15:53 2012 -0700

    Move NEWDESC sample events to alphabetic position
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 58c48067ba1b901bdb1d4151181949591c97b60a
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 16:12:28 2012 -0700

    Add test case for invalid STATUS_* event type
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 146f32473a3c0f9f91c0e4a721e92906d293a63e
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 16:11:37 2012 -0700

    Change status event type for CONSENSUS_ARRIVED
    
    According to control-spec.txt, CONSENSUS_ARRIVED is in the STATUS_GENERAL
    family.  While it is not vital for the faked test events, let's make
    it accurate.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 04633d4243f88f232a39daf5cbffbb228ee8f6bf
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 15:44:35 2012 -0700

    Add test cases for ORCONN events
    
    Rename ORCONN_CONNECTED to ORCONN_LAUNCHED and use ORCONN_CONNECTED
    for more complex _parse() method test.  Adds three bad events to trigger
    exceptions in _parse().  Move sample lines to alphabetic position among
    other sample cases.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit df01f3737143545493652437680db52ea8e7f337
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 15:20:13 2012 -0700

    Add test cases for CLIENTS_SEEN event
    
    These sample events hit every exception possible in the ClientsSeenEvent
    _parse() method.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 6cf5422575dd1e3e6fe2237d6bbc2f45e2096e1a
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 15:09:00 2012 -0700

    Add test cases for CIRC_MINOR events
    
    Checks the same error conditions added to CIRC event tests in previous
    commit.  Also, swap test_circ_minor_event() and test_clients_seen_event()
    to alphabetic order of appearance.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit c5db5cb1be07da0da92b30559557e72285d3fc37
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 15:00:49 2012 -0700

    Add test cases for CIRC events
    
    Check ProtocolError is raised when the TIME_CREATED time cannot be
    parsed or the CiruitID is not valid.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 48daff039d95d8223ecd83cbd3af59a2ee00dc37
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 14:58:08 2012 -0700

    Add test cases for BUILDTIMEOUT_SET events
    
    Check integer and float conversion failures raise a ProtocolError.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 97dc4108404a1c0f676845538a542f98085ffb4a
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 14:34:59 2012 -0700

    Add test case for missing BW read value
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit f1b81e062c2b1ec646f99b346b98dd3d4ec38057
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 14:30:40 2012 -0700

    Test error conditions for ADDRMAP events
    
    Add tests for missing quotes before (ADDRMAP_BAD_1) and after
    (ADDRMAP_BAD_2) the expiry positional argument.  Add a test for an
    error event (ADDRMAP_BAD_3), constructed from control-spec.txt.  Also,
    move the sample event lines to their alphabetic position in the list of
    other event samples.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 26a39c9ac5bf77860f2a10dd776c95ce46ef6380
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 14:13:38 2012 -0700

    Add tests for stem.response.events.Event
    
    This tests the base class used to build named event classes (e.g. ADDRMAP,
    BW, CIRC, etc.).  Mainly this documents existing behavior so we can
    check future changes to the class do not surprise us with breakage.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 8a0c9f58046670c7a808e2e4a79b82244405e847
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 14:05:13 2012 -0700

    FIX: include positional argument name which is missing proper quoting
    
    This was discovered with a spot coverage test.  These were the only
    lines not run in the Event class while I was adding tests for the
    stem.response.events module.  I just *had* to write a test for these
    two lines and it turned out they had a problem.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit e30f2df26221004fc2c535e5ef5e0b2c8518e3b2
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 14 17:12:27 2012 -0700

    Remove extra checks for circuit id and stream id validity
    
    CircuitID and StreamID may not be None, so raise ProtocolError if either
    ID is None.  Inside is_valid_circuit_id() and is_valid_stream_id(),
    isinstance(entry, str) already returns False if None is checked.  Add None
    as invalid IDs in tests.
    
    NOTE: Do not try this with is_valid_fingerprint() and is_valid_nickname(),
    as the fingerprint or nickname may be None in certain cases.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 22286e9dc3fea710182984dd320f4e419cc5a425
Merge: 6711458 9a1601d
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 14 09:27:50 2012 -0800

    Script for republishing stem's site
    
    Now that we're moving our site from ReadTheDocs to a tpo host (staticiforme) we
    need our own method for automatically republishing our sphinx docs. To do this
    we're using a cron process that runs every five minutes which...
    
    * Pulls stem's git repo. If it has changed...
    * Run 'make html' to rebuild the docs.
    * Republish the content by calling 'static-master-update-component'.
    
    Many thanks to weasel for working with me on this, and pointing out the
    numerous mistakes I've been making.

commit 9a1601db33f58b044aa2375db8f97f364bdd849c
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 14 09:12:48 2012 -0800

    Logging site republication runtime
    
    Again, mostly a commit to exercise that site republication is now working with
    a cron setup.

commit 85241a8cdb145edce2655dda69d2c822276095a1
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 14 08:59:16 2012 -0800

    Only logging an init message when repeating
    
    I'm about to change the republication script to be a cron process rather than a
    daemon, but first I want to drop its initializtion log message when running in
    a one-off fasion. Otherwise we'll get a log entry every five minutes for it.

commit 91fc3b8a0d7b33eddfee52d4bef29604923fa19f
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 14 00:32:01 2012 -0800

    Correcting mistake in the tutorial
    
    Correction thanks to arma...
    
    23:57 < armadev> tutorial looks nice!
    23:57 < armadev> s/lets/let's/g though

commit ba930a28b5e2cd7f43c4b85cdd61366d58103892
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 13 23:40:33 2012 -0800

    Revising our site url
    
    Honestly I'm just making a trivial fix so I can see if my republisher script
    finally works or not (fingers crossed!).

commit 2fcdc4bbd6ebcb27725d45cffa3461b4d833913e
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 13 23:34:43 2012 -0800

    Propagating cwd for republish calls
    
    The script's a little nicer (and closer to working) when we pass through a cwd
    rather than trying to set it separately.

commit 0700b831ff4068bc6ea83e33e8b5c63fdeaaeb0b
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 13 23:10:01 2012 -0800

    Dropping disown from republish
    
    The disown command isn't available on the webserver, and on some
    experimentation seems to be unnecessary. The daemon process survives after
    logout.

commit ffde535388f2996a9adccdccc75a22a0176d0d35
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 13 22:56:17 2012 -0800

    Republishing --repeat should... well, repeat
    
    Forgot to make the script continually run when --repeat was provided.

commit d2b2b9757cf56be09d8841184f493ae478247c0b
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 13 22:53:52 2012 -0800

    Passing args along to republish.py

commit 9dbf340fde1a7c17b40cf3e01c0bd99684b91ce1
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 13 22:52:28 2012 -0800

    Adding execute to republish

commit 72832baa56dfbf0cff0980746db708c897017603
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 13 22:50:34 2012 -0800

    Script to republish stem's website content
    
    Script for stem's webserver (staticiforme) to provide a daemon that
    periodically publishes our site's content.

commit 6711458be83b2050b058b7528bccb4e15c2c6a25
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 13 08:39:18 2012 -0800

    Circuit and stream ids are strings
    
    Ids for streams and circuits presently look like ints, but stem should handle
    them as strings as per the control-spec...
    
    ; Unique identifiers for streams or circuits.  Currently, Tor only
    ; uses digits, but this may change
    StreamID = 1*16 IDChar
    CircuitID = 1*16 IDChar
    IDChar = ALPHA / DIGIT
    
    Issue caught by Lunar^. This should be invisible to callers (they can probably
    still provide us with ints), the only noticeable difference being that
    new_circuit() and extend_circuit() now provides a str rather than int.

commit 9b5ff19be35ebbf8c75772a63406ba23762b37de
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Dec 12 09:32:02 2012 -0800

    Lowering cache hits to the TRACE runlevel
    
    In practice logging GETINFO and GETCONF cache hits can be very noisy (within a
    minute of running it with arm there were around a hundred duplicates). Lowering
    the visability of those events.

commit a7b275a3d64301da4d23137dbadd4842cc7a041f
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Dec 12 08:13:39 2012 -0800

    log_to_stdout() only printed at TRACE level
    
    Ok, the logging module is *trying* to make me hate it. I haven't a clue what's
    wrong with the basicConfig() function, but when I try to use it the 'level'
    argument is ignored and it always spews its output at the TRACE runlevel. This
    is worthless (trace is often too noisy to be helpful) so making my own
    StdoutLogger.
    
    Tested by running this on a separate system, then copying the change over by
    hand.

commit 92a7c6f8224ce9e69caab7c52b027b77e77dc1e5
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 11 19:38:10 2012 -0800

    Default ignored when geoip was unavailable
    
    We ignored the default argument and always raised a ProtocolError if the geoip
    database was unavalable - oops.

commit 872af06dc1a5770fb57e0af72e2412aa879f7e1f
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 11 19:30:45 2012 -0800

    Logging a warning when the geoip db is unavailable
    
    Issuing a warning when we've made enough failed geoip queries to conclude that
    the database is unavailable. This is partly for parity with arm which issues a
    warning in this situation (and that seems appropriate for other callers too).

commit f4061ace5d87f2b0d4b47fb2a0aabbdcbd97b964
Merge: a8b20ad 6176abb
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 10 07:47:52 2012 -0800

    Adding a close_stream() to the Controller
    
    Feature being discussed on...
    https://trac.torproject.org/7687

commit 6176abb3f1c46192126a256aea6d6b27b4d66df9
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sun Dec 9 18:38:47 2012 -0700

    Minor revisions to the previous close_stream() revisions
    
    These tweaks change some pydoc verbiage and put NONE back in
    CircClosureReason while removing it from RelayEndReason.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 29a6d14fa054c05e4c3c6bb97b0a91fd2fe3569a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 9 14:41:53 2012 -0800

    Minor revisions to close_stream()
    
    * Dropping the NONE RelayEndReason. It feels a little hacky to include it just
      so the index_of() will align with the integer values. Instead having
      close_stream() account for the offset.
    
    * The reason parameter of close_stream() is a stem.RelayEndReason rather than
      int so fixing its pydocs.
    
    * When doing formatted strings it's not necessary to wrap values in str(). If
      it's being assigned to a '%s' then that happens as part of the insertion.

commit 33547a2b346b07f900d81be08f2dcd19451ea434
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 9 14:33:21 2012 -0800

    Deduplicating RelayEndReason vs StreamClosureReason
    
    StreamClosureReason is a superset of RelayEndReason so there's no need to list
    both. Merging the enum descriptions and having StreamClosureReason refer to
    RelayEndReason.
    
    Also added RelayEndReason to __all__ so wildcard imports will pick it up.

commit 786624951b627ae76370ab3afe38997552b80ae2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 9 14:23:42 2012 -0800

    Consolidating how StreamClosureReason is made

commit 77481da93b23a54d339133ba379c0a6f3741f799
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 8 19:46:56 2012 -0700

    Add a close_stream method to Controller
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 545f251b698281b5cd6bd3f704242c9fdafaa36b
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 8 19:24:28 2012 -0700

    Build StreamClosureReason as a superset of RelayEndReason
    
    Explicitly make StreamClosureReason rely on RelayEndReason for its
    contents.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 69a2c8193fd613ce057b6d0adb466a21437ef5ff
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 8 19:20:38 2012 -0700

    Add enumeration of RELAY_END reasons
    
    This is for the reason to be given with the CLOSESTREAM command.
    NONE is a filler to make the remaining reasons start at offset one
    and will be rejected by the server, so it is not listed in the pydoc.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit a8b20adb7044971935380b4170254c1c2d7f2051
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sun Dec 9 08:37:54 2012 -0700

    Improve tests for close_circuit exceptions
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit b3fe1659c0d2c7bf7c8ce282e0f858eb6350e80e
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 8 19:57:13 2012 -0700

    Raise different exceptions for different CLOSECIRCUIT errors
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit ec13ebb581fc7b57f055c59934f5f7f41c2111bb
Merge: 4cacd2f 3be3a8d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 9 13:54:20 2012 -0800

    Testing event version requirement checks
    
    Patches by robinson to both fix the mock_method() function and use it to test
    the tor version requirement checks for events.
    
    event version check: https://trac.torproject.org/7674
    mock_method() fix: https://trac.torproject.org/7686

commit 3be3a8d94bb666b56bdc65e732f83c88ccfe0a15
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 9 13:52:20 2012 -0800

    Minor revisions for the new add_event_listener tests
    
    Just some minor tweaks...
    
    * Replaced 'lambda x: x' with 'mocking.no_op()'.
    
    * Dropped the 'assertIsNone()' call. The purpose of this line is to make sure
      that add_event_listener() doesn't raise an exception, not to make an
      assertion on what the function returns.
    
    * Minor import tweaks to conform with 3da47d3.

commit f5e2d127581a487f9cba7edf86943c0f78445180
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 8 18:29:54 2012 -0700

    Add version requirement tests for Controller.add_event_listener
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 22389a8ffd94251e857d3dd9d0abd40a01bc574e
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 8 18:25:19 2012 -0700

    Update mocking.mock_method pydocs
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit a00e2dda071bb2d6c04b6c78ece33942ba0def53
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 8 18:08:39 2012 -0700

    Use getattr()/setattr() for mocking.mock_method
    
    mocking.revert_mocking also needs setattr because it may be trying to
    revert to the original method in an object.  This fixes the following
    error when trying to use mock_method:
    
        TypeError: 'dictproxy' object does not support item assignment
    
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 4cacd2f13d1cd003db0c15addcf265fa4262ee37
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 9 12:33:06 2012 -0800

    Dropping '--' from config options being used to launch tor
    
    The '--' prefix in '--[config option]' commands are unnecessary. Thanks to
    lunar on #7666 for pointing this out.

commit 120907822f06fd476f2c47b7135e816cd730b1c7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 8 23:41:21 2012 -0800

    Only using TAKEOWNERSHIP if we're on the local system
    
    We were using TAKEOWNERSHIP if our pid matched tor's __OwningControllerProcess.
    However, doing this doesn't make sense if we're connecting remotely. Caught by
    Robert on...
    
    https://trac.torproject.org/7666

commit aebe936d6e3f6a62d3d5f70944f06d599c207475
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 8 23:11:24 2012 -0800

    Providing '__ReloadTorrcOnSIGHUP 0' when lauching tor with a custom config
    
    Lauching tor with the '__ReloadTorrcOnSIGHUP 0' option when we provide it with
    a custom configuration. This is to prevent the process from erring out when it
    gets a sighup due to not having a torrc.
    
    Suggestion by Robert on...
    
    https://trac.torproject.org/7682

commit 6f94c909d21b710fe86030cf3c1824102fc52613
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 8 23:04:30 2012 -0800

    Dropping eronious regex from test
    
    A relay's LongName identifier can be missing the relay name. This was caught by
    Eoin and Robert on...
    
    https://trac.torproject.org/7558

commit 3da47d3b9d6d1ae5c6b2013a4247c45c13461a05
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 8 22:02:29 2012 -0800

    Reordering import statements
    
    Dropping my oddball import ordering scheme in favor of the python standard.
    Now that we're getting more contributors making the library newcomer-friendly
    defintely trumps my weird notions of aesthetically pleasant imports.
    
    Taking this opportunity to also drop use of the 'as' import keyword. I'm still
    a little unsure what the most 'proper' style for imports is, but I've defintely
    seen 'from' used a lot more than 'as'.

commit 9fddf70f2d1028dc6434ab866e010ca1fa53fb1d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 8 15:35:38 2012 -0800

    Minor correction for the get_info() pydocs
    
    The prior comment about when a stem.ProtocolError was raised was mostly right,
    though it missed an important detail: our 'detection' for the geoip db being
    unavailabe is based on repeated failures so the that exception likely *won't*
    be raised for db unavailability.

commit bd386445e7c6bf93348fde50503765fb13341971
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 8 15:31:57 2012 -0800

    FIX: ProtocolError is no longer in stem.socket
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 9fc23f64d3534a7e68bb0b66dddbd18b3e8db63c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 8 15:24:50 2012 -0800

    Unexpected TypeError when get_conf() called for non-existant value
    
    Calling the Controller's get_conf() method would result in a TypeError if
    the configuration option didn't exist. For instance...
    
    ======================================================================
    ERROR: test_reattaching_listeners
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/integ/control/controller.py", line 134, in test_reattaching_listeners
        controller.authenticate()
      File "/home/atagar/Desktop/stem/stem/control.py", line 954, in authenticate
        stem.connection.authenticate(self, *args, **kwargs)
      File "/home/atagar/Desktop/stem/stem/connection.py", line 358, in authenticate
        authenticate_none(controller, False)
      File "/home/atagar/Desktop/stem/stem/connection.py", line 435, in authenticate_none
        auth_response = _msg(controller, "AUTHENTICATE")
      File "/home/atagar/Desktop/stem/stem/connection.py", line 787, in _msg
        return controller.msg(message)
      File "/home/atagar/Desktop/stem/stem/control.py", line 308, in msg
        self._post_authentication()
      File "/home/atagar/Desktop/stem/stem/control.py", line 1511, in _post_authentication
        owning_pid = self.get_conf("__OwningControllerProcess", None)
      File "/home/atagar/Desktop/stem/stem/control.py", line 1009, in get_conf
        return _case_insensitive_lookup(entries, param, default)
      File "/home/atagar/Desktop/stem/stem/control.py", line 1644, in _case_insensitive_lookup
        for entry in entries:
    TypeError: 'NoneType' object is not iterable

commit 40827b365e3a5f3dac6d1f8e8ffddd589be6cd70
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 8 15:10:08 2012 -0800

    Checking event version prereqs before setting any
    
    Checking that our tor version meets all of the event's version prereqs before
    setting any. This has a few advantages...
    
    * It makes add_event_listener() an all-or-nothing operation (ie, failing one
      of the checks means that no listeners are attached, rather than a few)
    
    * Avoids repeated acquisitions of _event_listeners_lock (so other lock users
      can't slip in-between).
    
    * Lets me fix a bug where we were calling SETEVENTS for each event type, rather
      than a single call when we're done adding the listeners.

commit 8506a43e04df43166f009c523cb63762d2243aeb
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 8 15:04:18 2012 -0800

    Add version requirements for event types
    
    Patch from Sean to add the version requirements for numerous event types...
    
    * AUTHDIR_NEWDESCS (added in tor commit b9d37a2d58e76ee2)
    * BUILDTIMEOUT_SET (version information taken from control-spec.txt)
    * CIRC_MINOR (added in tor commit 5519e633ec4c48c5)
    * CLIENTS_SEEN (added in tor commit 5519e633ec4c48c5)
    * CONF_CHANGED (added in tor commit 5a801a8c8b71c955)
    * DESCCHANGED (version information taken from control-spec.txt)
    * GUARD (added in tor commit 31e09bb5242b24d5)
    * NS (version information taken from control-spec.txt)
    * NEWCONSENSUS (version information taken from control-spec.txt)
    * SIGNAL (version information taken from control-spec.txt)
    * STATUS_* (added in tor commit 4401c47f638a60da)
    * STREAM_BW (added in tor commit 4a74812c70b75df8)

commit 6da7588be3d0331d4e66a41901510cc680aa7cba
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 7 05:39:42 2012 -0700

    Note that some events are original to control-spec V1
    
    Do not explicitly assign a minimum version to some events.  Instead,
    use the Event class version, but add pydocs about when the event
    was added to Tor.
    
    While these events were all added in tor version 0.1.1.1-alpha, not
    all were in the same commit. Tor commit 9b4aa8d2abbce713 adds V1
    BW, CIRC, NEWDESC, ORCONN, and STREAM.  Tor commit c2f6fe9b85255185 adds
    V1 logging (i.e. DEBUG, INFO, etc.).  Tor commit 64fc1f7bae174d62 adds
    ADDRMAP.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 86524bf589014f2f6d7a4a9704cc0801663654b2
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Fri Dec 7 04:03:19 2012 -0700

    Begin version requirements for events
    
    Check that the tor process is of a minimum version able to generate
    the events requested.  Checking versions here improves on the
    'Unrecognized event "XXX"' response from the tor process.
    
    The SETEVENT control message was added to tor version 0.0.9-pre5
    (commit 347d3f9d629df74b).  But control protocol V1 changed the message
    syntax in tor version 0.1.1.1-alpha (commit 9b4aa8d2abbce713) and V0 is
    not supported.  Therefor, version 0.1.1.1-alpha is the baseline for any
    and all events in Stem.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit e2c69117b388abbbeee9240e67365e893b137cd3
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Tue Dec 4 18:38:27 2012 -0700

    Fix various pydoc typos
    
    Corrections from Sean for our is_alive methods, get_info, and other methods.

commit 7fba50ff561578f746ceb3383b1bb16d709a14e3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 8 14:13:43 2012 -0800

    Supporting TAKEOWNERSHIP when spawing a tor process
    
    Adding a 'take_ownership' option to the stem.process module's launch_tor*()
    functions. When it's set we'll do the following...
    
    * Assert ownership over the tor process that we spawn by setting its
      __OwningControllerProcess to be our pid.
    
    * If a Controller connects to it then we replace our ownership via the pid with
      ownership via the socket connection (issuing a TAKEOWNERSHIP request and
      dropping __OwningControllerProcess).
    
    Thanks to lunar for the initial TAKEOWNERSHIP patch on #7666, and rransom for
    his advice on #7667.

commit 885a294646703a537c37cd2a5ac9aa8728561744
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 8 12:35:34 2012 -0800

    Re-attaching event listeners
    
    When a controller disconnects then reconnects we want our event listeners to
    resume getting events. There's a couple things about this commit that I really
    don't like...
    
    * It includes a hack in our msg() method to add the hook.
    * Our tests include sleep() calls. This sucks since it makes our tests take
      considerably longer, and can break on systems under heavy load.

commit 3eb1da90792de75dc55492e37c3785dd0db28bc7
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 7 18:31:01 2012 -0800

    Reverting 'Tidying up event testing boilerplate'
    
    This reverts commit fb0aec5d95e9d2e6d64b7ab7c51018f3581565c5.

commit 700b54c10c4f5cf37f250189fd76ffa6bd24cca6
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 7 17:30:53 2012 -0800

    Prompt notice when tor is starting
    
    When the prompt script is first ran it can take a while for tor to start.
    Giving a notice that tor is starting.
    
    This also changes the Controller instance variable from 'control' to
    'controller'. I'm trying to standardize on the later to avoid a naming confict
    with stem.control imports.

commit 0d01ad60bb2a4ea2f0ac1f9626360d5db22858ea
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 7 09:13:46 2012 -0800

    Support for bridge-ip-versions extrainfo desc lines
    
    Adding parsing support for 'bridge-ip-versions' lines in extra-info
    descriptors. These are fields only relevant to bridges (though the spec doesn't
    prohibit them from appearing elsewhere). They're pretty much the same thing as
    the IPVersions in CLIENTS_SEEN events, so borrowing its parsing.
    
    Spec addition: https://gitweb.torproject.org/torspec.git/commitdiff/9a518d9
    Metrics-lib addition: https://gitweb.torproject.org/metrics-lib.git/commitdiff/17e9149

commit bd6e5cea6525f76b95966b023a1c1fa88065adc3
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 7 08:48:12 2012 -0800

    Support for geoip6-db-digest extrainfo desc lines
    
    Adding parsing support for the newish 'geoip6-db-digest' lines in extra-info
    descriptors. Lacking this support has been breaking our integ tests for a
    while, so I was able to exercise this both with the unit test addition and an
    actual consensus.
    
    Bug: https://trac.torproject.org/7354
    Spec addition: https://gitweb.torproject.org/torspec.git/commitdiff/9a518d9
    Metrics-lib addition: https://gitweb.torproject.org/metrics-lib.git/commitdiff/43b9390

commit 9c0cc18909a1dd16cc810cf3178acfbcb0bc064f
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 7 07:22:43 2012 -0800

    Avoiding address type query for cached get_mask() requests
    
    Oops. There's no reason to call get_address_type() when we already have a
    lazily loaded _mask attribute.

commit 45403096a550186cfbecc035cd54c125e3b0237c
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 6 22:44:17 2012 -0800

    Reducing ExitPolicyRule's memory requirements
    
    Frequently there's several ExitPolicyRule entries per an exit policy, so when
    we pull the consensus we get quite a few instances of this class. Making the
    following changes to reduce the memory requirements...
    
    * Dropping the rule attribute. The string representation of the rule should be
      good enough, if not better for callers.
    
    * Replacing the address_type attribute with a method for getting it. This lets
      us store the address type as an integer within our class.
    
    * Replacing the mask attribute with a method for getting it. The ip mask
      representation is very rarely useful, so there's little reason to store it
      unless it's requested.
    
    * Lazily loading the integer reprentation of our address and mask, both
      speeding up our constructor and avoiding it entirely if our caller never uses
      is_match()
    
    This lowers the memory requirement for loading the full consensus on my netbook
    from 5.5% to 5.1% (a 7% drop). This is a drop in the bucket compared to the
    prior commit, but between the faster constructor runtime and squeezing out a
    little more performance it's still worth it.

commit bbd702c81e1923bfbab236d0d3649fd2f87b3d95
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 6 22:05:07 2012 -0800

    Vastly lowering memory usage for exit policies
    
    A pox upon good indentions. When dealing with microdescriptor exit policies I
    enumerated all ports involved, the goal being to provide constant time
    is_match() lookups. We certainly did that, but there's quite a few valid ports
    and enumerating them for all relays has consistantly cost so much memory on my
    system that it triggers the OOM killer.
    
    Dropping the port set optimization. Memory usage trumps is_match() performance.
    Reading the full consensus into memory now takes 5.5% of the memory on my
    netbook, verses before where it would locked me up.

commit 51141d37b9a26a8a05cfa09cf85f3939ab79011a
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 6 09:02:35 2012 -0800

    Skip pointless extra read of network status documents
    
    When parsing a network status document we read to the end of the router status
    entries solely to determine where the end of the file is. This is a waste of
    time and memory. Instead, checking for the end of the section as we go along.

commit a71163021f1731302c7f922ce4e5b3191d2fb938
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Dec 6 20:05:18 2012 +0530

    Improve Controller.extend_circuit documentation
    
    Realized it might not be that obvious how paths are specified,
    and added some sample code to the Controller.extend_circuit
    documentation.
    
    Fixed a typo/grammer.

commit 7882d3cfeec98a478046a779e531a6013558c348
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Dec 6 19:39:16 2012 +0530

    Controller.extend_circuit now checks for failures properly

commit 69f72efc9367092c989819bb2e408598ecc9dcdb
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 4 08:37:59 2012 -0800

    Checking for None by identity
    
    According to PEP 290 comparisons to None should be done via identity (the 'is'
    keyword):
    
    http://www.python.org/dev/peps/pep-0290/#testing-for-none
    
    I had already been doing this, albeit inconsistently. Inequality checks were
    still using '!=' - swapping all None comparisons to be by identity.

commit d50b8a06744a8476bc16c04268fef75b20712b2b
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 4 08:21:09 2012 -0800

    Purging use of the variable 'i'
    
    A pox upon whoever taught me to use variables like i, j, and k for loop
    constants. It's ungrepable and a practice that should be purged via fire and
    pitchforks.
    
    I'm keeping it in for list interpolation (since it's localized to a single
    line), but replacing it with 'index' everywhere else. I'm also replacing a few
    points that do...
    
    for i in range(len(my_list)):
      my_item = my_list[i]
    
    ... with calls to enumerate().

commit 7011a6a2f1c306e204eb6e518604911d5561fc71
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 4 08:00:58 2012 -0800

    Including all entries in the Event's positional_args
    
    Fixing a semi-bug where the Event's positional_args only contained entries that
    we didn't recognize. This was a bug according to the Event's pydoc, but by
    design according to an inline comment. I'm not really sure what I did that - oh
    well...

commit 115607af56f7e8497c9c3343a4f7eae850b7a8a9
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 4 07:53:23 2012 -0800

    Comments for private Event attributes
    
    Adding comments to describe arguments like _POSITIONAL_ARGS and _SKIP_PARSING.
    Also dropping _QUOTED entries for key/value entries since it's no longer being
    used for that purpose.

commit fb0aec5d95e9d2e6d64b7ab7c51018f3581565c5
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 3 23:20:24 2012 -0800

    Tidying up event testing boilerplate
    
    Most event unit tests initially check two things...
    
    * That the event class matches an expected value.
    * That the string content for the event matches what we were constructed with.
    
    Blending these two checks with the function that provides us with events.

commit ae1f59b6b0ec1550434bc1508c7cd9ce866fa616
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 3 21:04:10 2012 -0800

    Better handling quoted key/value mappings in events
    
    We were being pretty picky about key/value mappings with quoted values, only
    accepting them if they were in both our _KEYWORD_ARGS and _QUOTED. The trouble
    with this was...
    
    1. It make our parsing code more convoluted than it needed to be.
    
    2. If tor added a new quoted key/value mapping then it would likely break our
       parser. This is because we'd mis-interpret the value as being positional
       fields. For instance if...
    
       650 MY_EVENT NEW_USER username=atagar
    
       ... became...
    
       650 MY_EVENT NEW_USER username=atagar message="hello world"
    
       Then we'd misinterpret the event as having no keyword arguments, and rather
       have four positional arguments (the last being 'world"').
    
       This isn't strictly wrong according to the control-spec. It's good about
       specifiying how new positional and keyword arguments should be handled, but
       quiet about quoted values.
    
       However, accepting quoted mappings by default is more intuitive and less
       likely to cause sadness down the road so making it so.

commit 22b4ced8548202a5715119a1d331900e6b3e8fff
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 3 20:41:53 2012 -0800

    Simplify check for if event attributes are unrecognized
    
    Whenever event parsing handles an attribute for which we have an enum it checks
    to see if the value is recognized by stem. If not then it logs at the INFO
    level.
    
    Replacing these boilerplate checks with a helper function. I'm also adding a
    test to check that we're actually logging these messages.

commit fb03666ebe9fa1ec134e6260bb79b6ed363e9994
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 3 19:22:55 2012 -0800

    Revisions for STREAM_BW support
    
    Couple very minor tweaks to the prior commits:
    
    * Adding STREAM_BW to the Event enum table.
    
    * Chaning is_valid_stream_id() to be a function rather than a module alias.
      This is effectively the same thing, but is less likely to confuse sphinx and
      imho is a bit more intuitive.

commit b3d91930be616ad7b7de10e3bbe581637583202c
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 1 13:50:48 2012 -0700

    Add unit test for STREAM_BW event handler
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit ac4eac1f15a3e8fce01bfe7e3f92aeff077894a5
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 1 13:28:04 2012 -0700

    Add STREAM_BW event handler
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 959e1506892c5f11e24a9a4fdfb3eafd73b4bd56
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 1 13:27:35 2012 -0700

    Add is_valid_stream_id helper function
    
    This is an alias to is_valid_circuit_id since stream and circuit
    identifiers follow the same pattern.
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit b3a1cbc538dced7c97da7cd3cd48f42937ed3565
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sat Dec 1 13:41:10 2012 -0700

    Add unit tests for is_valid_circuit_id function
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 3ba521dfa2f2a6f023e7b92fa24bcfb910f1fe78
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 3 09:00:49 2012 -0800

    Support for CIRC_MINOR events
    
    Implementation and testing for CIRC_MINOR events. Snagged test data by
    listening for events while bootstrapping.

commit bb458df4ac47c1405ce7e7f7e3e15ececf485b4b
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 3 08:21:50 2012 -0800

    Support for CONF_CHANGED events
    
    Implementation and testing for CONF_CHANGED events. Got the test data by
    issuing a SETCONF request...
    
    SETEVENTS CONF_CHANGED
    250 OK
    SETCONF ExitNodes=caerSidi MaxCircuitDirtiness=20 ExitPolicy
    650-CONF_CHANGED
    650-ExitNodes=caerSidi
    650-ExitPolicy
    650-MaxCircuitDirtiness=20
    650 OK
    250 OK

commit 47f608b05c0085a5acfb76b08a9efe916b1e09e9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 2 22:31:22 2012 -0800

    Support for SIGNAL events
    
    Implementation and testing for SIGNAL events. This moves tor's signals to an
    enum, which also makes the Controller's signal() method a little nicer.

commit 0687e99789827aa43dd8523006eb49aed837c2ef
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 2 20:20:33 2012 -0800

    Support for BUILDTIMEOUT_SET events
    
    Impementation and testing for BUILDTIMEOUT_SET events. Snagged test data by
    listening for the events while bootstrapping...
    
    650 BUILDTIMEOUT_SET COMPUTED TOTAL_TIMES=124 TIMEOUT_MS=9019 XM=1375 ALPHA=0.855662 CUTOFF_QUANTILE=0.800000 TIMEOUT_RATE=0.137097 CLOSE_MS=21850 CLOSE_RATE=0.072581
    650 BUILDTIMEOUT_SET COMPUTED TOTAL_TIMES=125 TIMEOUT_MS=8915 XM=1375 ALPHA=0.860984 CUTOFF_QUANTILE=0.800000 TIMEOUT_RATE=0.136000 CLOSE_MS=21850 CLOSE_RATE=0.072000
    650 BUILDTIMEOUT_SET COMPUTED TOTAL_TIMES=126 TIMEOUT_MS=8923 XM=1375 ALPHA=0.860563 CUTOFF_QUANTILE=0.800000 TIMEOUT_RATE=0.134921 CLOSE_MS=21850 CLOSE_RATE=0.071429
    650 BUILDTIMEOUT_SET COMPUTED TOTAL_TIMES=127 TIMEOUT_MS=8814 XM=1375 ALPHA=0.866231 CUTOFF_QUANTILE=0.800000 TIMEOUT_RATE=0.133858 CLOSE_MS=21850 CLOSE_RATE=0.070866

commit 42872dd08e81d6b341654ab85969df2ed77a7397
Merge: e1772bb c1e88e4
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 2 18:30:14 2012 -0800

    Tor event handling
    
    Merging support for tor's asynchronous events. This isn't entirely complete,
    missing...
    
    * STREAM_BW (being added in https://trac.torproject.org/7616)
    * BUILDTIMEOUT_SET
    * SIGNAL
    * CONF_CHANGED
    * CIRC_MINOR
    
    ... but better not to let the perfect be the enemy of good. This branch was
    partly code reviewed by robinson on...
    
    https://trac.torproject.org/7328

commit c1e88e4257c606593ecd2abf4212cc5266b34bf4
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 2 18:24:54 2012 -0800

    Support for NEWCONSENSUS events
    
    Implementation and testing for NEWCONSENSUS events. This is pretty much the
    exact same thing from our perspective as NS events.

commit 92419a54b83d80d7b124482df9bdf42fb931c7a0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 2 17:23:02 2012 -0800

    Support for CLIENTS_SEEN events
    
    Implementation and testing for CLINETS_SEEN events. This would be a difficult
    event to get examples for, but luckily the spec has one.

commit 92af0a5c0f9c6924cfddfb6388f581f3b6014732
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 2 16:16:20 2012 -0800

    Support for NS events
    
    Implementation and testing for NS events. To get test data I wiped my
    cached_consensus, then listened for NS events while bootstraping. The first
    time I did this it provided a flood of entries (probably the whole consensus),
    but then when I restarted tor and tried again it just proved one.

commit 2585969d6efbd8b1a6660d5de1ede437792dc22b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 2 14:02:21 2012 -0800

    Support for GUARD events
    
    Implementation and testing for GUARD events. The spec is a little vauge, so
    we're missing enum descriptions and a good understanding of the 'Name'
    attribute...
    
    https://trac.torproject.org/7619
    
    Got the test data by listening for GUARD events while bootstrapping...
    
    650 GUARD ENTRY $5D0034A368E0ABAF663D21847E1C9B6CFA09752A BAD
    650 GUARD ENTRY $071C2D9D0E03D75FF30DCCA850BB29CA4A4BB383 BAD
    650 GUARD ENTRY $36B5DBA788246E8369DBAF58577C6BC044A9A374 NEW
    650 GUARD ENTRY $071C2D9D0E03D75FF30DCCA850BB29CA4A4BB383 GOOD
    650 GUARD ENTRY $5D0034A368E0ABAF663D21847E1C9B6CFA09752A GOOD

commit 4959378f9aa83d96bc1431d72bea6fa7a1b6e2c7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 2 11:51:34 2012 -0800

    Allowing zero Terget port in STREAM events
    
    The STREAM event's target port can be zero in DNS_REQUEST requests. Caught
    thanks to robinson on...
    
    https://trac.torproject.org/7598

commit e59de80010d1894c5eecb7d4b9d953ee195a5259
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 2 03:00:51 2012 -0800

    Testing for STATUS_* events
    
    I had grand plans to have a StatusEvent subclass for each of the causes.
    However, on reflection that would encompass a bundle of work that's almost as
    large as all of the other events put together. It might still be a neat thing
    to do someday, but there's plenty of higher priority things to do first.
    
    Adding tests for all of the event examples that I collected earlier. This
    includes a bit of a hack to make quoted key/value mappings work. Previously
    we've only parsed quoted mappings if they belonged to the _QUOTED listing, but
    STATUS_* events sprinkle them all around so for forward compatability we can't
    simply enumerate them.

commit aeebbe762579951b4a78ddc8c68347d2d181fc49
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 2 02:05:09 2012 -0800

    Handling typo in STATUS_SERVER event name
    
    Prior to tor 0.2.0.22 STATUS_SERVER events claimed to be 'STATUS_SEVER'.
    Accepting both names as per the spec's suggestion.

commit 1af965f1d12f5a58e921559901a2b7f0ca1e4cba
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 30 23:04:07 2012 -0800

    Support for STREAM events with a DIR_FETCH purpose
    
    The SOURCE_ADDR field of STREAM events can have zero as their port. Fixing this
    and adding a test. This was caught thanks to robinson on...
    
    https://trac.torproject.org/7598

commit 628dd1f3246566bf4d7b4b2e2932f459bec32986
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Nov 29 18:49:50 2012 -0800

    Initial support for STATUS_* events
    
    The STATUS_GENERAL, STATUS_CLIENT, and STATUS_SERVER events all follow a nice,
    well specified format. The specific action instances for these events warrant
    subclasses, but to start with adding the StatusEvent class which covers the
    basic attributes.

commit f7cc47eae4e733872ba11687a81fc7cddb22fb1e
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Nov 27 08:48:50 2012 -0800

    Filling in AuthDescriptorAction descriptions
    
    The AUTHDIR_NEWDESCS spec was clarified in 'https://trac.torproject.org/7533'.
    I thought that I'd fixed all the TODO notes, but guess I missed this one.

commit 0f4e52bf20ae82a8951ed910639f0e4fa41aeac8
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 23 19:54:58 2012 -0800

    Parsing AUTHDIR_NEWDESCS events
    
    The specification for the AUTHDIR_NEWDESCS event type was expanded in...
    https://gitweb.torproject.org/torspec.git/commitdiff/a1331ca
    
    We still don't have an example of the event so no tests, but it's enough to
    implement the parsing. It would be nice if the event type included a field for
    the descriptor type, but oh well.

commit af527bdc1b87586c201ffd7fa288f40e29e33598
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 23 17:01:52 2012 -0800

    Dropped StreamPurpose.UPLOAD_DESC
    
    Dropping UPLOAD_DESC, which was evidently never used...
    https://gitweb.torproject.org/torspec.git/commitdiff/36d7d05

commit df9a051783a94b831aa1c93c8592d6c46397a229
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 23 12:36:28 2012 -0800

    Adding descriptions for ORStatus and ORClosureReason
    
    Spec has been expanded with descriptions for the ORCONN fields. I'm still not
    sure what these events mean nor what many of these values are actually trying
    to indicate, but it's enough for me to parrot the descriptions.
    
    bug report: https://trac.torproject.org/7513
    fix: https://gitweb.torproject.org/torspec.git/commitdiff/036e256

commit 404025ec9a7f0614742ce35ae13a1a001e4d9f9e
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 23 11:43:13 2012 -0800

    Documenting the StreamPurpose values
    
    Description for the STREAM event's PURPOSE field were added to the spec to
    reflecting that here.
    
    bug report: https://trac.torproject.org/7508
    fix: https://gitweb.torproject.org/torspec.git/commitdiff/e23a24a

commit 4ce0be17442a4c21612eefb6cbe426aebbfc722f
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 23 11:31:14 2012 -0800

    Adding CircClosureReason.MEASUREMENT_EXPIRED description
    
    Spec was updated with a description so reflecting that here.
    
    bug report: https://trac.torproject.org/7506
    fix: https://gitweb.torproject.org/torspec.git/commitdiff/de62160

commit 92ab865999a3e6a4ed21311560e4b15c22ad4139
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 23 11:02:44 2012 -0800

    Replacing GMT refrences with UTC
    
    Accouning for the spec change...
    https://gitweb.torproject.org/torspec.git/commitdiff/d581e63

commit 7f73e1c2c92bd949c29b1430200778d46be4d0cc
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Nov 21 09:05:40 2012 -0800

    Supporing DESCCHANGED events
    
    Adding support for DESCCHANGED events. These are trivial since... well, they
    don't contain any attributes. I'm also filling in some missing EventType enum
    documentation and changing the API docs for events to automatically pick up new
    event types (there's no point in using autoclass rather than automodule in this
    case).

commit 84ec250529b81cb86899897784602e6c41fb8d98
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Nov 20 09:37:54 2012 -0800

    Stub for AUTHDIR_NEWDESCS events
    
    Implementing the parts that I can for AUTHDIR_NEWDESCS events. The spec doesn't
    provide enough detail for us to parse them (https://trac.torproject.org/7533)
    and I need an example of an event before I can test them
    (https://trac.torproject.org/7534).

commit ab6e7a365cfa1e31b269f1df0e722ea11a2c53b1
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 19 00:09:09 2012 -0800

    Support for ADDRMAP events
    
    There's a special spot in hell for whoever decided to allow for quoted values
    in events. This implements and adds testing for ADDRMAP events. Unlike TorCtl
    we aren't falling back on a regex for the... er, 'wonderful' quoted stuff, but
    rather including quoted value support in the Event parser.
    
    Got test data by visiting a few sites in TBB...
    
    650 ADDRMAP check.torproject.org 38.229.72.22 "2012-11-18 22:48:34" EXPIRES="2012-11-19 06:48:34"
    650 ADDRMAP ocsp.digicert.com 5.63.145.124 "2012-11-18 21:53:42" EXPIRES="2012-11-19 05:53:42"
    650 ADDRMAP www.atagar.com 75.119.206.243 "2012-11-19 00:50:13" EXPIRES="2012-11-19 08:50:13"

commit 54f81be931196e92870b09119fa7e02a7a62cdbf
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 18 21:06:29 2012 -0800

    Support for NEWDESC events
    
    Implementing and testing NEWDESC events. Got test data by listning for events
    right after starting a TBB instance...
    
    650 NEWDESC $9C2007BE0D3785D5BAB204F0CFE07CF29F4B75EB~Unnamed
    650 NEWDESC $A10C4F666D27364036B562823E5830BC448E046A=DFRI1
    650 NEWDESC $A1130635A0CDA6F60C276FBF6994EFBD4ECADAB1~tama
    650 NEWDESC $A4E3D4A42E68D8302E4A3EA5D117EDCF0B157F72=tor1rufus
    650 NEWDESC $B3BE230EEA146002C4A2A436B4EF7D897DE6C9B6~WombleNode01
    650 NEWDESC $B3FA3110CC6F42443F039220C134CBD2FC4F0493=Sakura
    650 NEWDESC $B4BE08B22D4D2923EDC3970FD1B93D0448C6D8FF~Unnamed
    650 NEWDESC $B839D6CCC5438A21E547DB62043CB0192C068F51=brajah1
    650 NEWDESC $B8B29766ACFADF832ABD1BCAA86497A20F54B584=stillhavenoipinfo
    650 NEWDESC $BBE409F5791DAA52C2C3C9117CBA5AA55F3E2E88=Rarity
    650 NEWDESC $BE938957B2CA5F804B3AFC2C1EE6673170CDBBF8=Moonshine
    650 NEWDESC $C863FB2A6109C9CE2993C8855BAC59583B15475B=coinet
    650 NEWDESC $CA37697301883DE9E895F642D76F3CD31E855434=Bitcoin
    650 NEWDESC $CBC5F626F16052B2B987106D4845DF0B5197901D=mergeslittletornode
    650 NEWDESC $CC19411EDD0BE139E8828EECE2F11E5DFE5F5C09~RadiantPickle
    650 NEWDESC $CF334A712B78A84C6E8F2FF71987122E17951176=skye
    ...

commit 62b6f153e66a3b1a8adff3ac837c161e43576f2d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 18 14:47:27 2012 -0800

    Reordering event tests
    
    We're using an alphabetical order elsewhere for the events (except logging
    events which are batched together at the top), so matching that in the unit
    tests.

commit a1a5784d480421af4394b3e012d6f4bbb8ee6c8b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 18 14:44:05 2012 -0800

    Support for ORCONN events
    
    Implementation and tests for ORCONN events. These have several holes in its
    documentation (https://trac.torproject.org/7513) so I'm not really sure what
    these events actually are. Reguardless, got some samples by connecting to TBB
    and issuing a NEWNYM.
    
    AUTHENTICATE
    250 OK
    SETEVENTS ORCONN
    250 OK
    650 ORCONN $1D024F41EDBF3F061E1341D516543090D8A44B42=AccessNowKromyon21 CONNECTED
    650 ORCONN $7ED90E2833EE38A75795BA9237B0A4560E51E1A0=GreenDragon CONNECTED
    650 ORCONN $A1130635A0CDA6F60C276FBF6994EFBD4ECADAB1~tama CLOSED REASON=DONE

commit dff8a0edf35a6e52f2d0e1d52fee71f891146c23
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 18 13:03:46 2012 -0800

    Alphabetizing event types
    
    I had been following an order similar to the control spec but on reflection
    that doesn't make sense. Ordering these alphabetically would be better.

commit d1cd96db70da3a737f6fee24354d61f5e52893e5
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 18 13:00:50 2012 -0800

    Few spelling corrections

commit c96c76ce16d42aeef6641bfeaaf937673934dc5c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 18 12:55:52 2012 -0800

    Moving attribute enums to top level module
    
    Enumerations for our event attributes don't really belong in the control
    module. They're a bit more like our exceptions in that they're a data type that
    might be used in multiple spots. Hence moving the enums to live with them.

commit d735e1e267877d9e9e7fa0686c2cd49c63930f0b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 18 11:48:00 2012 -0800

    Support for STREAM events
    
    Implementaton and tests for STREAM events. I got the test data by...
    
    * starting TBB
    
    * used netstat to get the control port (shouldn't have needed to do this -
      https://trac.torproject.org/7512)
    
    * connecting to it with telnet
      * AUTHENTICATE
      * SETEVENTS STREAM
    
    * visited google's front page in firefox
    
    Full test data:
    
    AUTHENTICATE
    250 OK
    SETEVENTS STREAM
    250 OK
    650 STREAM 18 NEW 0 encrypted.google.com:443 SOURCE_ADDR=127.0.0.1:47849 PURPOSE=USER
    650 STREAM 18 SENTCONNECT 26 encrypted.google.com:443
    650 STREAM 19 NEW 0 encrypted.google.com:443 SOURCE_ADDR=127.0.0.1:47850 PURPOSE=USER
    650 STREAM 19 SENTCONNECT 26 encrypted.google.com:443
    650 STREAM 18 REMAP 26 74.125.227.129:443 SOURCE=EXIT
    650 STREAM 18 SUCCEEDED 26 74.125.227.129:443
    650 STREAM 19 REMAP 26 74.125.227.129:443 SOURCE=EXIT
    650 STREAM 19 SUCCEEDED 26 74.125.227.129:443
    650 STREAM 20 NEW 0 encrypted.google.com:443 SOURCE_ADDR=127.0.0.1:47851 PURPOSE=USER
    650 STREAM 20 REMAP 0 74.125.227.129:443 SOURCE=CACHE
    650 STREAM 20 SENTCONNECT 26 74.125.227.129:443
    650 STREAM 21 NEW 0 encrypted.google.com:443 SOURCE_ADDR=127.0.0.1:47852 PURPOSE=USER
    650 STREAM 21 REMAP 0 74.125.227.129:443 SOURCE=CACHE
    650 STREAM 21 SENTCONNECT 26 74.125.227.129:443
    650 STREAM 20 REMAP 26 74.125.227.129:443 SOURCE=EXIT
    650 STREAM 20 SUCCEEDED 26 74.125.227.129:443
    650 STREAM 21 REMAP 26 74.125.227.129:443 SOURCE=EXIT
    650 STREAM 21 SUCCEEDED 26 74.125.227.129:443
    650 STREAM 22 NEW 0 www.google.com:443 SOURCE_ADDR=127.0.0.1:47853 PURPOSE=USER
    650 STREAM 22 SENTCONNECT 26 www.google.com:443
    650 STREAM 23 NEW 0 www.google.com:443 SOURCE_ADDR=127.0.0.1:47854 PURPOSE=USER
    650 STREAM 23 SENTCONNECT 26 www.google.com:443
    650 STREAM 21 CLOSED 26 74.125.227.129:443 REASON=CONNRESET
    650 STREAM 20 CLOSED 26 74.125.227.129:443 REASON=CONNRESET
    650 STREAM 22 REMAP 26 74.125.227.147:443 SOURCE=EXIT
    650 STREAM 22 SUCCEEDED 26 74.125.227.147:443
    650 STREAM 23 REMAP 26 74.125.227.147:443 SOURCE=EXIT
    650 STREAM 23 SUCCEEDED 26 74.125.227.147:443
    650 STREAM 24 NEW 0 ocsp.thawte.com:80 SOURCE_ADDR=127.0.0.1:47855 PURPOSE=USER
    650 STREAM 24 SENTCONNECT 26 ocsp.thawte.com:80
    650 STREAM 25 NEW 0 ocsp.thawte.com:80 SOURCE_ADDR=127.0.0.1:47856 PURPOSE=USER
    650 STREAM 25 SENTCONNECT 26 ocsp.thawte.com:80
    650 STREAM 24 REMAP 26 199.7.52.72:80 SOURCE=EXIT
    650 STREAM 24 SUCCEEDED 26 199.7.52.72:80
    650 STREAM 25 REMAP 26 199.7.52.72:80 SOURCE=EXIT
    650 STREAM 25 SUCCEEDED 26 199.7.52.72:80
    650 STREAM 26 NEW 0 ssl.gstatic.com:443 SOURCE_ADDR=127.0.0.1:47857 PURPOSE=USER
    650 STREAM 26 SENTCONNECT 26 ssl.gstatic.com:443
    650 STREAM 27 NEW 0 ssl.gstatic.com:443 SOURCE_ADDR=127.0.0.1:47858 PURPOSE=USER
    650 STREAM 27 SENTCONNECT 26 ssl.gstatic.com:443
    650 STREAM 23 CLOSED 26 74.125.227.147:443 REASON=CONNRESET
    650 STREAM 26 REMAP 26 74.125.227.143:443 SOURCE=EXIT
    650 STREAM 26 SUCCEEDED 26 74.125.227.143:443
    650 STREAM 27 REMAP 26 74.125.227.143:443 SOURCE=EXIT
    650 STREAM 27 SUCCEEDED 26 74.125.227.143:443
    650 STREAM 25 CLOSED 26 199.7.52.72:80 REASON=DONE
    650 STREAM 27 CLOSED 26 74.125.227.143:443 REASON=CONNRESET
    650 STREAM 26 CLOSED 26 74.125.227.143:443 REASON=DONE
    650 STREAM 24 CLOSED 26 199.7.52.72:80 REASON=DONE
    650 STREAM 22 CLOSED 26 74.125.227.147:443 REASON=DONE
    650 STREAM 19 CLOSED 26 74.125.227.129:443 REASON=DONE
    650 STREAM 18 CLOSED 26 74.125.227.129:443 REASON=DONE
    Connection closed by foreign host.

commit a349a01fece5534195a6619d72df4944065001b8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 17 18:49:23 2012 -0800

    Support for CIRC events
    
    Implementation and testing for CIRC events. This work also concerns the
    'GETINFO circuit-status' method, which is defined as providing the same output
    as CIRC events. This is part of the reason why I put the enums for the event
    attributes in 'stem.control'.

commit d7dc386533f0b7885f2966922b9cb051c1b89384
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 11 12:36:37 2012 -0800

    Enum docs for stem.control.EventType
    
    I had been doing a table for the EventType enum under the add_event_listener()
    method. This made sense and matched with what I'm doing for enumeration
    documentation except that this should be a interlinkable :data: field in our
    header.

commit 119579bd8c53c15e949e9a77a0c30e12be07d1cd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 11 12:28:51 2012 -0800

    Enum docs for stem.control.State
    
    While documenting enums I must have messed up by search term for stem.control
    since, rather than adding enum documentation, I concluded that there weren't
    any enums and removed the import statement. Fixing this.

commit 7ba579494215a5353cff1db9fbf7e59d6b666af9
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 5 09:14:07 2012 -0800

    Support for logging events
    
    Adding an event class for DEBUG, INFO, NOTICE, WARN, and ERR.

commit 1d2d684e3975e0e55369485b3d75155e90ea6aef
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 5 08:36:05 2012 -0800

    BW event unit tests
    
    Adding a few simple unit tests for the BandwidthEvent class.

commit 1d56eb59e256977d34f9d3d77e4e0b8c2df49f86
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 5 08:21:30 2012 -0800

    EventType to Event subclass table
    
    Adding a table that'll list the EventType enumerations and map them to Event
    subclasses.

commit 1a6a4d55701aaa199c14e1ef1ee1b6455922d761
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 5 08:02:15 2012 -0800

    Unit test for add_event_listener() example
    
    Adding a unit test for the pydoc example in add_event_listener(). I really,
    really don't like how the tests for these documentation examples are turning
    out. The end result looks precious little like the example and is convoluted as
    hell.
    
    We should revisit how we do documentation example testing in the future, but
    that'll be a subproject in itself.

commit cc530bb2005f0e156c62fe322d5b43ffe16c3e99
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 5 07:20:49 2012 -0800

    Adding a add_event_listener() example
    
    On reflection the example that I included on ticket #7328 is pretty good at
    exemplifying the function's usage, so including it in our pydocs.

commit c35a626af19a5d22b8cc1ca1cef5bae212e3a726
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 4 21:06:05 2012 -0800

    BandwidthEvent class
    
    Adding a class to handle BW events. This is the easiest type of tor event to
    test because it's emitted every second. The other event types we'll probably
    need to go with unit tests.

commit cac0c434ca9a4ed1f9bb6aa940dba1e006b92471
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 4 18:38:28 2012 -0800

    Base class for events
    
    Adding a base class for tor events which does the initial parsing of positional
    and keyword arguments. This is using a similar design to the TorCtl patch I
    wrote for #3679 (but a little better).

commit ba280f94da9962e1a2b1bfaf22d5fbd17ffa3103
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 4 15:53:44 2012 -0800

    Adding event handling support to the Controller
    
    My initial plan was to add event listener support similar to TorCtl but with
    automated handling of the SETEVENT calls. While working on this meejah
    suggested accepting functors instead (like txtorcon). On reflection this is
    both much nicer for our callers and easier for us to support.
    
    Adding functions for adding and removing listeners, with a simple integ test.
    Next up will be the Event classes...

commit e1772bb9ad9ad2f6f085e2c0f489214d2f7fd6ee
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 23 12:14:12 2012 -0800

    Revised behavior for document signature methods
    
    The spec has been revised to explain how signature methods worked. I had
    assumed that they were included in microdescritor consensuses and excluded from
    standard consensuses. Turns out however that they can be included with both
    and, if excluded, have a default value of 'sha1'. This is much nicer from a
    parsing and behavior standpoint.
    
    bug report: https://trac.torproject.org/7072
    fix: https://gitweb.torproject.org/torspec.git/commitdiff/96427e8

commit 69c55aba4146861d0adead57850bea8fa9ded3ea
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Nov 29 18:55:10 2012 -0800

    Descriptor reader hung if malformed content was in an archive
    
    The descriptor reader was missing a catch clause when parsing archives, causing
    the reader to die (and callers to hang) if it contained invalidly formed
    descriptor data.
    
    Caught by eoinof on...
    https://trac.torproject.org/5810

commit 6f2c1c33931b6fda738468d03ee0f835f0caca6b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 25 13:50:04 2012 -0800

    Checking when EXTENDCIRCUIT is/isn't provided
    
    The EXTENDCIRCUIT's path argument is optional unless...
    
    * the ciruit id is non-zero
    * we're prior to tor version 0.2.2.9 where it was made optional:
      https://gitweb.torproject.org/tor.git/commitdiff/ac68704f07c2b703
    
    Pointed out by robinson and integrates his suggestions on...
    https://trac.torproject.org/6666

commit d82a70a4fb874ca295c1644e3c77f24afddcbf06
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Nov 28 21:16:04 2012 -0800

    Minor whitespace changes
    
    Just adding empty lines in a few spots to improve readability. Conventionally
    we include blank lines between pydoc description, :param:, :returns:, and
    :raises: entries. It also find that it helps to have empty lines around
    conditionals and comments, though that might just be a matter of taste.
    
    No functional changes.

commit 5da6b9790da266f96258c7c6d6a439ca2ef06529
Author: Eoin o Fearghail <eoin.o.fearghail@gmail.com>
Date:   Tue Nov 27 21:04:14 2012 +0000

    Refinements to previous checkin after code review/feedback
    
    cf https://trac.torproject.org/projects/tor/ticket/5810
    
    Removed most of the logging code
    _digest function now returns the digest in uppercase hex
    digest value is now calculated once & cached for evermore.
    moved key string manipulation code to a separate function as it is used
     more than once, cf _get_key_bytes()
    reverted change to test/integ/descriptor/server_descriptor as _digest now returns uppercase hex
    added some documentation to _digest()
    added some documentation to sign_descriptor_content()

commit e0095fbe54759c45cbf6d1b120d2b17b47a0ec21
Author: Eoin o Fearghail <eoin.o.fearghail@gmail.com>
Date:   Fri Nov 23 22:16:22 2012 +0000

    Implementing Relay Descriptor verification
    
    cf https://trac.torproject.org/projects/tor/ticket/5810
    
    1) Implemented relay descriptor verification using the python-crypto lib.
       Code is only run if python-crypto can be imported. [cf stem.prereq.is_crypto_available()]
       NOTE: constructing a RelayDescriptor will now raise an exception if invalid descriptor content is used.
    2) Refactored the digest() function in server_descriptor.py.
    3) Added a function to the mocking lib to sign a descriptor with an auto-generated key
    4) Add usage of new sign_descriptor_content() in unit tests where necessary.
    5) Updated the non-ascii-descriptor file to be correctly signed.
    6) Updated extra info descriptor test to use new fingerprint in non-ascii-descriptor file
    7) Removed server descriptor tests that do not make sense if data is being generated dynamically.
       e.g. Removed test fingerprint valid test, since data now dynamically generated.

commit 368fedfaffc6e86dc8febd1d6d82e54f91b6ad01
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 25 14:01:18 2012 -0800

    Whitespace fixes
    
    The prior couple patches lacked whitespace on empty lines, causing our
    whitespace checker to complain. I wonder if I should stop fighting with the
    world here - they're clearly wrong, but I'm pretty alone in this...

commit 3b6ac7bd10c38b9ef52d73bdbf134eced065e291
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Sun Nov 25 07:33:05 2012 -0700

    Fix close_circuit test
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit d682c8cd13a0f6bffcd056a2a938db2dfa65f500
Author: Sean Robinson <seankrobinson@gmail.com>
Date:   Thu Nov 22 08:09:42 2012 -0700

    Add a close_circuit method to Controller
    
    Signed-off-by: Sean Robinson <seankrobinson@gmail.com>

commit 4c65738d100ab0802a75cc4b9d64065c8a8ef881
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 25 13:03:46 2012 -0800

    Borrowing setup.py corrections from robinson
    
    Borrowing a couple things from robinson's patch on...
    https://trac.torproject.org/7565
    
    ... including the description, updated stem url, and provides/keywords args.
    Tested by simply running 'python setup.py build'.

commit 337f2221109ae61ed90fbf640d23fa434bed8049
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 25 12:11:59 2012 -0800

    Adding a setup.py
    
    I've had a setup.py sitting around since the GSoC mentor summit, but didn't
    check it in since I haven't made a method for testing the Python 3 conversion.
    
    robinson submitted a patch adding a setup.py so guess I should stop waiting on
    that.

commit e73c4f353612ad26b47ca2c66d8c6f4ac78f165b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 17 10:57:30 2012 -0800

    Disabling flaky bits of descriptor tests
    
    As per 'https://trac.torproject.org/7163' the network status and server
    descriptor integ tests are flaky right now. We read our cached consensus and
    network status document for a relay, then query the control socket for that
    relay. This really seems like it should work, but sometimes tor replies with an
    unknown key error (ie, saying the relay doesn't exist).
    
    There's no point in keeping this test while it's flaky. We should investigate
    what's going wrong at some point so keeping the ticket open for that.

commit 28d6bc73796a46ba0463768a2de35eeb3bc0a1fe
Author: Eoin o Fearghail <eoin.o.fearghail@gmail.com>
Date:   Fri Nov 16 18:57:38 2012 +0000

    Minor bug fix.
    The _get_descriptor_content function was adding whitespace
    to the relay descriptor that it was generating.
    This was incorrect.

commit 505b5a6f82222f4d16e1e1cdac96c66096a52bf1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 10 23:17:31 2012 -0800

    Adding link for stem's development wiki
    
    We don't have anything linking stem's front page to the development wiki. In
    general this is ok, the dev wiki is only of interest to people wanting to hack
    *on* stem, not use it. However, we should have at least one link to let new
    contibutors find it on their own.

commit 0e99a269b94f657575690734f4ffff15a17944e0
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 9 09:46:54 2012 -0800

    Revised enumeration documentation
    
    Our enumerations were only defined in our 'Module Overview' section which sucks
    for a few reasons...
    
    1. We can't interlink with the enum definitions.
    2. Blocks lack any formatting that could improve readablity.
    3. In the long term I'd like to replace the 'Module Overview' blocks with
       something better.
    
    Sphinx's python domain directives include a 'data' type, used for global
    values. That's exactly what our enums are so using the directive, with tables
    to list our enum values.

commit 07c5bae7fb4fd6139afcb2a14abdddad75580e0e
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Nov 8 08:29:01 2012 -0800

    Git image linked to gitweb rather than stem's repo
    
    Oops, not sure how I got this wrong. We want to link to our repository, not the
    list of projects.

commit 142450fc68d8df967bc99b632154f43d4b2e3451
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Nov 6 08:56:44 2012 -0800

    Including exceptions in api docs
    
    Moving the exceptions to stem/__init__.py meant that they were no longer
    covered my any of our automodule declarations. Adding a special 'Exceptions'
    section to the controller's api page.

commit 82747342cac4a7204b42928fe3a3f32fadea62c6
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Nov 6 19:08:03 2012 +0530

    Moving exceptions to stem's toplevel module
    
    Fixing #6357

commit a5c5ae8c3af4409b8fcac10a38b21a37723237c9
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Nov 6 08:09:06 2012 -0800

    Reverting comment change about tor error messages
    
    The prior spelling fixes condensed a comment that we had about the error
    message tor will provide when we use the wrong controller password. The change
    makes sense, but these lines are here to tell us the exact error message tor
    provides (after which it's hopefully reasonably obvious why we check for the
    "Password did not match HashedControlPassword" part of it).

commit bb0a3ace259e4f23eb53837908a3163c1c19e305
Author: Eoin o Fearghail <eoin.o.fearghail@gmail.com>
Date:   Tue Nov 6 00:00:51 2012 +0000

    Changing spelling of parseable to parsable, also one tiny whitespace fix.

commit 2a597415c0a8ca8203ee0282103ef0c606f1d887
Author: Eoin o Fearghail <eoin.o.fearghail@gmail.com>
Date:   Mon Nov 5 22:54:02 2012 +0000

    Squashed all my spelling changes into one commit

commit c58514337ec02d17260c9468ba17b1a15f959c82
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 4 11:04:42 2012 -0800

    stem.process integ test with RUN_SOCKET target broken
    
    Ok, this puzzled me for a while. Our stem.process integ tests make
    '/tmp/stem_integ' to be its temporary data directory for the run, then cleans
    it up afterward.
    
    This is great, except that the RUN_SOCKET causes us to run with the following
    in its torrc...
    
    ControlSocket /tmp/stem_integ/socket
    
    It's a coincidence that both use the same tmp directory, but it's not
    necessarily bad. What was bad, though, is that...
    
    * the stem.process tests broke since the mkdir call for stem_integ failed
    * if that hadn't failed our tearDown would have deleted stem_integ, and with it
      our control socket causing all further tests to fail
    
    We should probably use the tempfile module to make our stem.process data
    directory, but for now just being lazy and having it account for the RUN_SOCKET
    target.
    
    Caught by Ravi on...
    https://trac.torproject.org/7284

commit e6f95bb7d34d308b3cfdfe7c66c4d10ddb5756c2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 3 21:52:05 2012 -0700

    Adding a logo for stem
    
    Logo for the stem project, curtsey of the WP Clipart project.

commit 3581e963953805c9156589255b534d73e45e2156
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 3 21:18:47 2012 -0700

    Adding contact link
    
    Adding a link to my contact page. I'd prefer for people to contact tor-dev@,
    but I've learned from experience that I'll get quite a bit more input by
    providing a text field that doesn't require subscription.

commit 0df574edf8bf004e55acef809772c5e0735616ce
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 3 20:26:39 2012 -0700

    Image credits and licenses
    
    Noting where the site images come from. Almost all of them are from the
    collection I've gathered of free image resources
    (http://www.atagar.com/freeMedia/).
    
    I've tried pretty hard to avoid images with unknown licenses but there is one
    exception: the anvil schematic for the API link. This is a fantastic image
    and I've searched quite a bit to figure out where it originally came from
    (found it on a blog but doubt that's the source). Pity...

commit e095e101160a0a739909143ae22996fb5a376e64
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 3 19:46:27 2012 -0700

    Link for the bug tracker
    
    Adding a link to our front page for a bug tracker page on trac that's specific
    to our project.

commit 9fa93cb5cb84a73defd7ccfcdffe82896684d03a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 3 17:12:52 2012 -0700

    Error when trying to parse the last line in router_status_entry.parse_file()
    
    When calling the Controller's get_network_statuses() method the last line would
    cause an exception because we tried to parse an empty string as a router status
    entry. Aborting when we get to the end instead.

commit 86d6b6114ad0929ee787cf454c9d93f1af3f9033
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Nov 1 23:20:25 2012 +0530

    Rewrite Controller.repurpose_circuit test to use a newly created circuit.
    
    Attempting to fix https://trac.torproject.org/projects/tor/ticket/7259 by
    using a newly created circuit instead of parsing the GETINFO circuit-status
    output.

commit 54e3970b674665d58b786b52af02b18bdb686136
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Oct 31 08:46:55 2012 -0700

    Supporting files in whitespace checks
    
    We called os.walk() when determining the files for which we want to check
    whitespace. However, when presented with a file rather than a directory this
    causes us to not check anything...
    
    >>> list(os.walk("/tmp/foo"))
    []
    
    This broke our attempts to check 'run_tests.py', and let a couple whitespace
    issues slip in. Fixing get_issues()'s handling for individual files.
    
    Issue caught by Eoin on...
    https://trac.torproject.org/7263

commit d46292f0dce4a189457b02a1ea951e845094c509
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Oct 31 08:40:21 2012 -0700

    Using absolute paths for whitespace checks
    
    We were using relative paths for our whitespace checks, which caused varying
    behavior based on our cwd...
    
    atagar@morrigan:~/Desktop/stem$ ./run_tests.py --unit
    ...
    TESTING PASSED (7 seconds)
    
    atagar@morrigan:~/Desktop/stem$ cd ..
    atagar@morrigan:~/Desktop$ stem/run_tests.py --unit
    ...
    WHITESPACE ISSUES
    * stem/example.py
      line 18   - indentation should match surrounding content (2 spaces)
      line 19   - missing 'with' import (from __future__ import with_statement)
      line 23   - indentation should match surrounding content (2 or 8 spaces)
      line 35   - indentation should match surrounding content (4 spaces)
      line 72   - line has trailing whitespace
      line 76   - indentation should match surrounding content (0 spaces)
      line 77   - indentation should match surrounding content (0 spaces)
    
    * stem/run_tests.py
      line 289  - indentation should match surrounding content (2 spaces)
      line 486  - line has trailing whitespace
    
    TESTING PASSED (19 seconds)
    
    Note that 'example.py' isn't part of stem. It's an untracked file that I have
    in the stem directory. The reason that it's being included in the whitespace
    check is that we're grabbing all python files under 'stem' which, now that
    we're one level up, is the whole project.
    
    Using absolute paths that are relative of run_tests.py so we get consistent
    results.

commit 56aac96d6213f28a6b597c640affc7a5a963bf75
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 29 19:26:42 2012 -0700

    Conficting DataDirectory for process tests
    
    The process module's integ tests start and stop tor instances. However, if
    you're already running tor then the DataDirectory of these instances will
    conflict with what you're already running...
    
    Oct 29 19:18:11.532 [notice] Tor v0.2.1.30. This is experimental software. Do not rely on it for strong anonymity. (Running on Linux i686)
    Oct 29 19:18:11.540 [warn] ControlPort is open, but no authentication method has been configured.  This means that any program on your computer can reconfigure your Tor.  That's bad!  You should upgrade your Tor controller as soon as possible.
    Oct 29 19:18:11.541 [notice] Initialized libevent version 1.4.13-stable using method epoll. Good.
    Oct 29 19:18:11.542 [notice] Opening Socks listener on 127.0.0.1:2777
    Oct 29 19:18:11.543 [notice] Opening Control listener on 127.0.0.1:2778
    Oct 29 19:18:11.543 [warn] It looks like another Tor process is running with the same data directory.  Waiting 5 seconds to see if it goes away.
    Oct 29 19:18:16.546 [err] No, it's still there.  Exiting.
    
    Issue spotted by gsathya on...
    https://trac.torproject.org/7251

commit 89ec87cfa3c119d96ee32d7debec0dde9e97cb51
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 29 19:07:03 2012 -0700

    Not properly aborting test when skip() is unavailable
    
    The test_cached_microdesc_consensus test is missing a return after its skip,
    causing us to fail if...
    
    1. We don't have a microdescriptor consensus.
    2. We're running python 2.5 or 2.6
    
    Spotted by gsathya on...
    https://trac.torproject.org/7250

commit 081228149f80448dc9097d7e260b2738792250ca
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 29 18:59:23 2012 -0700

    Helper function for descriptor tutorial
    
    The 'Mirror Mirror on the Wall' tutorial had some repetition due to having both
    an example for getting descriptors from the disk and through the control port.
    Moving that code to a helper function so I can drop the common parts from the
    second example.

commit aac44fe0a23f409e61484f8ce6c1b5c79a312714
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 29 09:03:52 2012 -0700

    DictWriter.writeheader() introduced in python 2.7
    
    The writeheader() method of the DictWriter class was added in python 2.7,
    breaking our 2.5 and 2.6 compatability...
    
    ======================================================================
    ERROR: test_multiple_descriptor_types
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/unit/descriptor/export.py", line 91, in test_multiple_descriptor_types
        self.assertRaises(ValueError, export_csv,)
      File "/usr/lib/python2.6/unittest.py", line 336, in failUnlessRaises
        callableObj(*args, **kwargs)
      File "/home/atagar/Desktop/stem/stem/descriptor/export.py", line 39, in export_csv
        export_csv_file(output_buffer, descriptors, included_fields, excluded_fields, header)
      File "/home/atagar/Desktop/stem/stem/descriptor/export.py", line 91, in export_csv_file
        writer.writeheader()
    AttributeError: DictWriter instance has no attribute 'writeheader'
    
    Noting the prereq in our pydocs and ignoring the 'header' flag if we can't
    support it.

commit b09de1093e7a7237d3825d5728126cbf2f15842b
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 29 08:47:26 2012 -0700

    OrderedDict substitute for older python versions
    
    In d30a628e I simplified the _get_descriptor_components() by using the
    collections.OrderedDict builtin. Unforutunately OrderedDict was introduced in
    python 2.7, breaking our 2.5 and 2.6 compatability...
    
    ======================================================================
    ERROR: test_voting_delay
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/unit/descriptor/networkstatus/document_v3.py", line 359, in test_voting_delay
        document = get_network_status_document_v3({"voting-delay": "12 345"})
      File "/home/atagar/Desktop/stem/test/mocking.py", line 782, in get_network_status_document_v3
        return stem.descriptor.networkstatus.NetworkStatusDocumentV3(desc_content, validate = True)
      File "/home/atagar/Desktop/stem/stem/descriptor/networkstatus.py", line 458, in __init__
        self._header = _DocumentHeader(document_file, validate, default_params)
      File "/home/atagar/Desktop/stem/stem/descriptor/networkstatus.py", line 541, in __init__
        entries = stem.descriptor._get_descriptor_components(content, validate)
      File "/home/atagar/Desktop/stem/stem/descriptor/__init__.py", line 289, in _get_descriptor_components
        entries = collections.OrderedDict()
    AttributeError: 'module' object has no attribute 'OrderedDict'
    
    Caught by eoinof on...
    
    https://trac.torproject.org/7244
    
    I'm fixing this by adding an OrderedDict substitute that's under the MIT
    license...
    
    http://pypi.python.org/pypi/ordereddict

commit 0afe836c94f83f3e11435cdb9f8a3f33dbb9f106
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 28 14:37:39 2012 -0700

    Fixing spelling of check_requriements()
    
    Spelling mistake caught by eoinof...
    https://trac.torproject.org/7245

commit 1d8cdb206e54354fa563f578615b3079da8c7abb
Author: Eoin o Fearghail <eoin.o.fearghail@gmail.com>
Date:   Sun Oct 28 20:45:23 2012 +0000

    Added a count of how many, if any, tests were skipped.
    
    This count is then displayed at the end of the test run as
    skipped tests may not otherwise be noticed!

commit 39d90df0f9f48c318a9e1ba4ca32836b40a6eb81
Merge: a00fb8a f31b19b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 28 13:54:05 2012 -0700

    Revised API documentation
    
    Stem's API docs were a monolithic dump of our pydocs. This was utterly
    unreadable so breaking it down into individual module pages. I also applied
    some uniform formatting changes to all of our pydocs to make them more reader
    friendly.

commit f31b19b966bc2f1243f58eebf8a1db0ce2876ff0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 28 13:49:17 2012 -0700

    Making the logging NullHandler private
    
    We have a null logging handler to avoid having the logging module give a
    warning about no handlers being present. There's no reason for stem's users to
    care about it so making it private.

commit 07c75fa254926b2e14ef91d8edc9e8e20170ce81
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 28 13:48:04 2012 -0700

    Spelling corrections
    
    Spell checked all of the stem/* files and the test utilites. Quite a few
    mistakes...

commit 1e1d84143b26212d3158f30555145687d3c653d3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 28 12:31:04 2012 -0700

    Moving AuthMethod docs to the module overview
    
    We were using definition entries for the AuthMethod enums which was nice, but
    gave them more emphasis than they deserved. Users won't usually touch a
    PROTOCOLINFO response directly so the enumeration really isn't that important.

commit a78ea67a521197b8c57c8556df4250ec5161b6b0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 28 12:17:20 2012 -0700

    Linking image and clone command to gitweb
    
    I disliked having a '(browse)' link for a couple reasons...
    
    1. people might copy it as part of the clone command if they're new to git
    2. it was pretty small considering its relevant importance
    
    Instead linking the clone command itself and the git image.

commit ebb3de86c9af92001845f3560f857615e1f633eb
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 19:32:43 2012 -0700

    Revised API docs for stem.response
    
    This is a little different from the other modules in that we're including
    stem.response.__init__ as an automodule then autoclasses for all of the
    contents. The response classes each have precious little documentation to
    having them each on their own page would be pointless.
    
    Also moving the AuthMethod enum to the connection so we can drop the
    protocolinfo module documentation (it wouldn't be included since we're using
    autoclass).

commit d2cd6ccb0c9df470478278e570f732beec973df6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 18:42:32 2012 -0700

    Revised API docs for stem.process

commit 04ea42cba22be4defeddf0d11803c23476612bdd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 18:16:34 2012 -0700

    Revised API docs for stem.control

commit 08fa765d219b24a8c0bb0142b71a940cd6fd7bf2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 17:42:38 2012 -0700

    Revised API docs for stem.socket

commit ec185b901de8829f12c7e0c1778f079e444ce14a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 17:26:35 2012 -0700

    Revised API docs for stem.connection

commit 4167d59d8fa9d72656381dec39c6a7315176ebed
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 16:57:34 2012 -0700

    Collecting API stubs under a directory
    
    Moving all the autodoc stubs under an api directory to keep things tidy.

commit 8459d01c2ce4feb9c661e9bf958da5169eae58a3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 15:53:41 2012 -0700

    Revised API docs for stem.util.tor_tools

commit 6de23cdc3fb45eb9493c9d3819585f414fecb9cf
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 15:49:06 2012 -0700

    Revised API docs for stem.util.term

commit d2a3acff92383ea95602acd4382cf648b5693da4
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 15:42:02 2012 -0700

    Revised API docs for stem.util.system

commit cc37ede99703c0db0a34a3b3125339af473fe810
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 15:35:37 2012 -0700

    Revised API docs for stem.util.proc

commit 9217bb1b4bfe07235604cd2ac94d41995b8cbd33
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 15:28:28 2012 -0700

    Revised API docs for stem.util.log

commit e4014bd0e70fc3af567f6e762d585127141279e6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 15:21:30 2012 -0700

    Revised API docs for stem.util.str_tools

commit 15ff02a5d2d654fb6afd6791d635ee2421f1b604
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 15:14:26 2012 -0700

    Revised API docs for stem.util.connection

commit a30d89064f857e355019dbd70a37997abe475699
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 14:02:09 2012 -0700

    Revised API docs for stem.util.conf
    
    The utility's header was pretty sucky so rewrote it to better exemplify how
    this module should be used.

commit 907fb84cbc11d054718ba045ac7028676f4c1746
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 10:37:21 2012 -0700

    Revised API docs for stem.util.enum

commit 8ee4c14fe0784c82637f7d3bb0b6d496c44717c9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 10:23:51 2012 -0700

    Revised API docs for stem.descriptor.router_status_entry

commit 670334b61dddec3b53ab5ea47d80149a81795e54
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 10:08:44 2012 -0700

    Line wrapping pydoc clauses
    
    I had been avoiding line wraps in :var:, :param:, and :return: entries to avoid
    confusing the spinx compiler but on reflection it's output is uneffected when
    we indent the following line.
    
    Just wrapping the lines that were really unweildy rather than going for a
    strict eighty character width. Sorry
    lines-of-code-must-be-shorter-than-my-iphone-screen zelots.

commit 29f500874cb90aa6d9c3626a609dfc957950590f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 09:42:03 2012 -0700

    Revised API docs for stem.descriptor.networkstatus

commit 934a475cb1e07126847acdbe1166abf1d48128c1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 27 08:19:06 2012 -0700

    Revised API docs for stem.descriptor.extrainfo_descriptor

commit 2c4e518adf545633e05fc97bba4718f9aebe3396
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Oct 26 09:05:00 2012 -0700

    Revised API docs for stem.descriptor.server_descriptor

commit 6925f6147120b645338f20abb8597d7b7787cf31
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 25 18:31:02 2012 -0700

    Revised API docs for stem.descriptor

commit 4b528c5e75bea5892a63ed5d63d3135d46971503
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 25 09:37:01 2012 -0700

    Revised API docs for stem.descriptor.export

commit 3c04eb6092371526126ef745903bb3e96782511a
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 25 09:17:35 2012 -0700

    Adding toctree declarations
    
    Sphinx gives a warning saying "document isn't included in any toctree" if all
    *.rst documents aren't included in a toctree somewhere. I'm not using a toctree
    as our primary form of navigation so it doesn't really make sense, but it's
    easy enough to make a contents page off to the side that satisfies sphinx.

commit 31cc49aa7606ce0f2b18a18fcdb0823a8d4448ec
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Oct 23 09:44:51 2012 -0700

    Revised API docs for stem.descriptor.reader

commit d00d2ac5a4679a428358f06242cb0980660bb100
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 21 21:26:26 2012 -0700

    Revised API docs for stem.exit_policy

commit 0cc97d2dcc15ce227983cfc72c234bf995b4857a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 21 20:04:30 2012 -0700

    Only including prepared API docs
    
    Our sphinx-apidoc made it so all our pydocs were included in our built
    documentation. This is great, except that those documentation dumps are
    completely unreadable. Dropping the sphinx-apidoc call in favor of including
    specific autodoc pages we've prepared.
    
    Made some revisions to stem.version's documentation and including that to
    start.

commit a00fb8ae0e2f7e9fe4aac4f28f87198a412feb52
Merge: fc7c5cb be5af88
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 20 16:00:03 2012 -0700

    String utility toolkit
    
    Few utility functions for string generation (mostly for making human readable
    labels of size or time metrics).
    
    cr: neena (https://trac.torproject.org/7158)

commit be5af88894666350f74a1944f9d3a37c4c827cfe
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Oct 19 13:50:33 2012 -0700

    Using get_size_label() in the tutorial
    
    Making the "Mirror Mirror on the Wall" tutorial examples a little nicer by
    using our new str_tools.get_size_label() helper.

commit e06b76b8ae68446edc3c3123412a514a00e2fba7
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Oct 19 13:33:19 2012 -0700

    Expanding str_tools unit tests
    
    Testing the common edge cases that come to mind, and making a few small fixes.
    I was also missing examples for a couple functions.

commit 70227efa08d77f65b1eea069ae93bd0ba0bf1993
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Oct 19 09:41:34 2012 -0700

    Utils for human readable time and size metrics
    
    Snagging the utilities from arm's src/util/uiTools.py for making human readable
    labels of time and size metrics. They'll not only be handy for stem's users,
    but also our exception messages and something I want to use in our tutorials.
    
    I've simplified these functions a bit and added unit tests for the pydoc
    examples. More tests are needed, though...

commit fc7c5cb0e0b06f26c7d8a1e0af348453127e3d4c
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Oct 19 14:00:17 2012 -0700

    Parsing or-address in all server descriptors
    
    As mentioned in 19ba050 the 'or-address' lines initially just applied to
    bridges, but would later be relevant to all relay types. Looks like that day
    has come since 'or-address' now appears in my cached consensus (indicated via
    integ testing failures).

commit e33f55d57b100d9f2e6f697e849811b28cd3bcfb
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Wed Oct 17 17:54:35 2012 +0530

    Fix missing paranthesis

commit 9ee778c6e723962ca8e3aa8470153a75a650d97b
Merge: f717e88 a0470a6
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Oct 16 09:05:20 2012 -0700

    Controller methods for querying descriptors
    
    Adding convenience methods for querying descriptor information (except
    microdescriptors, for which we don't yet have a parser).

commit a0470a6285c2462535f53536e7ed969725356049
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Oct 16 09:03:27 2012 -0700

    Using get_server_descriptors() in the tutorial
    
    Using the controller's new get_server_descriptors() method to exemplify an
    alternate way of doing the 'Mirror Mirror on the Wall' tutorial without reading
    the cached files directly. Both methods have advantages so I definitely want to
    exemplify each.

commit c52521efe273e8091ec95554d24ed172e755d29e
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Oct 16 08:47:03 2012 -0700

    Controller methods for fetching all descriptors
    
    Adding a couple methods to iterate over the server descriptors or router status
    entries. This definitely isn't what I want in the long run - we should iterate
    over the contents as they're being read from the socket rather than reading the
    whole thing into memory.
    
    Meh. Might as well get the functionality that we want now. The fix will be
    transparent for users later anyway.

commit 79612002788e1f9f80733974b17eefa976055d06
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 15 19:12:15 2012 -0700

    Controller methods for querying descriptor info
    
    Adding a get_server_descriptor() and get_network_status() method for querying
    server descriptors and router status entries.

commit f717e8801374939ea969ef4c73a556f15db02dab
Merge: 2a9d3c8 f996e24
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 15 09:14:57 2012 -0700

    Basic stem website
    
    Making a front page and pages for our three most important sections...
    
    * Tutorial
    * API
    * Download
    
    All three are pretty minimal, but definitely better than what we had.

commit f996e24b794c0f96e8e8221cdde5f21d8c2cf900
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 15 09:13:58 2012 -0700

    Adding stub for API page
    
    Moving the old API docs to its own page. It's presently pretty minimal and
    sucky, but at least moving it to the right spot for now.

commit 031ba36e8d67e13e0562b280ad5a89ca42c70236
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 15 09:07:01 2012 -0700

    Adding a download page
    
    Page that'll later have our download options. For now it just has git
    instructions and a 'coming soon' filler.

commit 8c767d35514ca345d051c94246668e97151d1c73
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 14 18:09:09 2012 -0700

    Moving tutorial to dedicated page
    
    Beginning of my grand scheme to make stem's front page a simple listing of its
    sections (tutorial, API, release notes, download, etc).
    
    Moved the tutorial to its own page with a summary of the tutorials listed at
    the top. The tutorial is linked from the main page by an image link.

commit eb28305879647bd628794674517da4377c127819
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 14 13:20:16 2012 -0700

    Moving tutorial tests to be unit tests
    
    Both of our integ tests for the tutorial examples were a bit clunky, the
    controller test because it could only run with newish tor versions and the
    descriptor test because it had to make use of metrics descriptors. Oh, and they
    didn't actually assert anything.
    
    Moving the tests to be unit tests instead. This included greatly expanding our
    mocking capabilities, supporting arbitrary mock objects. I'm not sure if the
    end result is any better than our prior integ tests since we're mocking pretty
    much everything (and in the case of the descriptor test it's pretty ugly). On
    the other hand the tutorial tests are primarily for basic syntax and
    compatability with our current version of stem so guess it's ok...
    
    All this said, I *really* like our arbitrary object mocking capability...

commit 11edc074fcd9f5d178899d3f4d8b6545bcd5d429
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 14 11:30:01 2012 -0700

    Replacing reversed() in tutorial
    
    Dropping the reversed() call in the tutorial in favor of providing the
    'reverse = True' argument to sorted(). Thanks to gsathya for the suggestion.

commit 2a9d3c856c5dfe4310a836318ae45069a64fde3b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 13 19:27:17 2012 -0700

    Clarifying what the tutorial's reader provides
    
    Just realized that newcomers wouldn't know that the DescriptorReader provided
    RelayDescriptor instances in the example (and hence where the methods that we
    use come from). Clarifying that detail.

commit 23513eb220ed195ee73fcadd6fb826c3581810fc
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 13 19:09:06 2012 -0700

    Adding a tutorial to our sphinx front page
    
    Tutorial for basic stem usage, with tests for the examples we provide. Our
    documentation is still incredibly beginner unfriendly, but at least this gives
    them a place to start.
    
    On a side note one of our integ tests kinda sorta killed our test instance by
    calling...
    
    controller.signal("INT")
    
    We didn't notice this because it happened in our very last test.

commit 5e29e28d16dc050d4cc5dcac9ae92d20935651d7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 13 16:34:47 2012 -0700

    Merging testrc.sample with settings.cfg
    
    We run the tests all the time but never with a configuration file so
    deemphasising the capability by merging the sample rc file with the
    settings.cfg (which is actually... well, used). If someone really wants to
    configure the tests this way then they might as well look at the settings.cfg
    anyway since it'll provide the full range of options.

commit ea356b738fff3bab4a9e468298ea7807aff37960
Merge: 5cd457b ec335d1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 13 11:33:47 2012 -0700

    Network status document handling
    
    Parsing and lots 'o test for handling network status documents. This includes
    both v2 and v3 documents (votes, consensus, and microdescriptors).
    
    https://trac.torproject.org/6569

commit ec335d1b5bf4368982bc9736a271f675a6a7c60d
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 11 19:54:05 2012 -0700

    Supporting v2 documents in the descriptor reader
    
    The v2 descriptor's integ data broke the descriptor reader tests. Turned that I
    had forgotten to add v2 documents to the types that it recognizes.

commit 372ee9836b98af582eecf3d35844397d8935bd9b
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 11 19:27:36 2012 -0700

    Parsing and tests for network status document v2
    
    Parser, unit, and integ test for version 2 network status documents. These
    documents are deprecated and no longer generated, however we still need a
    parser to read older consensuses.
    
    Unlike the v3 parser I'm cutting a few corners...
    
    - not validating parameter ordering
    - no validation that header/footer parameters haven't swapped places
    - only the bare minimum unit test, no tests for invalid content
    
    We can remedy these if necessary but with the growing irrelevance of v2
    consensus parsing I doubt we ever will. Plenty of more important things to do.

commit 0793113b37afa737862440363983105e820640bd
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 11 07:53:41 2012 -0700

    Renaming NetworkStatusDocument to NetworkStatusDocumentV3
    
    Changing the name to make room for v2 documents and future versions.

commit 925e8a3e0232f734c1067e88ac8898e446dee35e
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Oct 10 19:15:29 2012 -0700

    Reordering document's integ test attribute checks
    
    Changing the order of the attribute assertions to match the classes, and adding
    a few checks that were missing.

commit 9170c0ff411bca266bd90af4bf423c5c0decb44e
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Oct 10 09:35:59 2012 -0700

    Replacing _strptime() helper with normal datetimes
    
    I kinda liked the idea of the _strptime() helper and briefly moved it to the
    runner so I could use it in other tests, but on reflection it doesn't really
    help readability nor is it appreciably shorter. Swapping its usage to normal
    datetimes for uniformity.

commit b0d24bee1d2a00f20aba8953a15df1e006860b33
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Oct 10 09:15:13 2012 -0700

    Refactoring network status integ tests
    
    General refactoring for the network integ tests, mostly moving things around
    and renaming to make things more readable. The only functional difference is
    that a couple tests were using the 'assert' keyword rather than testing
    assertions. This was a bug - it meant that we'd skip those tests unless running
    in assert mode and even then it would cause exceptions rather than testing
    failures.

commit c20cfdcd6d87a04d23c508b8b333086c77000d77
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Oct 9 09:15:41 2012 -0700

    Addressing issues spotted by integ tests
    
    Enough of this unit testing, time to run our new parser against actual network
    status content. Unsurprisingly this ran into a couple issues...
    
    - Microdescriptors have an extra field on their 'directory-signature' lines.
      This is undocumented so it'll also need a spec fix...
    
      https://trac.torproject.org/7072
    
    - Our parser for 'directory-signature' was only reading the first one, rather
      than iterating over all entries.
    
    Most of the rest of the changes are just revising the integ tests that Ravi
    wrote to accomidate changes I've made to the classes.

commit 78e09a971b7c242dc1d0a6b95718b78edd5cb32b
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Oct 9 07:56:33 2012 -0700

    Adding tests for the network status document examples
    
    Including unit tests for the header pydoc examples to check that they're
    runnable. I also changed the first example to show that the consensus file
    doesn't need to remain open when using that method.

commit 3ddd3c55da9b859dcce218283ee91aab251bf230
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 8 09:27:57 2012 -0700

    Rewriting network status module pydocs
    
    Revising the header documentation for the network status document module. The
    bit about v2 support is presently a lie, and I need to include the examples in
    the tests. Otherwise, this module is almost done!

commit 27356d41908c70fdec23d26d4f9933f8fefb71f9
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 8 08:50:13 2012 -0700

    Simplifying _get_descriptor_components() usage
    
    The _get_descriptor_components() provided the first and last keywords since
    those are often needed for validation. However, this is pointless now that
    we're using an ordered dictionary (we can simply check the key listing).

commit 98d8e4685bab0dd107f5dd6f185492d6f3a5c136
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 8 08:36:47 2012 -0700

    Dropping test for empty dir-source fields
    
    Nick replied on 'https://trac.torproject.org/7055' saying that the fields can't
    be empty strings, so dropping the test. Also adding a validation check that the
    hostname isn't an empty string.

commit 60868f5f161dc0db59f4c19019b75c29f417b37c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 7 18:53:18 2012 -0700

    Dropping unused descriptor helpers
    
    Removing the helper functions for the prior parsers of network status
    documents.

commit 4216b5f1d5762d229945306508ea078c9fd1902c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 7 18:47:27 2012 -0700

    Supporting microdescriptor flavored consensuses
    
    Adding support for microdescriptor flavored consensuses into the
    NetworkStatusDocument class. It made sense to have a separate class for it, but
    on the other hand it *is* still a v3 consensus and the only impact the flavor
    has is alternate router status entries so just blending a 'flavor' and
    'is_microdescriptor' attribute in.

commit a1d2a0ffffd5aeec96c37fbbf1b4134c9b34b631
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 7 18:18:14 2012 -0700

    Unit tests for minimal v2 and microdescriptor v3 router status entries
    
    Tests to exercise basic parsing for the v2 and microdescriptor v3
    RouterStatusEntry subclasses. They're both largely a subset of the v3
    RouterStatusEntry so I don't plan to add additional tests for the moment.

commit b8ca825a7f6dcde9f7689bca41a0938598749b4f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 7 17:59:08 2012 -0700

    Module for V2, V3 and Microdescriptor router status entries
    
    Moving the router status entries to their own module and adding classes for V2
    entries and microdescriptors (both still completely untested). The abstraction
    is based on the ServerDescriptor breakdown but with more liberal use of helpers
    for the actual parsing.

commit 72cb24dde90c928135a686707db6baa5a88d5147
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 6 14:09:34 2012 -0700

    Unit test to include directory authorities in document
    
    Adding a document test that includes authority entries. The especially
    interesting bit of this is checking that validation propagates to the
    KeyCertificate constructor (ie, it's checking that the 'validate' flag is being
    passed from the document to authority and authority to cert classes).

commit dc89b293bb3f8a282766971597ada90723f17fa7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 6 13:51:29 2012 -0700

    Unit tests for DirectoryAuthority fields
    
    Tests for the DirectoryAuthority's individual fields, and fixes for a couple
    issues they uncovered.

commit 5ac628e7c8bd95c6c2cade6477aceef51a941b2e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 6 12:26:10 2012 -0700

    Adding unit test for empty dir-source values
    
    Just realized that the nickname and hostname fields, which lack any
    constraints, might also allow for empty values. I'm not really clear about this
    so asking for clarification...
    
    https://trac.torproject.org/7055

commit 347d4c03a63b8a3f81bee83ee990e5ea4ada9125
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Oct 5 08:53:17 2012 -0700

    Boilerplate descriptor tests for DirectoryAuthority
    
    Copying a few general tests from the other descriptors. I should move these
    into a descriptor test superclass so I don't need to keep copying them. That
    said, there are some minor tweaking that might prevent that...

commit a09b1e038fc97dada370e01ad8fd0c4f81b90c12
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 4 09:39:27 2012 -0700

    Fix when router status entries have duplicate lines
    
    Spotted a bug with the RouterStatusEntry where we have an undefined 'key'
    variable when validating that we don't have any duplicate entries. Turns out
    that this use case wasn't being tested so added a unit test to exercise the bug
    first.

commit bf192cb70a770fe2551add747bbdb6aed2f94d9d
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 4 09:04:53 2012 -0700

    Parsing for DirectoryAuthority
    
    Rewriting the parser for the DirectoryAuthority class. Still passes the minimal
    descriptor tests so next gonna add the field specific unit tests.

commit 11070d9a91e0c00b4b5854fc935c5fc7a871e97f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 29 12:41:24 2012 -0700

    Unit tests for minimal DirectoryAuthority instances
    
    Adding unit tests for the minimal vote and consensus directory authority entry.
    Presently this is just exercising the old parsing code (which I broke in a few
    places during some of my earlier refactoring).

commit 17ce11ac3822ec811fc9b25ed28f24391ae50137
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 29 11:19:59 2012 -0700

    Dropping _read_keyword_line_str() and line_matches_keyword()
    
    Dropping a couple of the descriptor helper functions that are no longer used.
    There's a couple more that I'd like to get rid of but they're not yet fully
    replaced.

commit e52f0ec96b01db7da8002479b9a864a6f18f6db8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 29 11:17:09 2012 -0700

    Dropping the old KeyCertificate class
    
    We have a new shiny and tested KeyCertificate class so dropping the old one.

commit 7986afa40937657391346dea5b4aa6aeb561c870
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 29 11:13:34 2012 -0700

    Unit tests for KeyCertificate fields
    
    Remaining unit tests that I'm planning to do for the KeyCertificate. Oddly this
    is the first time that I've added a unit test for malformed content in a key
    block field (oops).

commit 92b691c5e9ace6803a5ed24d42e7ec41665595c8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 29 10:19:16 2012 -0700

    Boilerplate KeyCertificate unit tests
    
    General unit tests that I've been including with most descriptor types. I
    should probably include these in a more systematic fasion...

commit 50cc258286cd31da9571f8010f5f24b2e56bf386
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Sep 28 09:32:21 2012 -0700

    Dropping a coupel unused imports
    
    Ran pylint over the recently refactored files. It caught a couple of unused
    imports.

commit 8f89166499dafcd155022e07bbf39eafd7cccf4e
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Sep 28 09:05:35 2012 -0700

    Moving network status document mocking into util
    
    The network status document has a few oddities compared to the other descriptor
    types. It includes...
    
    - ordering constraints outside of the mandatory fields
    - keyword-only lines
    - router status entries
    - different mandatory fields based on if it's a vote or consensus
    
    Imho this new version handles these a lot more elegantly than the prior helper
    that was embedded in the network status document unit tests.

commit 6f5d91666562e562fd773636ffcd292abff0d939
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Sep 28 08:03:56 2012 -0700

    Moving key certificate mocking into util

commit d1ed7c2f9be307d8968ba56fd1906fcd44c1ab25
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Sep 27 09:40:06 2012 -0700

    Moving router status entry mocking into util

commit 3923916993535c87d95c223bbf9fd0479fc0dc68
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Sep 27 09:10:27 2012 -0700

    Moving extrainfo descriptor mocking into util
    
    The extrainfo descriptor unit tests had its own helper function for getting
    mock descriptor content. This is the same thing as what the server descriptor
    unit tests use so moving it to the mocking module.

commit a5e8d26381fc9787a8cbc147497e4fa397d70fc5
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Sep 27 08:22:12 2012 -0700

    Generalizing mocking.get_server_descriptor()
    
    All of the descriptor unit tests (server, extrainfo, and document components)
    need a similar type of mock data. I've been doing this in an adhoc fashion but
    this can stand for a bit of uniformity. Starting off by cleaning up the server
    descriptor mocking.
    
    As per the convention for the mocking module, making it provide a mock object
    by default with an option to get the raw content.

commit 5d1a11fa4f479cac4d4987d9841889ea251b2eba
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Sep 26 09:00:22 2012 -0700

    Unit test for parsing a minimal key certificate
    
    Test for minimal key certificate parsing, and related fixes for the
    KeyCertificate class. The main gotcha in this is that KeyCertificates don't
    need to have a prescribed order (unlike other network status document
    fields).

commit 6ee5abbe5c1137b4a311d5b7ef5003874982008e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 23 14:20:49 2012 -0700

    Initial KeyCertificate implementation
    
    First stab at parsing the authority section's key certificates. This is
    completely untested, next step is to write some unit tests for it.

commit 2e65e58568469f460fedf965164b1355fae61204
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 22 19:01:16 2012 -0700

    Begin revising the DirectoryAuthority
    
    God damnit. I thought I was almost done but, on reading the spec some more, the
    key certs and authority entries deserve their own descriptor type just like the
    router status entries. Does this document have no end?
    
    Saving the top-down approach I was taking but putting it on ice for now. I
    should work at this from a bottom-up fashion instead, starting with the key
    certs and starting a new batch of tests for it. Fun fun. :(

commit 7a13a14c337ecf30162e82d75c904910af884b11
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 22 18:06:20 2012 -0700

    Unit test for misordered content
    
    I added a test a while back to check that we detect misorderd document
    attributes, but it didn't pass at the time so I skipped it. The document
    parsing rewrite is almost done so enabling the test.

commit 72c561b9adb7e3963aa14da98b8d9bc77df533dd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 22 17:57:49 2012 -0700

    Unit tests for parse_file() and router entries
    
    On reflection the reason that the prior changes passed the unit tests so easily
    was because the parse_file() function and inclusion of router status entries
    was completely untested by my unit tests. The RouterStatusEntry class itself it
    tested, but not its inclusion in a document. Integ tests would certainly cover
    this, but I want the unit tests to exercise everything too.
    
    Adding the missing tests and some fixes for issues that they revealed.

commit a5babce203784eb45f7c585edbcee4caf2362212
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 22 14:12:21 2012 -0700

    Generalizing how router entries and authorities are parsed
    
    The _get_routers() and _get_authorities() were essentially doing the same
    thing. Replacing both with a more general _get_entries() helper that reads a
    range of the document and constructs instances for it.
    
    Taking advantage of this nicer helper's keyword arguments to make the code more
    readable (functions that take a ton of positional args are is a pita).
    
    I'm a bit surprised (and concerned) at how easily this passed unit tests. No
    doubt I've broken the integ tests but I'm not putting any effort there until
    I've finished the document parser rewrite.

commit e4185194801b6afd10671e6e7a1a53977a598c29
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 22 13:18:16 2012 -0700

    Getting rid of the _get_document_content() helper
    
    Too many helper functions and the code becomes fragmented. The
    _get_document_content() was only used a couple places, and both of those were
    actually better with their own slightly different implementations.

commit 71240065e2a89ab1c8eeb9f865fddc294742cb8a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 22 12:46:03 2012 -0700

    Refactoring how authority information is fetched
    
    Authorities are another section, like the header, entries, and footer so
    fetching it from the _get_document_content() helper. I'm not really happy with
    this code yet, but it's a step in the right direction and lets us finally get
    rid of _parse_old().

commit 641bed527c019a92e47dd769bca23fa04411d9c3
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Sep 21 09:00:19 2012 -0700

    Breaking up the header and footer from NetworkStatusDocument
    
    The NetworkStatusDocument class was starting to get monsterous, and it was just
    gonna get worse. A network status document consists of four sections...
    
    - header
    - authorities
    - router status entries
    - footer
    
    Making the NetworkStatusDocument a thin container for these four, and making
    separate classes for them. This has made the code much nicer.
    
    The only disadvantage that I've seen is that validation is done in pieces so
    if, for instance, you're missing mandatory fields from both the header and
    footer you now won't be told about both in a single error message. Instead
    the header will be parsed first, fail, and just tell you about those.
    
    That said, this is a pretty minor regression and well worth the improved
    maintainability.

commit 1f868090e2d641ddcb49d02bd15b5894f5bf6923
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Sep 18 09:36:56 2012 -0700

    Parsing the directory-signature and unrecognized lines
    
    Finishing up with the footer. It doesn't make sense for the DirectorySignature
    or DirectoryAuthority to be Descriptor subclasses (cuz... well, they aren't
    descriptors). However, I like having this struct class rather than providing
    our callers with a tuple list. I should probably do this for other descriptor
    documents too...

commit ea3102387729daa17587f12b47f249ff49baf1bd
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Sep 18 08:03:12 2012 -0700

    Adding bwweightscale parameter constraints
    
    My spec fix to clarify the parameter was merged [1] so enforcing the default
    and constraints in our parser.
    
    [1] https://trac.torproject.org/6872

commit 802f96e94c81dc7270d1c8311485e3cd1bf2e10d
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Sep 17 08:55:53 2012 -0700

    Parsing the bandwidth-wights attribute
    
    The 'bandwidth-wights' line is pretty similar to the 'params', so sharing most
    of the parsing code between them. Testing for the following...
    
    - negative and zero values
    - malformed entries
    - ordering
    - that this can't appear in a vote
    - missing values or empty content

commit 53f16d8546302621f786c30ef0fb1b69786444fe
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 16 18:10:08 2012 -0700

    Noting that we might want to add 'bandwidth-weights' later
    
    The "bandwidth-weights" field mentions a 'bandwidth-weights' parameter and a
    default value. This isn't listed in the params section so getting confirmation
    of what I should be making of this param first.

commit b3444ace427dca9a66a1afed1d24a09971612810
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 16 17:43:46 2012 -0700

    Rejecting all footer fields in outdated consensus-method
    
    We were only checking that we lacked the mandatory footer fields when our
    consensus-method indicated that we shouldn't have a footer. Instead checking
    that we have no footer at all. This also makes the code a little nicer...

commit 5c4a3ec4cb22fe0fff6c44f3eecf2a2639788ed6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 16 17:07:03 2012 -0700

    Parsing the directory-footer attribute
    
    For being an empty attribute this sure is a strangely big change. Checking the
    following...
    
    - that footers don't appear prior to consensus-method 9
    - that the directory-footer lacks any content
    - that prior to consensus-method 9 we're happy to not have the line (bug I
      introduced because the footer has mandatory fields)

commit 84fe0b46adc684a663af8536da35648adfd841e2
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Sep 14 09:35:03 2012 -0700

    Tidying up NetworkStatusDocument._parse() start
    
    Much of the _parse() method's start was unnecessary. Shortening it to improve
    readability.

commit f60e60006a54ba2fc1eba0cb2fa5fade55b670ff
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Sep 12 09:14:30 2012 -0700

    Validating params values and including defaults
    
    The 'params' line of a network status document has several known entries, with
    their own constraints on the value. Validating that the document obeys those
    constraints.
    
    Also, the path-spec has default values for a handfull of params so optionally
    defaulting our params attribute to that.

commit 355e474b3d62dceb09699f1be27b5b27b925e7fa
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Sep 10 09:12:19 2012 -0700

    Parsing the params attribute
    
    Being a fair bit more anal about field validation, checking the following...
    
    * bounds on the value to be an 32 bit int
    * order of the values (keys should be ascending)
    * reject values like '+11' (the int() function accepts them)
    
    This also includes unit tests for these use cases and better error messaging.

commit 634c8f1d32dab7717b518593225b43b9e0349285
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 9 16:44:37 2012 -0700

    Basic validation attribute testing
    
    The network status document's unit tests were mostly just checking invalid
    inputs. Changing the attribute unit tests to test some valid input too.

commit 7218845d7bfce16b9ae4a0812cc74cd54b07f7c2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 9 16:02:58 2012 -0700

    Defaulting to consensus-method 1
    
    As per the spec clarification in commit b829224, when the consensus-method or
    consensus-methods fields are missing we should default to method 1 (since in
    practice that's the only time the field doesn't appear).

commit 0e4f8bfbdcf1bc9e427748dd37b53ad6b4a32d93
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 9 15:13:32 2012 -0700

    Parsing the known-flags attribute
    
    Only changes are...
    
    * Moving the Flag enum to stem.descriptor.Flag since it's something that users
      will commonly use. As a rule of thumb I'd like users to only need
      'stem.descriptor' unless they're doing something fancy. However, this is just
      a temporary move - I plan to move Flag to is own module, like Version.
    
    * Excluding empty strings from our known_flags attribute (our prior behavior
      was to parse things like "   " into ['', '', '', '']). The spec doesn't set
      any constraints on what a flag can be so I suppose technically one could be
      the empty string, though this'll never be the case. Still waffling back and
      forth about if this should cause a validation error or not...

commit fd0deb96f30fc691019e0d2a09068a4787d80e97
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 8 19:11:13 2012 -0700

    Parsing tor version list attributes
    
    Handling the 'client-versions' and 'server-versions' lines, comma separated
    lists of tor versions.

commit fb808fde025a8c4f7f04d243f6e679f2f8b5e255
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 8 18:52:58 2012 -0700

    Parsing the voting-delay attribute
    
    Nothing special here. Simply a pair of integer fields.

commit cb96c3e094719b096e5425faeadc01a15dfcb388
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 8 18:42:29 2012 -0700

    Parsing consensus-method and datetime attributes
    
    Handling the published, valid-after, fresh-until, and valid-until fields. Doing
    them as a batch since they're all simply datetime fields without any other
    constraints.

commit 65e92e80711999591441abbdf23a21321e4d9271
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 8 18:18:38 2012 -0700

    Parsing consensus-methods attribute
    
    Only noteworthy bit is that the spec says 'Method "1" MUST be included.' so
    checking for that.

commit b931e980b62600d5cc9cd60069102c2ffa26cb84
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 8 12:35:01 2012 -0700

    Parsing vote-status attribute
    
    Changing our 'vote_status' string attribute to 'is_vote' and 'is_consensus'
    boolean attributes. The spec specifically says that anything else is invalid so
    there's little reason to allow arbitrary content in the field.

commit 5e12da0a50ee60577f3a571eea0ef3350e146d10
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 8 12:15:34 2012 -0700

    Parsing network-status-version attribute
    
    I don't like our assumption that this is a v3 network status document, but I'll
    need to think more about how to handle other versions later. For now keeping
    the validation assertion that parsed documents are v3.

commit db87c88965806ec13f9888c220879af45fc65e44
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 8 11:54:33 2012 -0700

    Validating that fields aren't duplicated
    
    Almost all of the fields in a network status document can only appear once.
    Checking that this is a case. This also includes a unit test for the
    validation. I'm a little surprised that the old parser doesn't choke on
    duplicated content, but I'm about to replace it so probably not worth
    investigating.

commit f23c5312b671dd2bd41baa5a0c24568592ac3241
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 8 11:23:08 2012 -0700

    Validating the order of fields in documents
    
    As mentioned in the prior commit, to be valid network status fields need to
    appear in a particular order. Checking for this as part of document validation.

commit bc09f1e697f4a34475b4b9b55bcb91c14494427b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 8 10:51:51 2012 -0700

    Unit test for misordered fields
    
    To be valid a network status document's fiends need to appear in a particular
    order. I'm about to add vaidation for this, but the parser is being finicky.
    I'm tired of trying to fix incrementally fix it, so skipping this test for now
    and moving on to the parser rewrite.

commit 725e2f9deb1a22b5cfca0242ea79d5c0548a40e3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Sep 8 10:26:48 2012 -0700

    Unit tests for minimal vote and missing fields
    
    Unit tests for a couple important use cases and lots 'o fixes for the issues
    they uncovered. As mentioned earlier the 'validate' attribute took the wrong
    meaning in this parser so valid content errors and invalid content triggers
    stacktraces.

commit 7b38323ee41aa890f5b777eff4bd365d5a6c1fb1
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Sep 6 09:29:25 2012 -0700

    Splitting up entry and document unit tests
    
    Splitting up unit tests for the RouterStatusEntry and NetworkStatusDocuemnt.
    Both will be pretty lengthy and there's little reason to try to mash them
    together.

commit cf475d58dfa4d42e982eca6307e2a61e7545147e
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Sep 6 09:14:09 2012 -0700

    Minimal unit test for network status documents
    
    Adding a unit test for the minimal valid network status document (plus a
    consensus-method field since that influences validation). This uncovered
    some bugs with the NetworkStatusDocument class...
    
    * The network_status_version field misdocumented as being an int (it was
      actually a str). We need it to be a str for microdescriptors so simply
      changed the pydoc.
    
    * The consensus-method and bandwidth-weights are documented in the spec as
      being optional fields. The parser errored with a stacktrace when
      consensus-method was missing, and gave a validation error if there isn't a
      bandwidth-weights.
    
    * Inappropriate validation error if there was unrecognized content.
    
    * The get_unrecognized_lines() method is documented as providing a list of
      lines. The NetworkStatusDocument returned a string instead.
    
    * Off-by-one error that caused consensus-method 9 documents to skip parsing
      footers.

commit d30a628e5da06699d8533f92c4b1c496510213b8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Sep 2 17:06:20 2012 -0700

    Checking for mandatory and disallowed fields
    
    There's several restrictions on a valid network status document, some of which
    are which fields it does and does not contain. Validating that manitory fields
    apprear, and that fields which shouldn't appear don't.
    
    This also drops my _get_entries() helper function in favor of
    _get_descriptor_components() since we're now dealing with documents which can
    contain key blocks. There wasn't really enough of an advantage to
    _get_entries() to justify the duplicate code anyway.
    
    Also fixing the _get_descriptor_components() I wrote so it return footer
    content in the header when the content doens't contain any 'r' lines.
    
    This currently keeps the old parser (renamed to "_parse_old()") since I haven't
    addressed any of the actual parsing yet.

commit 239d9642bfc800b4f720880f359cdc92a713e63f
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Aug 21 16:52:56 2012 -0700

    RouterStatusEntry unit tests
    
    Unit tests for the RouterStatusEntry use cases that come to mind. As normal
    they uncovered some bugs with the class.

commit dc15e682be37ea450a9b5e45f05958f676eabded
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Aug 20 22:58:11 2012 -0700

    Setting zero ports to None in server descriptors
    
    When writing the server descriptor parser I faithfully translated zero ports as
    zero because I wanted users to be able to differentiate between 'malformed' and
    'zero' when validation was disabled. However, on reflection I don't really care
    about catoring to non-validation use cases.
    
    Ravi made the router status entries interpret a zero port as being None which
    makes perfect sense - zero means that the port doesn't exist. Doing this for
    the server descriptors too.

commit be0c484de0be3cfa98a32c9ef5210b3e1d6eb6ee
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Aug 20 22:53:17 2012 -0700

    Unit test for minimal RouterStatusEntry
    
    As usual, starting the class unit testing by parsing a minimal example. Next
    gonna go through field by field to exercise interesting use cases.

commit a33297c8f65552c1ef721277e056a89f2b2727b7
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Aug 20 22:20:44 2012 -0700

    Rewriting RouterStatusEntry parser
    
    The networkstatus module has spotty error checking, for instance...
    
    * the unrecognized_lines attribute is never initialized, so anything touching
      it will trigger a NameError
    
    * missing values on an 'r' line will result in an IndexError, and extra values
      are ignored
    
    * malformed nicknames, ip addresses, and ports go undetected
    
    * non-integer ports or bandwidth/measured values give an 'invalid literal'
      error, which isn't terribly helpful
    
    * extra entries on a 'w' line inappropriately cause an error
    
    * malformed exit policies can cause an error when validation is disabled
    
    * we error if there's new keywords (the spec doesn't seem to say they're
      disallowed...)
    
    From the look of this code the 'validate' flag seems to have been
    misinterpreted as meaning "raise an exception if we run into something
    unexpected, like a new keyword or flag". This is wrong. Stem should *never*
    error when processing spec conformant content. Rather, the 'validate' flag
    means "raise an exception when we parse something that violates the spec". When
    that flag is False the parser should *never* raise an exception, making a
    best-effort attempt to parse even malformed content.
    
    I'm a bit sad to see that there are absolutely no unit tests for this module.
    That was the bulk of the testing for the server_descriptor and
    extrainfo_descriptor modules. Guess I'll write that next.

commit 0752e699fcfb657eb27ed97e0b188376cd9368a1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 19 21:47:06 2012 -0700

    Dropping unknown_flags attribute
    
    I suggested an unknown_flags attribute that would have flags which stem didn't
    recognize but, on reflection, this wasn't a very good idea. Splitting the
    attributes mean that users have multiple places to look for the relay's flags.
    
    We were also failing validation if there were any unknown flags and that is
    definitely wrong. New flags are perfectly valid and will happen in the future -
    those new flags should not break stem's parser. This was probably confusion
    from when I asked for the cached-consensus test to check for new flags. This is
    something separate and added it with this change.

commit b1a1674cd5df2b5443699738b6880c6a0077cf0f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 19 21:31:49 2012 -0700

    Dropping known_flags arg from RouterStatusEntry
    
    The RouterStatusEntry accepts a known_flags argument and I'm not sure why. It's
    not used, and the known/unknown flag distinction is simply based on if stem
    knows about it (and hence is in the Flag enum) so I'm not sure how or why users
    would utilize it.

commit dd071270d705507f0131bfceac7358f7223ec32d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 19 21:25:59 2012 -0700

    Asserting that _decode_fingerprint() output is a fingerprint
    
    Checking that the _decode_fingerprint() function returns a fingerprint,
    throwing a ValueError if it doesn't. Also adding some tests for this.

commit 2b4ec7c1c6e1e1848ca3080d9da41cb02d51b0ea
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 19 18:11:49 2012 -0700

    Error due to KEYWORD_LINE regex change
    
    The KEYWORD_LINE regex fix a few commits back broke one of the server
    descriptor unit tests because it caused lines with only a keyword (ex.
    "platform") to have a None value. Before the regex change they had an empty
    string value.
    
    I'm not really sure which is proper, but for now keeping the prior behavior.

commit fd58bc5a565bca11d372188a4add167775f091c1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 19 17:55:25 2012 -0700

    Replacing RouterStatusEntry's identity with fingerprint
    
    Truncated base64 encoding of a relay's fingerprint? Not useful. Common hex
    encoded fingerprint that's used everywhere else? Very useful.
    
    Decoding the RouterStatusEntry's identity value and adding a unit test for it.
    Unfortunately the networkstatus module didn't have any unit tests at all so
    this is gonna take a while to remedy...

commit 393c4de2817a27c1cbdaa7bdf881e9cd08e61a76
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 19 16:35:42 2012 -0700

    Renaming RouterDescriptor to RouterStatusEntry
    
    The dir-spec calls these "router status entries", so using that name in stem
    too.

commit f9a2ef8d0dd60a73c7cfe5a329b20329c4b4218c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 19 11:54:29 2012 -0700

    Hybrid approach between iterator and attribute
    
    Ok, changed my mind. At least partly. If the user is using parse_file() then we
    want to provide them with router entries and a bare document, like I was.
    However, if they're calling the document constructors then they've already read
    the complete document into memory so there's little reason not to parse it.
    
    In short, documents now have a 'routers' attribute. That attribute is filled
    with the parsed routers if we use the constructor directly, and left empty if
    we used parse_file().

commit 35c5eaf2933dad64bab61cc65657d60db276a54a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 19 00:21:08 2012 -0700

    Inverting network status' document and router relationship
    
    This code was trying very hard to produce a network status document that
    contained router entries. This is right and proper from an object oriented
    standpoint, but not from a practical one. The document cannot contain the
    router entries unless we keep them all in memory, and for the vast majority of
    users that is not desirable.
    
    In later revisions Ravi addressed the memory concerns by turning the
    router_descriptors attribute into an iterator. This works, but it's confusing
    and the router_descriptors' itertor only works as long as we keep the file open
    (or buffer the content).
    
    Instead, inverting the relationship so this module provides an iterator of
    routers and those routers have a reference to the document they came from. As a
    side benefit this lets us tidy up the code a bit.
    
    There like are some use cases where we would prefer a document object that
    contains routers, so I'll probably add that option too later.
    
    This also gets rid of the "Flavour" enum. I'm not entirely sure what it was for
    - it wasn't documented and its NS value was unused. Maybe it was vestigial from
    some prior changes.

commit e1ce38ebd4e49faff993e212e6959f9a46e93e50
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 18 19:08:15 2012 -0700

    Removing _skip_until_keywords()
    
    The _skip_until_keywords() is almost entirely a duplicate of
    _read_until_keywords(), its only difference being that it doesn't return the
    content. Adding an argument to _read_until_keywords() to do the same.

commit 756625453f5ac0ff9415d4cb04bc373c70bf79ef
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 18 18:53:07 2012 -0700

    Keyword parsing broken for tab dividers
    
    Most descriptor types (server/extrainfo descriptors, V2 network status
    documents, etc) have their divider whitespace defined as one or more spaces or
    tabs. God knows why they chose that - there's no good reason for outputing that
    and it makes parsing a pita. An exception is V3 network status documents, which
    redefine SP as being a single space.
    
    I've never seen this in the wild so this is mostly an academic concern, but I'm
    trying very, very hard to make stem spec conformant.
    
    The _read_until_keywords() and _skip_until_keywords() functions only worked if
    the keyword was divided by a space, not a tab.
    
    I sunk two hours into fixing a bug with the KEYWORD_LINE regex which caused it
    to match against signature content. Long story short, we were accidently
    fuzzing our parser by letting there be a 'keyword' and 'value' when there was
    no whitespace between the two.

commit 7927579282e6b6a3071dc1901c0a725a94fa7563
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 18 16:45:43 2012 -0700

    Iterating over routers with parse_file()
    
    Having parse_file() iterate over the RouterDescriptors in a consensus rather
    than providing the document itself. 99.9% of the time our callers want to do
    something with the routers, and providing them a document instead just forces
    them to unwrap it.
    
    I'm thinking about changing the networkstatus module to provide
    RouterDescriptors that reference the NetworkStatusDocument they came from
    rather than the other way around. We'll see how it works.

commit 21e5a4cd578dc81496d8355e377048c3707488b0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 18 16:30:03 2012 -0700

    Combining TestNetworkStatusDocument and TestMicrodescriptorConsensus
    
    Having separate test classes was a fine idea, but resulted in our tests
    claiming that 'test.integ.descriptor.networkstatus' ran twice. That's
    confusing, so we either need to put them in separate files or merge the
    classes. Opting for the later for now since TestMicrodescriptorConsensus only
    had one test.

commit a85e6f77ee1b4ba40e86bd148062bf8e479484e1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 18 16:12:35 2012 -0700

    Raising TypeError when metrics header is unrecognized
    
    When we get a metrics header for a format we don't recognize the
    _parse_metrics_file() function is supposed to throw a TypeError (and says it
    does in a comment). However, it didn't.
    
    This caused unrecognized metrics types to be parsed like a valid descriptor
    file containing zero descriptors. Fixing and adding a test for this use case.
    
    Also fixing a couple other minor bugs I ran into while fixing this:
    
    * Some tests had an assertTrue() rather than assertEqual(), causing the
      assertions to almost always evaluate to True.
    
    * The UnrecognizedType we threw for text files didn't have its mime type set.

commit f1ecbde2b5f00019ec171c44628c03c2f09e1952
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Aug 16 18:18:09 2012 +0530

    stop skipping unexpected "opt" lines

commit 6c3717b65acc9d208ef3bf90b5b54f3983e507df
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Aug 15 20:47:29 2012 +0530

    Fixes to document parsing
    
    One major change is that stem.descriptor.networkstatus.parse_file now
    returns a NetworkStatusDocument object instead of iterating over the
    router descriptors in the document

commit 8ad310114b1ea7b743a868a8b70832eea5b8f3e2
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Fri Aug 10 17:49:23 2012 +0530

    Add microdescriptor parsing

commit dcf2092607c8be29d065def1cb104290709bef87
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Aug 9 17:21:09 2012 +0530

    RouterDescriptor objects store a list of flags

commit 9ad3395400cd21a26c0106e665aa6e1e2026a353
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Aug 9 13:01:44 2012 +0530

    Removed DescriptorParser

commit eb0e424ed9459188b70f33ff401e23e9fd89138b
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Aug 8 12:39:55 2012 +0530

    Implement lazy router descriptor reading

commit 1b26a46ed3c2b26293474d9349efb247f5888a65
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Aug 8 08:33:48 2012 +0530

    Extend _read_until_keyword to read until multiple keywords

commit 5f61f4fe91014dbe44e5acc8facd69e0c3467797
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 21:39:57 2012 +0530

    Remove unnecessary imports

commit cba98e26f16214925f7f5e6d61b51767da32b66e
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 21:24:30 2012 +0530

    Fix incorrect stem.descriptor.networkstatus import

commit e7cc83b4e382d209dda4660b5d30b9691f3fd713
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 13:48:24 2012 +0530

    More documentation

commit 9e6fb6d9d99144832b78e5cf1b28fc4232d17c15
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 13:25:43 2012 +0530

    Fix incorrect docstring and reduce 1 LoC elsewhere

commit 92f46907b3d84c53ab10c8f19851d65be86f7b9f
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 13:19:57 2012 +0530

    stem.descriptor.networkstatus.KeyCertificate => stem.descriptor.KeyCertificate
    
    From section 4.7 the dir-spec,
    
    A concatenated set of all the current key certificates should be
    available at: http://<hostname>/tor/keys/all.z
    
    The key certificate for this server (if it is an authority) should be
    available at: http://<hostname>/tor/keys/authority.z
    
    ...
    
    we might want to parse these seperately at some point, so, moving it out
    of stem.descriptor.networkstatus

commit a288e1e43f2cecc55b1b46cbfab6734aee1a5395
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 13:10:49 2012 +0530

    Update comments

commit 9df15326fecd223bf7d4e5a40465a2fa80069025
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 15:01:49 2012 +0530

    Remove trivial comment

commit 847d1caec38e110f7d75ae1db526a85803094d13
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 14:43:42 2012 +0530

    Fix cached-consensus parsing

commit 21cfed42b859becabe71a803164004fea2372929
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 14:14:51 2012 +0530

    Add support for votes from Metrics

commit 4e29ae8dfc69cb68a25c9cf159e56db4e943565b
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 12:09:25 2012 +0530

    stem.descriptor.networkstatus_descriptor => stem.descriptor.networkstatus

commit 1ed53535e19b1c065dfc9c4589c31283cdac7649
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 12:03:05 2012 +0530

    Initialize instance variables

commit f955d14f25041846cdc82a8e59fe704378a6f6ac
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 12:02:03 2012 +0530

    Fix version parsing in NetworkStatusDocument

commit cc0c897908bb0d1bd97672675badb1286d71bfbc
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 11:58:33 2012 +0530

    Fix DescriptorParser.remaining

commit 789145ee6ca6dea2617cdd641771cbce1b50296a
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Aug 7 10:16:43 2012 +0530

    Add tests for vote and consensus documents

commit 4d9a7a38040add8a3e48254ddffd9f32189ccb26
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Aug 6 14:08:26 2012 +0530

    Add integ tests for networkstatus_descriptor

commit 4a58b7a9fb30584f70ade4a6d75bc8f9821e57d0
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Aug 6 14:07:25 2012 +0530

    networkstatus_descriptor.parse_file iterates over RouterDescriptors in a document

commit 1cf9a62f46d9b9c4ffcbae90dbb59abaa66ff3c8
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Aug 6 14:05:29 2012 +0530

    Fix DescriptorParser and make parse_file iterate over router descriptors

commit fc1ad3f21f8d23233551514dc87524a40697621d
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Aug 6 05:28:41 2012 +0530

    match Measured instead of Measured=

commit fb710843f6f21ec1779d80462a1a943594623281
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Aug 6 05:16:56 2012 +0530

    Fix documentation formatting
    
    stem.descriptor.server_descriptor's module overview is formatted
    incorrectly. (see stem.socket)

commit 8ed384d31aa446dd3c5954dd4353e059657ed7a3
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Aug 6 05:14:51 2012 +0530

    Move DocumentParser to stem.descriptor.DescriptorParser

commit a9f35d5e86bffc4b04cb525736a23a7e72009efc
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Jul 19 16:46:09 2012 +0200

    Add support for network status parsing

commit 5cd457bebedb6e1582ce7d0d7782df80e51a7291
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Oct 12 08:49:56 2012 -0700

    Cleaning up orphaned *.pyc files
    
    A common gotcha when deleting or renaming files is that the python bytecode can
    still be imported and used without the source file. In other words if you
    delete 'foo.py' but forget to also delete 'foo.pyc' then your 'import foo'
    statements will still work.
    
    This has often bitten us because the tests still pass, so the problem goes
    undiscovered until someone new checks out the repository.
    
    Having stem's test runner check for orphaned pyc files and remove them prior to
    running its tests. Idea suggested by Ravi.
    
    https://trac.torproject.org/7103

commit 705b61674e8cec9e5608a32c6a50a708fee8c5f7
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Sep 25 09:51:26 2012 -0700

    Running apidoc when being built by read-the-docs
    
    The read-the-docs service runs sphinx-build, bypassing our custom Makefile. The
    only gotcha with this is that it means we don't call apidoc, causing our
    autogenerated documentation (which is... well, just about everything) to not
    exist.
    
    Making the apidoc in the conf.py file, using the READTHEDOCS environment
    variable to deterine if we're on that platform or not.
    
    Tested by first running a regular sphinx build to repro the apidoc not being
    called...
    
    sphinx-build -b html -d _build/doctrees . _build/html
    
    ... then did it again after setting READTHEDOCS to confirm that the autodocs
    now existed.

commit 66dea047a651d595b755c754d84250cd1dd687ac
Merge: 56a06b6 5a0da76
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Sep 24 09:01:38 2012 -0700

    Adding a method for MAPADDRESS
    
    Method to map address aliases.
    
    cr atagar (https://trac.torproject.org/6951)

commit 5a0da767e086aab1f196f60ed9fd042eb6825db4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Sep 24 09:00:08 2012 -0700

    Couple minor fixes for tests
    
    Correcting an ImportError due to test.util being moved, and preventing the 'is
    valid IP' check from being a no-op.

commit ddf7aa7a21bfc8418488b27e05fa7e53ef4f514c
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Sep 24 20:16:18 2012 +0530

    Fixing a bunch of things after cr #6951
    
    https://trac.torproject.org/projects/tor/ticket/6951

commit 060c82b3a874c88159ee4a20803c93f22512851a
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sun Sep 23 11:45:33 2012 +0530

    Fix a typo

commit 8e54f37f64ea04a0772d825276f4e5e4a34d4df2
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Sep 5 18:29:10 2012 +0530

    Implement Controller.map_address

commit 56a06b6283d5bcd3c5eb6d6c85b98b2949858cf9
Author: Ramkumar Ramachandra <artagnon@gmail.com>
Date:   Thu Sep 13 12:27:52 2012 -0700

    test/unit: correct a typo in rule.py
    
    A tuple was being initialized as a dictionary.  Correct this.
    
    Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>

commit 223ebc5617ce46007dfd5a9210e4ff91fb9767ea
Merge: 3df65c0 d97a0c3
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Aug 27 11:12:59 2012 -0700

    Adding a method for SETCIRCUITPURPOSE
    
    cr atagar (https://trac.torproject.org/6670)

commit d97a0c30002062ac0fbc45b32c2be6061cc35594
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Aug 27 11:08:10 2012 -0700

    Skipping repurpose_circuit() test if offline
    
    The repurpose_circuit() test uses 'GETINFO circuit-status' results which are
    empty when offline, causing an error...
    
    ======================================================================
    ERROR: test_repurpose_circuit
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/integ/control/controller.py", line 386, in test_repurpose_circuit
        first_circ = controller.get_info('circuit-status').splitlines()[0].split()
    IndexError: list index out of range
    
    ----------------------------------------------------------------------
    Ran 14 tests in 1.390s

commit d27fbbb06a247d3fb12684f76f2a7f5a51a36b04
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Aug 27 11:02:48 2012 -0700

    Implement Controller.repurpose_circuit

commit 3df65c080072d11d77664e2738652f3c0c08134e
Merge: 4f6a16c 74ba2e9
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Aug 24 12:51:00 2012 -0700

    Adding methods for SIGNAL and EXTENDCIRCUIT
    
    Feature additions by Ravi.
    
    cr atagar (https://trac.torproject.org/6659 and https://trac.torproject.org/6666)

commit 74ba2e9db42508a2bc2873dd078ae5bf2cbd140e
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Aug 24 12:38:20 2012 -0700

    Revisions for EXTENDCIRCUIT addition
    
    Couple minor changes:
    
    * The EXTENDCIRCUIT integ tests fails if we don't have an internet connection.
      Making it dependent on the ONLINE testing target.
    
    * Revising the pydocs a bit. For instance, there's no point in saying "If the
      purpose isn't provided, "general" circuits are built." when we can simply
      provide a default for the argument. ;)
    
    However, there's still a couple mysteries...
    
    * When I was offline I got the following stacktrace...
    
    ======================================================================
    ERROR: test_extendcircuit
    ----------------------------------------------------------------------
    Traceback:
      File "/home/atagar/Desktop/stem/test/integ/control/controller.py", line 373, in test_extendcircuit
        circ_id = controller.extend_circuit(0)
      File "/home/atagar/Desktop/stem/stem/control.py", line 1109, in extend_circuit
        raise stem.socket.ProtocolError("EXTENDCIRCUIT returned unexpected response code: %s" % response.code)
    ProtocolError: EXTENDCIRCUIT returned unexpected response code: 512
    
    ----------------------------------------------------------------------
    
      However, according to the control-spec the 512 response code is for "Syntax
      error in command argument". That doesn't make sense if we're failing because
      we lack a connection.
    
    * Is the 'path' argument for EXTENDCIRCUIT the circuits that we build through,
      or the relays to be chosen from when building a single hop? The EXTENDCIRCUIT
      description doesn't say, which seems to me to be a weakness in the spec.

commit fea0a43ded650b78d9453c1bc2ce15fdb99e394f
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Aug 23 22:07:01 2012 -0700

    Implement Controller.extend_circuit

commit 4f8b48ba3a7ee34a5725c9df242a961da9e7da01
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Aug 23 22:01:31 2012 -0700

    Collapsing SIGNAL documentation
    
    Shortening the pydocs for the signal() method. Might as well list both signal
    aliases on the same line.

commit aeb39b813961d02509a9c3e4e3db585b4190cac1
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Aug 23 21:52:01 2012 -0700

    Implement wrapper method for Controller to handle SIGNAL requests

commit 4f6a16c9526b45d3772abb797df2771bed6f884f
Merge: 92a0bf0 6d1f362
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Aug 23 21:44:33 2012 -0700

    Adding a method for USEFEATURE
    
    Methods to enable a feature and check if a feature is on.
    
    cr atagar (https://trac.torproject.org/6417)

commit 6d1f3625710eb382a810907a49b481ca14762783
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Aug 23 21:32:36 2012 -0700

    Minor revisions for USEFEATURE
    
    Few very minor tweaks...
    
    * 'GETINFO orconn-status' provides empty results when we don't have a
      connection, causing the test to fail.
    
    * Making is_feature_enabled() case insensitive.
    
    * Short circuiting is_feature_enabled() checks if it's in enabled_features.
    
    * The enabled_features attribute is simply used for an existance check, so we
      don't need to be too careful about deduplicaion. That said, we *do* need to
      be careful about the case of its entries.
    
    * Renamed the version constants since I'm trying to list their category at the
      start, to make a sort of namespace.

commit 51a00e9a3a2f2fa10277759616582d4acd2942a3
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Aug 23 00:05:10 2012 +0530

    Fix documentation formatting

commit 17d536657c43984665138ee131e77a6a291176d5
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Aug 22 23:45:49 2012 +0530

    Add Controller.is_feature_enabled
    
    Also made some minor modifications based on CR in #6417

commit 9a70131cf2bf0b25f89faa89c62fb70d5e891ee1
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Jul 3 01:34:46 2012 +0530

    Implement use_feature wrapper method

commit 92a0bf0d6afe39db45108626d2e89cde70997a2a
Merge: ac5be8c 683eeb2
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Aug 21 18:11:11 2012 -0700

    Caching expansion and test prompt
    
    I was planning to simply add caching for GETCONF and SETCONF requests, but
    ended up making a few other changes on the way. The most prominent is a new
    'prompt' script that you can use to quickly get python interpretor with a
    controller instance. Handy for testing.
    
    cr neena (https://trac.torproject.org/6631)

commit 683eeb25e812346cfd01790fe7729b686b8a2c2c
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Aug 21 18:10:30 2012 -0700

    Fixing issues spotted by Ravi
    
    Typos and whitespace issues spotted by Ravi.

commit 75d7f077e4b23361bc163171a4ec38b137f019ee
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 18 11:36:38 2012 -0700

    Setting string attribute of OperationFailed
    
    All OperationFailed subclasses had an empty string attribute, making stack
    traces pretty unhelpful.
    
    Before:
    
    >>> control.get_info('blarg')
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "stem/control.py", line 622, in get_info
        if default == UNDEFINED: raise exc
    stem.socket.InvalidArguments
    
    After:
    
    >>> control.get_info('blarg')
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "stem/control.py", line 622, in get_info
        if default == UNDEFINED: raise exc
    stem.socket.InvalidArguments: GETINFO request contained unrecognized keywords: blarg

commit f999b627dc3ab00cdd7c245d4811ad7b0a2cc20c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 18 11:21:08 2012 -0700

    Adding an easy method for getting a controller instance
    
    Adding a 'prompt' script that kicks off a python interpretor with a readily
    available stem.control.Controller instance. This starts a tor test instance if
    it isn't already running and offers to shut it down when done.
    
    The purpose of this script is to make it easy to get a controller instance we
    can exercise. The sh shebang makes this *nix specific, though I suppose windows
    users can still use it by running 'bash prompt'.
    
    Example usage:
    
      atagar@morrigan:~/Desktop/stem$ ./prompt
      Welcome to stem's testing prompt. You currently have a controller available
      via the 'control' variable.
    
      >>> control.get_info('version')
      '0.2.1.30'
      >>> quit()
    
      Would you like to stop the tor instance we made? (y/n, default: n): y

commit a6c5a063d1f963a53e012b4c31478f90dea0d795
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Aug 17 10:24:16 2012 -0700

    Better validation of controller's from_port() function
    
    I scratched my head due to an awefully misleading stacktrace when I passed a
    port value into the method's argument slot. Validating the inputs to avoid
    having this cause confusion in the future.
    
    Before:
    
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "test/prompt.py", line 67, in controller
        controller = stem.control.Controller.from_port(CONTROL_PORT)
      File "stem/control.py", line 457, in from_port
        control_port = stem.socket.ControlPort(control_addr, control_port)
      File "stem/socket.py", line 287, in __init__
        if connect: self.connect()
      File "stem/socket.py", line 171, in connect
        self._socket = self._make_socket()
      File "stem/socket.py", line 311, in _make_socket
        control_socket.connect((self._control_addr, str(self._control_port)))
      File "/usr/lib/python2.7/socket.py", line 224, in meth
        return getattr(self._sock,name)(*args)
    TypeError: coercing to Unicode: need string or buffer, int found
    
    After:
    
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "test/prompt.py", line 67, in controller
        controller = stem.control.Controller.from_port(CONTROL_PORT)
      File "stem/control.py", line 459, in from_port
        raise ValueError("Invalid ip address: %s" % control_addr)
    ValueError: Invalid ip address: 2779

commit fc736ff07c2ded068ee3e5b9ad1028923ee89c47
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Aug 17 10:07:55 2012 -0700

    Fixing circular import
    
    Correcting a circular import between the connection and control modules...
    
    >>> import test.prompt
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "test/prompt.py", line 15, in <module>
        import stem.control
      File "stem/control.py", line 48, in <module>
        import stem.connection
      File "stem/connection.py", line 106, in <module>
        def connect_port(control_addr = "127.0.0.1", control_port = 9051, password = None, chroot_path = None, controller = stem.control.Controller):
    AttributeError: 'module' object has no attribute 'control'

commit c87b1005ad045b4554c3ed4c2ba6ce6e54aa81dc
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Aug 6 10:04:40 2012 -0700

    Caching and logging for GETCONF/SETCONF
    
    Caching and logging in a similar fashion to GETINFO. Remaining todo items...
    
    * This is broken if another controller tampers with our configuration values.
      We should listen for configuration change events to catch that.
    
    * Calling 'GETCONF blarg' causes an exception without a message. I should
      figure out why.

commit 2a34547cc6dc157617c4a485d7e9e41816d46e66
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Aug 6 08:45:07 2012 -0700

    Caching parsed controller version
    
    Our controller version is something that we'll frequently look up. There's no
    reason to leave it uncached.

commit 9fe8da962aadec5d56bfeb6d6635a01228514ca2
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Aug 6 08:24:16 2012 -0700

    Pydoc mistake
    
    Actually, I'm not sure what I was trying to say there. Issue spotted by
    Karsten.

commit ac5be8cee5f58517c09217fd6860f7ef483dacc2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Aug 5 20:14:18 2012 -0700

    Fully parsing transport lines
    
    The extrainfo descriptor's new transport lines were a little funny in that
    they're only relevant to bridges, and when they apprear in bridges they're
    scrubbed of everything except the transport name. This in turn meant that they
    only appeared with transport names in the wild.
    
    However, now that transport lines can appear in non-bridge relays I'll
    sometimes see complete transport lines, so parsing and validating them
    properly.

commit 3f380f95876baffe48725828c23706dd9b46b760
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 4 19:49:39 2012 -0700

    Parsing transport lines in non-bridge relays
    
    Spotted a relay in the wild with a transport line and asn thinks that, though
    it's an oddity and doesn't make sense for non-bridges, this isn't a bug...
    extra-info toorvoid 0B37323298FF98CD86ED404895BB27B7426E8AE1
    published 2012-08-04 12:19:13
    transport obfs2 83.212.96.201:33570
    
    Expanding the parser to allow for it in any extrainfo descriptor.

commit 73fcfaf8ccbc2c1cb714f35fcc8e2823daf39808
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 4 18:52:33 2012 -0700

    GETINFO caching and logging
    
    Caching static GETINFO results, short circuiting geoip lookups when the geoip
    db is unavailable, and logging the call runtimes.

commit f7157ae4f576acb1b90e458f479e2a820aefbe86
Merge: 051ed27 7022021
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 4 12:18:56 2012 -0700

    CSV export functionality for tor descriptors

commit 7022021c4207a0066cb93261629f2ba020d307f6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 4 12:13:47 2012 -0700

    Revising descriptor exporter and associated tests
    
    Several changes for the prior csv export functionality, especially its tests.
    Most notably this includes...
    
    * rewrite of the tests to be simpler and break down by test cases
    * newline breaks rather than windows style '\r\n'
    * excluding private attributes (attributes that start with an underscore)
    * writing directly to a file in export_csv_file() rather than buffering to a string first
    * general refactoring to hopefully improve readability

commit ff43d9e7ad306499dce6e20909d0ce2e842541b4
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Aug 4 10:48:06 2012 -0700

    Moving descriptor contructor to mocking module
    
    We'll need to construct server descriptors in both their unit tests and when
    testing the exporter. The mocking module is both for function mocking and
    creating mock objects so this is the perfect place for it. I might move the
    extrainfo descriptor constructer here later if it seems useful.
    
    Personally I'm finding the current exporter tests and mocking changes very
    confusing so reverting the mocking changes and writing some simpler tests.

commit bb0dcfc7fdf9303251dbe7304005d349f9956fd6
Author: Erik <eislo@wesleyan.edu>
Date:   Wed Aug 1 16:40:56 2012 -0400

    Submitting for code review.

commit e94e00a61d262fd52706c0891276811ca0369e52
Author: Erik <eislo@wesleyan.edu>
Date:   Tue Jul 31 12:38:27 2012 -0400

    minor documentation update

commit aafb7b429e7220c938da7a591d17ca1b2463aa09
Author: Erik <eislo@wesleyan.edu>
Date:   Tue Jul 31 12:35:09 2012 -0400

    First complete build of export.py tests.
    
    Everything should pass, and testing export_csv_file() promted some changes,
    which include the change from specifying a file for writing to providing
    a document object.  This code is now ready for review.

commit 9440f4ffa20aa7ba2a0e3297cbf5b245bf2c1c91
Author: Erik <eislo@wesleyan.edu>
Date:   Mon Jul 30 18:09:13 2012 -0400

    First complete build of unit tests for torexport.
    
    There is a bug in test_export_csv_file which will be fixed tomorrow
    morning, otherwise all tests are functional and should be passing.

commit 4ea5cf2e2d8f7bb27db96d35d8fc0538b2745e70
Author: Erik <eislo@wesleyan.edu>
Date:   Tue Jul 17 17:46:16 2012 -0400

    Early unit test builds with lots of holes to fill.
    
    Problems encountered include mocking functions that utilize keyword args,
    passing descriptor objects to functions being tested rather than just
    dictionaries, and determining if, in the case of the user defining both
    include_fields and exclude_fields, an error should be thrown or the
    functions should simply remove any overlap from the two lists.

commit 305f7fdb30f420a751f2de29f250170c19569bb6
Author: Erik <eislo@wesleyan.edu>
Date:   Thu Jul 12 15:24:50 2012 -0400

    Updated some of Damian's suggested revisions.  Not sure of others.

commit 508373f1a7824b1ff9b86a597a718918d4250e24
Author: Erik <eislo@wesleyan.edu>
Date:   Thu Jul 12 14:13:59 2012 -0400

    First draft of torexport, 3 functions defined to handle different use cases.
    
    First draft of torexport, 3 functions defined to handle different use cases.
    csv_file_exp() takes a list of descriptors and exports their attributes as a
    csv file. descriptors_csv_exp() takes a list of descriptors and returns a
    generator.  Each iteration, this function returns a line of csv representing
    the attributes of a single descriptor object.  descriptor_csv_exp() returns
    a single line of csv based on the attributes of a sole descriptor object.

commit 051ed27b99a5b8796777b86360a41bec52527b12
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jul 24 08:42:56 2012 -0700

    Descriptor pydoc mistake
    
    The parse_file() helper no longer runs over bridge descriptors, just relays.
    The reason for this is that only relay descriptors exist in the
    cached-descriptors file so there's little need for this to include hacks to
    differentiate between the two.

commit e2528b85f6b0bf1fb82c42599e71d1c925fcd6ae
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jul 23 10:09:38 2012 -0700

    Calling QUIT prior to disconnecting socket
    
    Making the Controller's close() method try to be a nice controller by issuing a
    QUIT call before detaching the socket. This is a best-effort call (errors
    ignored) since the caller just wants the socket shut down and we don't need the
    QUIT to succeed to do that. Original change by gsathya and discussed on...
    https://trac.torproject.org/6234

commit e4699ba9545a99ec58b7d211dc7bff8d236fbb8e
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jul 23 10:04:23 2012 -0700

    Including common tag comparisons in version unit tests

commit 63519a16c31b6478e912c41cd7d47e1fcb634750
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 21 22:28:41 2012 -0700

    Including tor's git commit in Version
    
    Expanding the Version class to include an 'extra' and 'git_commit' attribute
    that reflected a proposed expansion of the spec...
    https://trac.torproject.org/6445
    
    This is already the de-facto form of tor versions so moving ahead with
    including it in the Version class.

commit 43db1dea2487ed7517c8be3f44a315fdeac731a8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 21 19:55:51 2012 -0700

    Sorting tor versions lexically
    
    According to the version spec we should "compare them lexically as ASCII byte
    strings". This seems a little weird since it means a reverse alphabetical order
    (ie 'Z' comes before 'A') but oh well.

commit fd2da106d0f90061532f8880702138b297ca4230
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jul 20 18:19:29 2012 -0700

    Pylint fixes
    
    Variety of issues, most of them caught by pylint.

commit d7ca32639205ca99ca053270b98b90dee4cc7986
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jul 19 09:08:25 2012 -0700

    Adding exit_policy to stem's __init__.py
    
    Forgot to add the new exit_policy.py module.

commit 78a997ea34aaee8cc4515382fb0f0ec4814c32a4
Merge: e68adb7 e670090
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jul 19 08:58:07 2012 -0700

    Adding an ExitPolicy class
    
    First draft was ported from arm by gsathya in...
    https://trac.torproject.org/5454
    
    I've since rewritten most of it to support IPv6, masks, and properly comply
    with the Tor spec for an exitpattern (arm's class was written to support torrc
    policies, which are different).

commit e6700902cd914922515f17179ee1e8e97a5af507
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jul 19 08:55:30 2012 -0700

    Minor post-rebase revisions
    
    Just a couple things I spotted after rebasing my exit_policy branch onto the
    current master.

commit 59ed7fdd4f90fe6ac804b7b24dfdc0c3f2bc68dd
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jul 19 08:47:43 2012 -0700

    Using the ExitPolicy class for ServerDescriptors
    
    The ServerDescriptor class had a todo item to replace the lists of strings that
    it was using for an exit policy with the ExitPolicy class when we had one.
    Making it so...

commit af9e5b4e9f4dcb149d37c7f7ad0249abaa56e689
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jul 18 09:55:06 2012 -0700

    Revised MicrodescriptorExitPolicy
    
    Rewrite the MicrodescriptorExitPolicy and expanded its tests.

commit 6d8dcbec4aacfc27ea43e28deea19675c715a28d
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jul 17 09:53:51 2012 -0700

    Tests for the ExitPolicy's __str__ and __iter__ methods

commit c1ced39c89092a81dad3bd3d8ad3545296a64190
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jul 17 09:47:21 2012 -0700

    Test and fix for the ExitPolicy's summary() method
    
    Ok, I was being stupid. The performance issue for the summary() method was due
    to an O(n) lookup when checking if we had seen a port (which made, for
    instance, to full-port-range policies choke on a 65535^2 operation). Swapped
    to using a set instead to make it constant time.

commit 22ec4c9226da797778f56d8d1303b61c4e3bf710
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jul 17 09:29:32 2012 -0700

    Dropping the ExitPolicy's test_parsing()
    
    The test_parsing() test is redundant with the ExitPolicyRule tests, so dropping
    it.

commit ac5b8e74f4b07ee0e2ab25745519417711bdf576
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jul 17 09:27:31 2012 -0700

    Tests for is_exiting_allowed() and can_exit_to()
    
    Tests for a couple more ExitPolicy methods, and fixes for issues with
    is_exiting_allowed() that they revealed.

commit 1936c4d5bdb861c34409a3e366e63035cce3a4a3
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jul 17 08:29:56 2012 -0700

    Test for the set_default_allowed() method

commit 6f36f95711d54fee1bdf7391842f5e76dde8538b
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jul 17 08:10:17 2012 -0700

    Adding a test for pydoc header examples
    
    Little test for our pydoc example, and fixing a couple little bugs. There's a
    larger bug due to our naive summary() implementation though a better solution
    isn't immediately coming to mind.

commit 384e78df0594c7d60af4b4a0096b12d4419db7f8
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jul 17 07:58:14 2012 -0700

    Correcting and expanding summary() example
    
    The ExitPolicy class no longer has a from_str() function. Also adding an
    example that better exemplifies how ranges work.

commit 5ccc2ca7b4d7ab1974871400e0b8fb3ff1478df4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jul 16 09:55:06 2012 -0700

    Test for the ExitPolicy's constructor
    
    Test to make sure that we can handle both string and ExitPolicy lists. This
    also checks that we can easily handle the split() output when breaking up a
    csv.
    
    Added __eq__() methods for the ExitPolicy and ExitPolicyRule to make policy
    comparisons easier.

commit 363e87f5bd55b4ecac9514d6f423fba52d4f55ec
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jul 16 09:05:25 2012 -0700

    Revised ExitPolicy class
    
    Revising the ExitPolicy class to something resembling its final incarnation.
    Next gonna revise the tests (they pass, but could use some love).

commit 004650ae2b09ec295f6e7ce4d70993c1764761d9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 15 14:53:53 2012 -0700

    Splitting up ExitPolicyRule parsing
    
    The ExitPolicyRule constructor was unpleasantly huge (and by extension, hard to
    read). Moving most of it to helper funcitons for parsing the addrspec and
    portspec.

commit 77c3a5ffa871accde7cb62fb3f9c40b30051c9c2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 15 14:06:16 2012 -0700

    Removing ExitPolicyLine and ExitPolicyError
    
    The ExitPolicyRule is a drop-in replacement for ExitPolicyLine, and there's
    little reason to introduce a custom ExitPolicyError exception when a ValueError
    effectively describes its use cases.

commit 0c069727eb5337cc7c3f422387e3f4d0ad5caa9c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 14 18:32:50 2012 -0700

    ExitPolicyRule class
    
    Adding a class that conforms with the dir-spec's exitpattern entity. Plus lots
    'o unit tests that told me over and over and over again that I was being
    stupid. Hopefully this version is at least kinda sorta close to being right...

commit ebec5fbd4a8080b1003ae87681cd5b804f5a81d7
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jul 13 09:54:42 2012 -0700

    Utilities for IPv4 and IPv6 addresses
    
    Adding the utilities that I need for the ExitPolicy to handle and translate
    addresses and their masks. Pity that these aren't provided by the python
    builtins. The IPy package seems to do it but this isn't worth adding a new
    dependency.

commit 30ba1019fcbeca9220e85fa5216b6bac1b5db16f
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Tue Jul 3 02:03:12 2012 +0530

    Add microdescriptor exit policy
    
    The microdescriptor policy class can now parse
    exit policy summaries. Add tests for this

commit fdb140763097c67c9612c5e7090534330e89a68f
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Mon Jul 2 19:19:56 2012 +0530

    Add exit policy test to run_tests

commit 03f625b328ba9148909dc523196a16f4356b9dc2
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Tue Jun 26 08:51:24 2012 +0530

    Add unit test for exit_policy
    
    Test the validation and parsing of the exit policy

commit 289063fefe8857c6ca10e95d2a3a90fcbd925f51
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Tue Jun 26 08:49:53 2012 +0530

    Remove MicrodescriptorExitPolicy, add validation and exception
    
    The micro desc exit policy is currently useless until we figure
    out what it's used for.
    
    Validate all the ip address and ports. Raise exception if wrong

commit b51d8da9bd2a6d4377019b37bbad768ad8ba17ca
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Tue Jun 26 08:48:23 2012 +0530

    Make stem.util.connection.is_valid_port accept a list
    
    Iterate through the list and check if each item is a valid
    port or not.

commit a8b654e046e143bb89c40021c0602dd9510912fe
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Mon Jun 11 23:27:47 2012 +0530

    Add MicrodescriptorExitPolicy
    
    This class has four methods -
    add() - which is the public method that the user calls to add
    exit rules or policy. If a policy is added, then this instance
    can't be changed further. Rules can be added till we we get a
    wildcard('*') as a rule.
    add_rule() - parses the rule
    add_policy() - parses the policy
    __str__() - creates a string representation of the policy

commit 0a1c6238cc95f37a1b26369247d5d9876ba2b804
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Thu May 17 22:19:10 2012 +0530

    Add ExitPolicy.get_summary()
    
    This provides a summary description of the policy chain
    similar to the consensus.

commit 4d88884b0cfdc98ca4c4741d4e1073267418038c
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Fri Mar 23 15:18:44 2012 +0530

    Add doc and fix whitespace

commit 607e32a3cf3d69ed185ecee4e62f55e7101d587a
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Fri Mar 23 14:10:54 2012 +0530

    Parse "private" alias better

commit 80228a52e3f9507547cf924156a1bcafdea6540b
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Fri Mar 23 12:21:01 2012 +0530

    Overhaul exit_policy.py
    
    - Fix Indentation issues
    - Remove ExitPolicyIterator class
    - Add ExitPolicyLine class
      - This class represent a single line from the Exit Policy.
        (provides much better abstraction IMHO)
      - __init__(), __str__(), and check() are copied from ExitPolicy
    - Changes in ExitPolicy -
      - Now, it acts a wrapper class to ExitPolicyLine
      - It contains a list ExitPolicy._policies which stores all the policies
        and each item in _policies is an object of ExitPolicyLine
      - check(), isExitingAllowed(), __str__(), and __iter__() are
        changed to provide a wrapper of sorts for ExitPolicyLine
      - add() is now used to add an exit policy line - This creates
        an object of ExitPolicyLine and adds it to ExitPolicy._policies

commit 88c436276f1aa0d20206499099b689eb50ace573
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Thu Mar 22 23:20:43 2012 +0530

    Add ExitPolicy, ExitPolicyIterator from arm/torTools

commit e68adb7fdd465e1780bc4ee1243e0c226add77d9
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Jun 27 18:34:15 2012 +0530

    Implement convenience method for handling SAVECONF requests
    
    This also changes the Exception structure a bit. I have added an
    InsatisfiableRequest class which will be raised when a valid request
    couldn't be satisfied by Tor.
    
    I've also added a super class for non-socket non-protocol errors called
    OperationFailed. Any error that was raised by an error response returned
    by Tor (i.e., one with an error code & a message) should subclass this.
    
    The exceptions now look like
    
      ControllerError - Base exception raised when using the controller.
        |- ProtocolError - Malformed socket data.
        |- OperationFailed - Tor was unable to successfully complete the operation.
        |  |- UnsatisfiableRequest - Tor was unable to satisfy a valid request.
        |  +- InvalidRequest - Invalid request.
        |     +- InvalidArguments - Invalid request parameters.
        +- SocketError - Communication with the socket failed.
           +- SocketClosed - Socket has been shut down.

commit 611e75dbeb5a81a655ca0f8ec1b4495850c3459a
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jul 10 08:22:55 2012 -0700

    Allowing extra-info dirreq-v*-share to be above 100%
    
    Spotted an extra-info descriptor with dirreq-v2-share and dirreq-v3-share
    values over 100%...
    https://lists.torproject.org/pipermail/tor-dev/2012-June/003679.html
    
    Karsten suggested simply removing the restriction on this field's upper bound
    since it's soon going away.

commit a0f980c996587ef112379891339d463bda3d487a
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jul 9 12:00:08 2012 -0700

    Supporting multiple extra-info transport lines
    
    The 'transport' lines being added to extra-info descriptors can appear any
    number of times, so storing them as a list. These options are weird in that
    they have an "address:port [arglist]" which never appears in the wild since
    they both only appear on bridge descriptors and are scrubbed from bridge
    descripotrs.

commit d52e934c24c9d327e277eb60bfee060697cb43ec
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jul 9 11:52:56 2012 -0700

    Only consulting major descriptor versions
    
    Minor descriptor version bumps are backward compatable changes. We might not
    recognize new additions, but it won't break us either (they'll just be
    unrecognized). Pointed out by Karsten on...
    https://trac.torproject.org/6257

commit ab553f54e70ea21d8bd117cf912e1b208417e5d5
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jul 9 11:34:33 2012 -0700

    Minor revisions for LOADCONF and prior cleanup

commit 6cc4e195452bc573e34f0b2de295b1028a20eb4a
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Fri Jun 29 00:25:17 2012 +0530

    Implement Controller.load_conf

commit c5472856c9895bd118af79297ccdce59fcba2e13
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Jul 9 10:12:23 2012 +0530

    Remove ugly if-elif tree in favour of a dict

commit ac0d2183f04838790c64ff3194ae29f90fd4ed93
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Jul 9 09:55:38 2012 +0530

    Aesthetic changes to __init__.py files
    
    Making __all__ lists multiline when they're long and awkward.

commit 6a91cda3db2a1ab1c4bf485fad03fd5cea24657a
Merge: a325a40 3d15387
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 8 13:13:17 2012 -0700

    Controller methods for GETCONF and RESETCONF
    
    Feature addition by Ravi, addressing...
    https://trac.torproject.org/6239

commit 3d1538746adad85bb7235d8bc7f4b3d93b1f3d50
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jul 8 12:39:29 2012 -0700

    Splitting set_conf/reset_conf back up
    
    Ravi made a good point that the set_option() syntax for resetting config
    options is now worse, so trying a compromise of our approaches.
    
    This reintroduces set_conf() and reset_conf() methods, but also keeping my
    set_options() method to handle the more complicated use cases (batch setconf
    requests, resetconf with values, and context sensitive options).
    
    Personally I found the old set_conf() method confusingly overloaded (python
    does not handle overloading use cases very well). This solution lets us keep
    the set_conf() and reset_conf() methods simple, while also providing something
    that's reasonably intuitive (imho) for the complicated use cases.

commit e2057f1beb158ce111ddc735c8b1a58fab79f6d1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 7 19:26:56 2012 -0700

    Reintroducing tuple support to get_options
    
    Damn I hate it when I'm right. The reason that arm supported tuple lists is the
    fucking HiddenService* options. Adding support for them back in, though I gotta
    admit that I flirted with the idea of just saying "screw hidden services".
    These things are such a hack...

commit f9b370c455d4128aabe73702a7a445dfe17d4a2f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 7 18:50:17 2012 -0700

    Revisions for prior SETCONF and RESETCONF additions
    
    Several revisions to the prior changes, the most prominent of which is to use
    an API more similar to arm's. The SETCONF and RESETCONF methods are
    functionally identical except how they handle undefined values, so there's
    little reason to provide separate methods in our controller. Just using an
    optional 'reset' arg instead.
    
    One thing that arm got wrong (and so did stem) is that our method to change
    multiple options should accept a dict rather than a tuple list. Order *might*
    matter for the accursed HiddenService options, but for the 99.9% use case a
    tuple list is a very strange argument to accept for this.
    
    This also includes some testing fixes, for instance...
    * Testing when we have a list argument (for instance, setting our ExitPolicy).
    * Failing our test if one of the calls that should fail actually succeeds.
    * Reverting the configuration options that we change after the test.
    * An identical test for context sensitive options was added to both the getconf
      and setconf test. Definitely a good use case to check and it could belong
      with either setconf or getconf, but doing the same test twice is pointless.

commit da66a5dfa3818de5376e252b667a649abccf9702
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Fri Jul 6 21:04:53 2012 +0530

    Make get_conf_map work properly with HiddenService* options

commit 9af2616aabb3bd515d695f3ab4ed1c839428bd4b
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Jul 3 00:07:36 2012 +0530

    Fix documentation formatting

commit 7d8b1fdef6e6b2cd38a7c2f3aa09a957b03af1a5
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Jun 27 01:28:39 2012 +0530

    Make tests cross-platform & fix things that broke during a rebase

commit ea57e936e08400d63a423c669311643afd4dd84d
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Jun 26 10:04:52 2012 +0530

    Add Controller.reset_conf and fix Controller.set_conf

commit 85c923bfc784aaf8080e08053ef6b577e8c23944
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Jun 21 22:08:30 2012 +0530

    Add SETCONF integration tests, SingleLineResponse unit tests and the bugfixes that come with them

commit 51025794e68b1607e6a15741ebdab45650df0c25
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Jun 21 20:30:04 2012 +0530

    Add unit tests for SingleLineResponse and add missing imports

commit 4ed9a1fba65257ccd0caba41da79ac6b879737f2
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Jun 21 13:58:23 2012 +0530

    Add set_conf wrapper method

commit ba53da1772007aca88bc75f1fed60d49170eb964
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Fri Jun 15 18:21:47 2012 +0530

    Add a generic SingleLineResponse class

commit 0bfa815f85337cbd216d0fbb2965b6ab305d3874
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Jul 2 22:48:09 2012 +0530

    Fix Controller.get_conf_map's documentation formatting

commit a325a40ff04df1a83614c529995749f2da8e79f8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 7 20:42:45 2012 -0700

    Replacing platform specific %D date formatting
    
    Python's strftime() method has a set of platform independent flags, but also
    accepts anything that our platform's libc does...
    http://stackoverflow.com/questions/10807164/python-time-formatting-different-in-windows
    
    I was using '%D' date formatting for our logging, but that doesn't work on
    Windows...
    https://trac.torproject.org/6206
    
    Replacing with the equivilant platform independent flags. Issue caught by Beck.

commit 74597a78159acf20146c72d0392e74f07d2bf9e3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 7 20:23:27 2012 -0700

    Skipping proc tests broken by tor's disabledebugger features
    
    Stem's proc integ tests exercise that utility by running it against our tor
    instance. However, as discussed in ticket #3313 the DisableDebuggerAttachment
    feature screws up our proc permissions. Skipping those tests if they would fail
    due to this.
    
    Caught by Ravi in...
    https://trac.torproject.org/6286

commit 6aa207434661434984a7ea5b8cf3d2ab013dea69
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 7 20:14:19 2012 -0700

    Processing proc stat files with more than 44 entries
    
    A long-standing arm and stem bug is that it checks if '/proc/<pid>/stat'
    contains exactly 44 elements and, if not, refuses to parse it. This is a
    mistake, stat files may have more entries depending on the platform. Changing
    the check to account for this.

commit bbcff94dee48e96db3e7c9b396926a9fac7f64b9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 7 20:07:18 2012 -0700

    Accounting for logging.Handler bug in python 2.5
    
    In python 2.5 logging.Handler doesn't extend object, causing our super() call
    to fail with a stacktrace. Hacking around it.

commit 48404ed766a775b2d958d9b4b686e96cc88914c0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 7 19:55:24 2012 -0700

    Skipping tests that make repeated connections on OSX
    
    Stem's tests currently hang on Mac OSX, and that doesn't look to be getting
    better any time soon...
    https://trac.torproject.org/5917
    
    I've narrowd this down to being either a python, tor, or mac bug with how
    repeated socket connection/closures are handled on that platform. For now
    skipping the tests that encounter this issue if we're running on a mac.

commit 4c7831fee70848b002b97d8d8dd4b30519ec9eb9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jul 7 13:15:52 2012 -0700

    Clarifying comments for the get_pid_by_open_file test
    
    The prior fix is better for the test (working == better!), but makes it a bit
    less self-descriptive. Adding a couple comments.

commit bdc71d9c70e6184f7450e1fecef2cb2ce7bc4ac5
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Jun 27 16:44:01 2012 +0530

    Fix #6248 and make test_get_pid_by_open_file cross-platform while doing so

commit 6631bf536adb73f7499bde68a0f1289a51e8f2aa
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jul 6 22:33:15 2012 -0700

    Removing broken stem.util.control import
    
    I deleted stem/util/control.py but forgot to rm the pyc afterward so stem
    continued to work for me despite having a broken import. Caught by Beck.

commit c9e2f7be38e7dede69a7898c7627294f857ea395
Merge: af27984 c09b80e
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jul 4 14:32:25 2012 -0700

    GETCONF handling

commit c09b80ef912bb590a37e0d4533a34bbba14cabf9
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jul 4 14:25:32 2012 -0700

    Revisions for GETCONF handling
    
    These changes have gone through several review iterations so there wasn't much
    to tweak with it. Only substantial twiddling on my part was...
    
    * The super() method is new to me, and looks like it's the preferred method
      (especially in Python 3.x when it's syntactically much nicer). Swapped all of
      our constructors to use super() rather than Parent.__init__(self, args). This
      also forced me to fix some base classes that weren't inheriting from object.
    
    * Moved _case_insensitive_lookup() helper into controller module rather than
      having a separate util.
    
    * Removed the "GETCONF reply doesn't match the parameters that we requested."
      check since this'll always produce an error if tor adds new context sensitive
      options (my bad).
    
    * The get_conf_map() method wouldn't accept context sensitive options like
      HiddenServiceDir.
    
    * The get_conf_map()'s default value is what we should return if the lookup
      fails, not the mapped values (ie, get_conf_map("blarg", {}) should give an
      empty map).
    
    * Integ test was missing a requirement check that we could make a control
      socket, causing the RUN_NONE target to fail.

commit ffd68ccc660a92cc3cff61ae8baafebb6799e52e
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Mon Jul 2 09:39:35 2012 +0530

    Fix a bug because of which get_conf wouldn't work for HiddenService*

commit 765c497f824fde380d8d1b864e3dec36fcce3b11
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Fri Jun 29 23:12:34 2012 +0530

    Changes for review

commit a035e89b522950a0bef54840ace887d2af31095c
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Jun 21 03:39:34 2012 +0530

    Documentation fixes & some code refactoring

commit 9e40210c83c2af32fecb08e4ba4ea71833ce8050
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Jun 21 03:39:34 2012 +0530

    Behave smartly with context-sensitive config keys

commit 0b83af2922b5c377161626c7aef2af14a0152463
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sun Jun 17 22:27:55 2012 +0530

    Fix a bug where getconf parses didn't set the default correctly for multiple keys
    
    and write a test to check for this

commit f6231a9dd9ed52dae4e77e153cae13e344a4e094
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sun Jun 17 20:29:58 2012 +0530

    Add getconf test to check for multiline configuration keys

commit b8959dfe154180018071a07d7ad8eceb68a4c516
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sun Jun 17 18:05:54 2012 +0530

    Various modifications to GETCONF parsing
    
    * Change the way entries are stored. Now only stores one value unless the caller
      explicitly mentions that he want to retrieve multiple values using the
      'multiple' argument.
    
    * Stop checking if the received configuration options are the ones that were
      requested. (The HiddenService options do this)
    
    * Minor documentation fixes.

commit a80a784b29a91fb3ee9a6ec8a30b829b68648e41
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Jun 14 12:16:44 2012 +0530

    Add attributes to InvalidRequest and InvalidArguments
    
    Adds attributes to store the error code and error message
    to InvalidRequest and InvalidArguments

commit 42ae08cfdda0ea2f90bccb2b3621b182ddf99e0a
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Jun 13 07:38:13 2012 +0530

    Make the GETINFO parser raise InvalidArguments instead of ProtocolError when appropriate

commit ece7f85dd675633de3b866c868bba1560fd22060
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Jun 12 21:07:13 2012 +0530

    Minor change to GetConfResponse documentation

commit 3b25147b7fcaadbc7e2aec6ec89158ed1d738687
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue Jun 12 19:59:30 2012 +0530

    Multiple fixes to GETCONF parsing
    
    * Add stem.socket.InvalidArguments and make the GETCONF parser raise it instead
      of stem.socket.InvalidRequest.
    
    * Fix the parser to parser to correctly parse multivalue configuration keys
      (such as ExitPolicy). Add tests for the same.
    
    * Fix tests to run against a Tor client using a control socket file.
    
    * Minor changes to documentation.

commit 16d392ac74996ae38a8f3a33191ebc1899a5cf8f
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sat Jun 9 09:56:17 2012 +0530

    Add integration tests for testing Controller.get_conf

commit f01a50e5bd1b24cb30676d869057fea3c8e04af9
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sat Jun 9 09:55:01 2012 +0530

    Add get_conf convenience method to Controller

commit 0db367a151dc75fde8a3a90fd273bda7ad9227ab
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sat Jun 9 09:49:07 2012 +0530

    InvalidRequest now inherits from ControllerError
    
    Moved stem.response.InvalidRequest -> stem.socket.InvalidRequest and made it a subclass of stem.socket.ControllerError

commit f908ce9c44ffad9237d0f571351adb2143001f52
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Fri Jun 8 17:50:06 2012 +0530

    Add unit tests for GETCONF parsing

commit 5701329478dfc363da17d80cafca26f07a9a9def
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Fri Jun 8 12:41:09 2012 +0530

    Add GetConfResponse class for parsing GETCONF responses

commit af27984ffcaa418fb71e68992976967a06c4ba56
Merge: 2fe0f2c c043464
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 30 15:51:59 2012 -0700

    Testing fixes for Windows

commit c0434646354a6031fd8fb0750af4700f1a450f1a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 30 15:50:14 2012 -0700

    Minor formatting tweaks

commit 489c7bab48001e8af82b897f2d7ab5a487169643
Author: Beck <csybeck@gmail.com>
Date:   Fri Jun 29 23:22:36 2012 +0800

    Skipped some unrelevant integ test cases on windows

commit 7515471fa13c92873307054ea9e40ef5eea64c4a
Author: Beck <csybeck@gmail.com>
Date:   Wed Jun 27 00:31:55 2012 +0800

    Replaced a hardcoded path in test_relative_cookie

commit c001394928e01eb9807c747fed0283c7748578f9
Author: Beck <csybeck@gmail.com>
Date:   Tue Jun 26 23:58:32 2012 +0800

    Fixed test_expand_path_unix, and added test_expand_path_windows
    
    test_expand_path_unix was changing os.path.sep in order to change the
    path seperator used by os.path.join. However, this did not work. The
    path seperator was hardcoded in os.path.join, which is posixpath.join
    on *unix systems and ntpath.join on windows. We now mock os.path.join
    to use posixpath.join or ntpath.join in different test cases.

commit 8edfe7123997e99885ae514bbc1ae76d4c9bc1f8
Author: Beck <csybeck@gmail.com>
Date:   Fri Jun 22 23:31:39 2012 +0800

    Fixed bugs in calls to test.runner.skip()

commit 62e51e9e98ef2d2f309eece027d7d2417e18dedd
Author: Beck <csybeck@gmail.com>
Date:   Fri Jun 22 23:22:42 2012 +0800

    Fixed bug that util.system.expand_path does not expand tildas on windows

commit 9724963081f269eb68b17cf116c950dbd6591474
Author: Beck <csybeck@gmail.com>
Date:   Fri Jun 22 19:51:59 2012 +0800

    Replace calls to os.uname() by platform.uname()
    
    os.uname() is unix-specific, while platform.uname() is a more portable
    uname interface that would also work on windows. And platform.system()
    == platform.uname()[0].

commit 2fe0f2c84f0d41111571f2ae02575124da008743
Merge: c35c8a6 14087a7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 30 15:41:40 2012 -0700

    Unit and integration tests for stem.util.proc

commit 14087a70e2c0e40541d610ce41c4b08c8fab6b3e
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jun 29 08:42:08 2012 -0700

    Revisions for proc unit and integ tests
    
    Lots 'o changes, most just nitpicks to satisfy my OCD. The main functional
    changes are just...
    
    * The mock_fn() and find_subsets() helpers are useful for other tests so moving
      them to the mocking module. The authentication unit tests, for instance, had
      a helper that this has replaced.
    
    * The default target for the mock_fn() calls doesn't seem to be needed, and
      could mask issues that prevents proper system abstraction (ie, if we're
      passing through to os.readlink() then we probably won't notice that the test
      only works on Linux).
    
    * The stem.util.proc.is_available() check was broken since we were using the
      function as a boolean, which is always True...
    
      >>> def foo():
      ...   return False
      ...
      >>> bool(foo)
      True
    
    * We created a controller to spawn a control connection, but forgot to close
      the controller afterward.
    
    * Test broken if we were running without a control port (for instance, with the
      RUN_SOCKET target).

commit af9d226e851b3ca1ab552a3fbd249a61602b8fd3
Author: Megan <mchang01@wesleyan.edu>
Date:   Thu Jun 28 16:51:48 2012 -0400

    Removed commented out code that was forgotten.

commit 830bc5fa28f9683109a9ed7bee2980312f550910
Author: Megan <mchang01@wesleyan.edu>
Date:   Thu Jun 28 16:31:59 2012 -0400

    Added integration tests for proc.py.
    
    In the cases of test_get_memory_usage(), test_get_stats(),
    test_get_connections(), there were issues of doing direct tests. First,
    the only way to have values to check against is with code written
    in proc.py, which would mean we are checking the code against itself.
    Second, the results of these functions change frequently, so by the time
    the functions are called, their results will differ from the expected
    values. To get around this, we simply checked that these values were
    nonzero.

commit 10ae1304d02c390237d865c5fc439992249bc9d0
Author: Megan <mchang01@wesleyan.edu>
Date:   Tue Jun 26 17:20:37 2012 -0400

    Made additional changes regarding sphinx documentation. This completes Code Review 1.

commit 48fcbc69e4ed13a44ff160ba35997a46487f997f
Author: Megan <mchang01@wesleyan.edu>
Date:   Tue Jun 26 17:05:35 2012 -0400

    First code review changes.

commit e94e48ecf3209273b6de13bd405fa87d3f1203b4
Author: Megan <mchang01@wesleyan.edu>
Date:   Mon Jun 25 18:05:44 2012 -0400

    First complete draft of proc unit testing code.
    
    Most mocked functions utilize mock_fn(), which checks for expected input
    and either returns the appropriate output given the mocking or lets the
    original function be called.  This is necessary in cases where outside
    code calls these functions and the mocked version is not able to supply
    them with the needed output.  Private methods in the stem/util/proc.py
    file are not tested, nor is is_available().  This decision was based on
    the conventions set in system.py testing code.

commit 3fd005a27ca35bf802d3917c992d2626534e87c9
Author: Megan <mchang01@wesleyan.edu>
Date:   Thu Jun 14 16:17:32 2012 -0400

    Created proc.py testing code and imported unit tests to run_tests.py.
    
    Wrote mock_get_lines to mock the proc module's _get_lines
    function. Completed unit testing code for proc utilities such as
    get_system_start_time, get_physical_memory, and get_memory_usage.
    The test_get_memory_usage function was written to test the functionality
    of mock_get_lines as it is the only function that takes in multiple
    line prefixes.

commit c35c8a6cbb3659841f79159512c24a08a5e1d622
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jun 29 09:59:16 2012 -0700

    Handling version 1.1 of bridge extrainfo descriptors
    
    Version 1.1 of the bridge extrainfo descriptors adds a new 'transport' field.
    I'm not entirely sure yet of its attributes, but recognizing the new descriptor
    version and parsing the field.
    
    I'm also expanding the tests a bit so unit tests cover bridge extrainfo
    descriptors, and parsing the metrics header line rather than just doing string
    matches.
    
    This is to address...
    https://trac.torproject.org/6257

commit 86baf8e562c0486d2ac2e2be893480eb149742b4
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Sat Jun 23 19:16:38 2012 +0530

    Remove single letter variable
    
    According to the guidelines of our benevolent dictator we
    must abolish the use of single letter variables. The sole
    infiltrator has been abolished. Once again, there is peace in the land
    of stem.

commit 54a013552266540d5029af2ff08db2fed1c1dc92
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 24 14:14:49 2012 -0700

    Stopping tor initialization timeout when done
    
    Ok, I'm kinda surprised that we haven't tripped over this before. We never
    terminate the tor initialization timeout on either or success or failure,
    causing us to arbitrarily kill our process after 90 seconds.
    
    This bug probably survived so long becuase our integ tests take less time than
    that to complete, and then spawns a new tor instance resetting the timer again.

commit 4b79530c47abdeebb33bc7c06a6a6bda4e2815b6
Merge: 09acb5e 1c01521
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 24 18:20:59 2012 -0700

    Python 2.5 and OSX compatability fixes

commit 1c0152185a9ff27a49f6491bcd2ea06f64158820
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 23 14:36:08 2012 -0700

    OSX sucks at terminating processes
    
    Our os.kill() call evidently doesn't kill tor right away on OSX, causing tests
    to disrupt each other (port conflicts).

commit 04a6225ffdccc81a431044d05f4f411c7e5a88f5
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 23 14:18:32 2012 -0700

    Inverted check when falling back to kill a process
    
    Oops, we're supposed to use os.kill() as a fallback if we *aren't* on windows.

commit e625e84b003420908d410ab10f51fdd78a1ccf17
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 23 12:16:53 2012 -0700

    Respecting reader stop() calls in mid-directory
    
    If we, say, attempted to read a directory with a million files then the
    reader's stop() method won't be respected until we've finished with that
    directory. That's stupid. Checking the _is_stopped Event after processing
    each file.
    
    Python lacks targeted breaks, so I needed to refactor the code a bit to
    accomidate this.

commit 4e02850db89006f256290e8bc7bea7169c6cd57c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 23 11:55:57 2012 -0700

    Missing kill() fallback when ending integ tests
    
    As an earlier commit mentioned, Popen.kill() is unavailable on python 2.5, and
    its fallback (os.kill()) is unavailable on Windows. Accounting for this in the
    integ test runner.

commit 2fc1c420818bff19437a2b74713223a410d9898e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 23 11:47:24 2012 -0700

    get_pid_by_port() unavailable on Mac OSX
    
    Tests for get_pid_by_port() are failing on Mac OSX 10.5.8 because sockstat is
    unavailable and lsof neither provides ports nor accepts the flags we need.
    Noting this in the pydocs/comments, and skipping its integ tests when on a mac.

commit a5f6006b3e1d072e1cd9524aa57408d707ff3ba0
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jun 22 09:27:03 2012 -0700

    Respecting custom tor commands for process integ tests
    
    The process integ tests were attempting to run 'tor', which naturally doesn't
    work if it isn't in your path. Using the command that we're using for our main
    test instance instead since that is sure to work.

commit ac34b8151c5b46369578922ba4f90fdce15128b3
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jun 22 09:18:39 2012 -0700

    Missed another tarfile 'with' declaration
    
    Missed that we were using the 'with' keyword for a tarfile in the integ tests.
    Using a try/catch instead so this'll work on python 2.5.

commit 9d54db62a7b918c3bc6a852275adaf70dfdfef44
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jun 22 09:14:20 2012 -0700

    Subprocess lacks a kill() method on python 2.5
    
    The Popen.kill() ad Popen.terminate() methods were first added in python 2.6...
    http://docs.python.org/library/subprocess.html#subprocess.Popen.kill
    
    We can work around this with os.kill() but that only works on *nix...
    http://stackoverflow.com/questions/552423/use-python-2-6-subprocess-module-in-python-2-5/552510#552510

commit e51fea598dced5e3c4a05c4cd0ad509beed530fe
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jun 21 09:45:14 2012 -0700

    Avoiding the 'with' keyword for tarfile
    
    The tarfile's __exit__() method was added sometime after python 2.5, causing it
    to produce a stacktrace if used via the 'with' keyword. Pity, yet another hack
    to accomidate a six-year-old version of python...

commit 60b0e8ba06a03e5ae1664959c2f9e1552f224b39
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jun 21 09:40:00 2012 -0700

    Using theading's currrentThead() rather than current_thread()
    
    The threading module's current_thead() function isn't availalbe in python 2.5.
    It isn't documented, but there seems to have been a mass aliasing in 2.6 to the
    underscore convention so using that instead.

commit 1098d9ad35dc0c04dcced034b1e564706c543c96
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jun 21 09:31:50 2012 -0700

    Python 2.5 doesn't have is_alive() method for Threads
    
    Though it's not documented [1], the Thread class' isAlive() method was first
    aliased to is_alive() in python 2.6. Also filed a ticket for the documentation
    bug [2].
    
    [1] http://docs.python.org/library/threading.html#threading.Thread.is_alive
    [2] http://bugs.python.org/issue15126

commit 9a8ac67f57fc203eb20dddea3aa09ba49bfe7a97
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 20 10:35:17 2012 -0700

    Python 2.5 doesn't have is_set() method for Events
    
    In python 2.5 the threading module used camel case method names. They also
    accepted the underscore convention (the 'official' style for python) in python
    2.6, but until we drop 2.5 compatability we need to use the camel case
    versions.

commit cec8ecfcf175af55d9503d45e169cd26f9664279
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 19 09:47:33 2012 -0700

    Integ test relied on tor being in path
    
    I thought that it was a safe assumption that tor was in our path since the
    integ tests will only run if it can start tor. However, the tests have a
    '--tor <path to tor>' argument which means that the tests may be running
    without our path including the tor binary (or even having the binary called
    'tor').
    
    Changing the test to look for 'ls' or 'dir' instead since every platform that I
    can think of has that. Caught by Karsten.

commit 09acb5e43ba6afd265bcf79e7534dbf369fa227b
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jun 22 09:56:17 2012 -0700

    Checking for windows newlines
    
    A while back I added windows/old-style-mac newline checking to the whitespace
    checker, but got distracted before committing. Reviving it from the stash. This
    should do the trick, but haven't checked it against windows files (guess we'll
    get confirmation that it works or doesn't when we have one submitted).

commit f724692fe512a2eaf2b2c4bfd491513f48bfca23
Merge: a4243fd 6f626f3
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jun 21 08:45:09 2012 -0700

    Merging Beck's windows test compatibility fixes

commit 6f626f3f1a43368f13b2679f3eab6322f26788c0
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jun 21 08:35:32 2012 -0700

    Minor changes for windows test compatibility fixes
    
    Just a few minor tweaks...
    
    * The 'path' and 'relative_path' were starting to get confused, so making the
      former the pristine input argument and having the later be what we modify.
    
    * A recent change removed direct usage of 'skipTest' since it breaks us on
      python 2.5 and 2.6...
      https://gitweb.torproject.org/stem.git/commitdiff/ed0db8f
    
    * Skipping the 'is multiple tor instances' check rather than always skipping
      the system integ tests.
    
    * run_tests.py somehow lost its executable permissions in 2b96648 (...?)

commit 03780a98ab131fb5ac6cf1bac192d1d7ec9f238e
Author: Beck <csybeck@gmail.com>
Date:   Wed Jun 20 01:41:37 2012 +0800

    Fixed a bug in stem.util.system.expand_path that may introduce a trailing slash.

commit be611a6ad6c7054c35e02571087950e133839592
Author: Beck Chen <csybeck@gmail.com>
Date:   Wed Jun 20 00:24:15 2012 +0800

    Fixed test_skip_listener_unrecognized_type on windows.
    
    This test case is included in stem.test.integ.descriptor.reader. Now it
    allows both "image/png" and "image/x-png" types.

commit 73277fc0dc9709b56fd69fa0d0c0dec2effcb2c9
Author: Beck Chen <csybeck@gmail.com>
Date:   Fri Jun 15 12:18:03 2012 +0800

    Skip two test cases in test.integ.descriptor.reader on windows.
    
    The two test cases are test_load_processed_files_permissions and
    test_skip_listener_read_failure. They both call os.chmod() to change
    permissions of the test file, but on windows, you can only set the
    file's read-only flag with it and all other bits are ignored. This
    would cause tearDown() to fail since test files are read-only and
    can't be removed, thus fail all the test cases.

commit 232858d8607270e00d42316e8655c70e0221c87e
Author: Beck Chen <csybeck@gmail.com>
Date:   Fri Jun 15 02:17:00 2012 +0800

    Implement stem.util.system.expand_path() on windows.

commit 67be8c6435e6d08b71327687ffa1097d01345bf2
Author: Beck Chen <csybeck@gmail.com>
Date:   Fri Jun 15 01:26:22 2012 +0800

    Skip test_stop() in test.integ.descriptor.reader on Windows.

commit 5c18901aa94ada77f43c1f67d91f4ee300ccea67
Author: Beck Chen <csybeck@gmail.com>
Date:   Thu Jun 14 19:32:36 2012 +0800

    Skip integ tests for util.system if pgrep is unavailable.
    
    Since pgrep is used in setUp() of test.integ.util.system.TestSystem, all
    test cases would fail if pgrep is unavailable in the current platform
    (e.g. windows). We'll simply skip them for now.

commit 57904d4540dc9e988a431986695e6e1f2f5cf8fa
Author: Beck Chen <csybeck@gmail.com>
Date:   Thu Jun 14 19:11:09 2012 +0800

    Uses stem.util.system.is_windows() for platform checking in run_tests.py

commit 417d7f8117cf40378901876d0e7b324f3f694a18
Author: Beck Chen <csybeck@gmail.com>
Date:   Wed Jun 13 01:17:10 2012 +0800

    Disable ANSI escape sequences in windows

commit a4243fd09384fcbe37f076cf4ec6163c80f00839
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 20 09:37:40 2012 -0700

    Adding pydocs for prior mocking change

commit 66376f5a6ebc35fd553506c226342ec8777fc9b5
Author: Erik <eislo@wesleyan.edu>
Date:   Mon Jun 18 17:01:53 2012 -0400

    Changed mocking.mock() to accomodate modules that link to other modules.
    
    We ran into a problem mocking os.readlink as in certain unix environments
    os links to an outside module, posix.  This resulted in the __dict__ of
    readlink remaining unchanged, and thus not mocking the realink function
    (which in our environment resides in the posix module).  To patch this,
    we added a third argument to mock() that can be used to explicitly
    specify the target_module.  A default value of None is provided so all
    prior calls on this method are compatible.

commit ed0db8f0defa6d94e62b20c7a2cc335f3801d91a
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jun 18 09:52:51 2012 -0700

    TestCase's skipTest method unavailable before python 2.7
    
    Blarg! The unit testing's skipTest method was a new addition in python 2.7.
    This is especially pesky since skipTest alters the control flow of the tests
    (raising an exception so we can place it in setUp methods).
    
    Options are to either monkey patch support into our testing module (ick!) or
    add a fallback everywhere we use skipTest. This change adds a best-effort skip
    method to the runner, then leaves it to the caller to return out of the method
    (reporting success for skipped tests).
    
    Sucks, but best that we can do until we drop *both* python 2.5 and 2.6
    compatability, which probably won't happen until we finally move to the 3.x
    series.

commit 385b05b3df8e07b4cd589a49c0f348a9541726bd
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jun 14 09:17:25 2012 -0700

    Patched mocking.mock() to handle standard library functions
    
    Although standard library functions have the same type as built-in functions,
    they cannot be mocked using the same setattr approach.  To fix this, we patched
    mocking.mock() to first look for built-in functions, then to handle
    non-built-in functions with the same type as built-in functions. This will
    catch standard library functions and mock them appropriately.
    
    This change is an adaptation of one submitted by Erik Islo and Megan Chang.

commit 122d0a6aa57f21fe2ef7d41c6fd14736074345c6
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 13 09:33:20 2012 -0700

    Tuples lack an index method in python 2.5
    
    Yet another python 2.5 compatability hack.

commit 4581c355d64f3241d649256d4e14d839aedc9ea5
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 13 09:27:47 2012 -0700

    Adding todo notes around the stupid concessions to 2.5
    
    I hate dirtying our code for ye old python 2.5. It's ancient and that fact
    isn't getting rosier very fast. However, a lot of people still use it so we
    should still support it for now. This won't always be the case so including
    TODO notes to remove these icky bits when we drop compatability for it.

commit d2eef839544eaf1ed55a85df2e4fd466a175fc40
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 13 09:24:34 2012 -0700

    The followlinks arg of os.walk needs python 2.6
    
    We were using the 'followlinks' argument two places: in the descriptor reader
    and whitespace checker. For the former I'm noting that 2.6 is a requirement for
    the argument and coding around it to work on 2.5. For the later I don't really
    care if the whitespace checker follows links so I'm simply dropping the arg.

commit b84a75638942e61a58b2af8edd97448ea72857d7
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 13 09:20:34 2012 -0700

    Adding module to check for requirements
    
    Stem requires at least python 2.5, and beyond that our python version and
    modules does impact what functionality we have available (yay). Adding a module
    for tracking what we have available.

commit e09e6b00a4b603db6342d27a11e3073acf8c74c6
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 13 08:46:21 2012 -0700

    Tor executable shouldn't be needed for unit tests
    
    We only need a tor variable when running the integ tests. Dropping the check
    that it exists in our path if we're just running the unit tests.

commit 98b3a8dfb10944ff3ec7b41a3b1f39297173bf26
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 13 08:39:57 2012 -0700

    Dumbing down syntax for python 2.5
    
    Evidently even with the 'with' keyword's future import it still chokes on comma
    delimited entries (instead you need to do nested 'with' blocks). Ick. Oh well,
    if that's all we need to maintain 2.5 compatability then I'll be happy.
    
    It's a pity we can't tell python 'run in 2.5 compatability mode' to check for
    these things. I doubt that we'll remember to run against a 2.5 instance to keep
    stem working there. Oh well, guess this is a problem that'll need to be
    addressed when figuring out a release procedure.

commit 408dbe96e88767ca55688d196287e41c57345219
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 12 09:58:08 2012 -0700

    Adding --test argument to selectively run tests
    
    Adding an argument that'll only run tests matching the given module prefix.
    This is to make it faster to write new tests (so we don't need to wait a minute
    for the full integ suite to run each time).
    
    Also adding examples to the '--help' output and rearranging the options so
    '--config' is at the end (it isn't especially useful and may be dropped in the
    future.

commit b8f4d3c75f0d4c085dcba02f47f4808c301e235a
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 12 08:49:01 2012 -0700

    Couple fixes for prior controller changes
    
    Two minor bugs...
    - our version parsing was broken for versions without a space in them
    - we were quoting the runner password constant rather than using it, breaking
      the RUN_PASSWORD target
    
    Also some minor whitespace tweaks and moving imports to the start of the file
    (we're only doing method level imports if doing otherwise would break things).
    We might change our import style later to resolve our circular imports, but
    that'll be a project-wide change.

commit 843f5568d42d4568459d0a4597e078380d5fa76a
Author: Beck <csybeck@gmail.com>
Date:   Tue Jun 12 16:41:12 2012 +0800

    Fixed indentation errors.

commit c4c242f589b39519ec7d71841fd4679ecac498af
Author: Beck <csybeck@gmail.com>
Date:   Tue Jun 12 16:03:46 2012 +0800

    Added integ tests for convenience methods.

commit 1d8ee15ef127073c384021c81e0ea07951d2c2fe
Author: Beck <csybeck@gmail.com>
Date:   Fri Jun 8 00:31:56 2012 +0800

    Move all convenience functions to Controller class, and minor changes on them

commit c4b5917be15fdfce5921b2dd704f025be59ae97d
Author: Beck <csybeck@gmail.com>
Date:   Thu Jun 7 10:45:57 2012 +0800

    Added convenient methods in BaseController and Controller.

commit fc327f29f66a84850ab89a32f3dd358756797029
Author: Beck <csybeck@gmail.com>
Date:   Fri Jun 8 00:31:56 2012 +0800

    Move all convenience functions to Controller class, and minor changes on them

commit de09fd34f4246b3b0323998de3829962bcad0dc7
Author: Beck <csybeck@gmail.com>
Date:   Thu Jun 7 10:45:57 2012 +0800

    Added convenient methods in BaseController and Controller.

commit c1450e440e388295a329e84b3e935bf86f88e673
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jun 11 19:01:34 2012 -0700

    Whitespace checker wasn't checking non-tests
    
    The whitespace checker was being ran with its default target, which is the
    directory of that file (in this case, test). Explicitely running over
    everything else, and fixing the issues that it found.

commit c52042c44b5a157dceb6013743fdea40bbc3f52c
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jun 11 10:01:17 2012 -0700

    Adding future 'with' import and checking for it
    
    Stem aims for python 2.5 compatability, but in that version the 'with' keyword
    wasn't available by default and would only function with a 'future' import.
    This has the possability of being a continuing pain in our ass so adding this
    as a check to the check_whitespace module. Admittedly this isn't a whitespace
    issue, but that module is just so damn conveninent for lintian style issues
    too...
    
    While doing this I discovered that we aren't running the checker over the stem
    module (just the tests, it seems), so something else to fix...

commit 65351e9eea2a30e81632e045c0b2825d0573719c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 10 15:16:57 2012 -0700

    Converting test.* contents to reStructuredText
    
    The test contents aren't currently included in the stem documentation, since
    they aren't generally of interest to users. However, might as well do its
    documentation as reStructuredText for consistency's sake.
    
    I'll probably include just the framework utilities with our documenntation
    later to help people hacking on stem.

commit 9654beb5acb7ad0e8e90bbdf7b9e186ccd946654
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 10 14:24:33 2012 -0700

    Adding extra-info digest() method
    
    Adding digest() methods for extra-info descriptors. I'm also swapping the
    server descriptor counterparts to provide a hex encoded digest as per Karsten's
    suggestion.

commit 860cd87212d6caca7f23423940ed1807cf3f0f84
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 10 13:58:23 2012 -0700

    Integ test with bridge extra-info content
    
    Adding a bridge extra-info descriptor from metrics. I'm currently being a bit
    lazy and only parsing some of the mapping fields. The rest should be covered
    by unit tests, so I'll only add the rest if this ends up bitting us.

commit 737018b7ee3da5b457689404376a7937de5a016f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 10 13:33:21 2012 -0700

    Making relay-signature field only apply to relay extra-info
    
    Parsing and handling the relay-signature field as only an attribute of
    extra-info descriptors for relays.

commit 7947d69d4e20ca402e9d1d35fe6e62012d5e61fe
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 10 12:31:14 2012 -0700

    Separating relay and bridge extra-info descriptors
    
    Creating separate classes for relay and bridge extra-info descriptors. At
    present they're identical to their common parent (ExtraInfoDescriptor), adding
    their differences next.

commit 560923cb7b572d02046c6ca2bd5eb4502fa591b3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 9 18:18:00 2012 -0700

    Revisions to change for SAFECOOKIE support
    
    The SAFECOOKIE change is pretty big so, unsurprisingly, it has needed quite a
    few changes. Many are stylistic nitpicks, though this also includes several
    more substantial changes including...
    
    * The SAFECOOKIE AuthMethod was added, but the ProtocolInfoResponse's
      _parse_message didn't recognize it causing it to still be listed as an
      unknown authentication method. Also, we called send/recv on the controller in
      authenticate_safecookie() at one point rather than using _msg, causing it to
      break when called with a Controller. I'm a tad puzzled by these... were the
      integ tests ever ran with the RUN_ALL target?
    
    * Slightly reducing the exhaustiveness of the authentication unit test in the
      name of runtime. The test exercises each combination of exceptions for each
      exception type which led to a combination explosion when we added SAFECOOKIE
      to the mix (raising the unit testing runtime from around two to twelve
      seconds). Made some minor tweaks to bring the runtime back down without
      really impacting what we test much.
    
    * Changing the CookieAuthFailed attribute that indicates if it arose from
      SAFECOOKIE or COOKIE auth to be a boolean. Providing the AuthMethod would
      only make sense if we added even more variants of cookie authentication
      (possible, but not anywhere on the radar).
    
    * Added AUTH_SAFECOOKIE to 'stem.version.Requirement'. This turned out to need
      some fairly big changes to how we check for version requirements (see prior
      commit).
    
    * Few issues with the integ tests, including calling 'remove' on a tuple and a
      bug where it would accidently remove all auth methods from the protocolinfo
      response before calling authenticate().
    
    * We only removed AuthMethod.COOKIE form a protocolinfo response's auth_methods
      when the cookie file was missing.
    
    * Few unused imports and lots 'o whitespace issues. I should probably improve
      the whitespace checker a bit.
    
    * Toned down the 'warning, we're under attack!' errors. The whole point of
      safecookie auth is simply to avoid providing the raw cookie contents to the
      thing we think is tor. Nothing about this handshake should be consitered to
      be an 'attack', nonce mismatches are more likely to just indicate some sort
      of bug.
    
    * No tor reply can be empty (if it was empty then we... well, wouldn't have
      read anything from the socket). Hence the ControlMessage can't be empty.
      It'll be a pita to check for this in every reply, so checking that as part of
      the ControlMessage constructor.
    
    * Added helper function to check if a value is hex digits. We do this a lot, so
      no need to repeat the regex everywhere.
    
    * Looks like we could misidentify IncorrectCookieValue verses
      CookieAuthRejected exceptions when we had both cookie and password auth.
    
    * I was a bit confused about what the AUTHCHALLENGE unit tests for an invalid
      response were attempting to exercise, so replaced with a shorter and simpler
      test for missing components.

commit 7ecd7c6a63baf08b638a378a7f4c237ec7ceeb84
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 9 14:20:56 2012 -0700

    Expanding capabilites of tor version comparison
    
    Requirements were simple tor versions which was fine for simple 'greater than'
    comparisons, but broke down for slightly more complicated requirements such
    as 'at least version X in the 0.2.2 series or version Y in the 0.2.3'.
    
    Introducing a 'meets_requirements' that will allow us to compare with arbitrary
    rules. For now we're just comparing with Versions or VersionRequirements, but
    could be expanded later if we need to.

commit 682cbb4a2b36b0d226ea2a7e8d9bb527fdb2e28e
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Tue May 8 01:05:20 2012 +0530

    Implement Safecookie support in Stem

commit 63c2ddb95e4c96e509e030240df12266012ce844
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 6 10:09:21 2012 -0700

    Descriptor typo corrections
    
    Fixes by Karsten

commit ee6e475a347277ff6a387107219fa5ba79a7ae99
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 6 09:53:03 2012 -0700

    Targeting reader test at /usr rather than root dir
    
    I've been running the reader's test_stop() integ test by running over the root
    direcctory because I simply needed something 'big' to keep the test occupied
    for a second or two. This turned out to inadvertantly fuzz the reader which is
    sorta a good thing since it discovered that block devices and encrypted
    partitions can be slow to read, causing stop() to block for a bit.
    
    However, we don't want to run over people's private stuff, and slow reads on
    encrypted partitions probably isn't a problem that we want to solve anyway, so
    running over '/usr' instead.

commit da1e0a75082efef9e24f86fbba68e6c47116619c
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 6 09:23:11 2012 -0700

    Correcting incorrect variable in logging
    
    Issue caught by Karsten.

commit ebed653be563f2e16c384d9d3b235ad2ecb11566
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jun 6 08:47:54 2012 -0700

    Skipping non-regular files when the reader crawls
    
    When the reader encounters block devices it can get stuck, causing the integ
    test for stop() to fail. Caught by nickm.

commit 38adeda918ffc1367ec8eb8dc0002bd737ad95c6
Merge: 2a21ec5 8badba1
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 21:28:53 2012 -0700

    Merging changes for sphinx documentation

commit 8badba152f5d9b2a83a9b9893c328cd23dd219a2
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 21:22:49 2012 -0700

    Targeting contents at top of modules
    
    For modules like stem.response and stem.util I was linking to the first anchor,
    which isn't the behavor I wanted (skips the title at the top, which is a nicer
    landing). I resisted making this change earlier because it means that those
    links have different styling from those above, but now that there's more of
    them I'm happy with it this way.

commit 4819e2fc97f76424a3dead741d327be711f53a8d
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 21:15:37 2012 -0700

    Converting stem.descriptor.* to reStructuredText
    
    Fingers so sore...

commit e13ea63c09e766d15c36983c9b37f6adde1a2bde
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 20:14:39 2012 -0700

    Removing double backticks from None documentation
    
    Double backticks should be monospaced, but with the haiku theme it's an
    inverted background with is far more noticeable. I didn't mean to accent 'None'
    this much so dropping the formatting from it.

commit 12af869b1f669117de736c86f3e094ad2e467be4
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 20:08:57 2012 -0700

    Converting stem.util.tor_tools to reStructuredText

commit fa60694958ed41a323a099e844ce92fb48faa373
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 20:06:31 2012 -0700

    Converting stem.util.term to reStructuredText

commit 3319642a609cd4995724877b90af76cd44067085
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 20:02:20 2012 -0700

    Converting stem.util.system to reStructuredText

commit 57e11a1345ed5547424b03fc5f7d9e663c4661ea
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 19:53:00 2012 -0700

    Converting stem.util.proc to reStructuredText

commit 1e04a08e75ae73f204c6ba3d4f8eef33104973f5
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 19:40:57 2012 -0700

    Converting stem.util.log to reStructuredText

commit d5ef02b04eccebef5d5bbf89003431c0d41f1401
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 19:33:10 2012 -0700

    Converting stem.util.enum to reStructuredText

commit c1392cd8b3322be4b39980dea54d5b224ed7062f
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 19:26:54 2012 -0700

    Converting stem.util.connection to reStructuredText

commit 1b23a2da711b202a9fba24f47c173f09024efa34
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 19:22:52 2012 -0700

    Converting stem.util.conf to reStructuredText

commit 68dc6af5c74cd8b733b07542c88f5af96bb146be
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 09:31:55 2012 -0700

    Converting stem.response to reStructuredText

commit c350257b05455509ab99c9019747e63553a668d5
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 08:25:48 2012 -0700

    Converting stem.socket to reStructuredText

commit cbcf257ee91f78a58984a2335814a589c51f14e9
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jun 5 07:54:20 2012 -0700

    Converting stem.control to reStructuredText

commit 1cfad693de89bc4769605f21d52ef9ba4f38205b
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jun 4 10:04:45 2012 -0700

    Adding module overviews back to docs
    
    One bit of documentation that I've found very useful, and sphinx lacks, is a
    brief summary of a module's contents. Hence adding that back to the headers.

commit c03ebb23ae39f9c9b68a1471dbf3debb30c399a7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 3 21:26:21 2012 -0700

    Defaulting connect_* to provide a Controller
    
    As a TODO comment mentioned, the connect_* convenience functions were supposed
    to provide a higher level controller class when one was available. It now does.

commit f46b5c7c0e543d90f26155e472a1658827c5f8fe
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 3 21:23:49 2012 -0700

    Converting stem.connection to reStructuredText

commit 1ccf6be7dec076afdf4c4a4150c38b5065a8f17e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 3 16:28:07 2012 -0700

    Switching to the haiku theme
    
    The haiku theme is lighter, and lacks a left-hand bar which makes the content
    far less cramped. All in all, looks far nicer for the content that we have so
    far.

commit 438dc79a512879473499211988a86e08099720de
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 3 16:21:26 2012 -0700

    Missed linking a funcion

commit ecc0e86d46ac2db77d45d46af5ac50168befa6c9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 3 16:18:00 2012 -0700

    Converting stem.version to reStructuredText

commit 6651e8304a3cee2ffed344371c830a6014853201
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 3 15:38:38 2012 -0700

    Converting stem.process to reStructuredText

commit 241359c043c3ffb2d2d15d24cacad6c9d7fd0ae7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jun 3 13:13:49 2012 -0700

    Removing autogenerated rst files via clean target
    
    The html target creates some rst files, so removing them when 'make clean' is
    ran.

commit a3a7b2055144ae4ccb06e464510df211fc4fb43a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 2 19:19:24 2012 -0700

    Skip recreating unchanged docs
    
    Greatly decreasing the time it takes to generate documentation when pydocs
    haven't changed. Sphinx is smart enough to avoid running against files whos
    last-modified timestamp hasn't changed, but not smart enough to check the hash
    of the content. Hence using rsync to avoid modifying our pydoc derived content
    unless the pydocs have changed.

commit 58566d6b2e20899c82206dba50d2aebfa008a96c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 2 18:55:25 2012 -0700

    Generating sphinx docs for all stem modules
    
    Using the sphinx-apidoc command to generate documentation for all stem modules.
    I'm doing this via the make file to simplify the documentation creation process
    (you still just need to run 'make html'). Also removing the windows make.bat
    file since I'm not gonna be keeping it up to date. We can resurrect it if
    someone volunteers to maintain it.

commit af83350dcf62f819d8e84c6e34051dc3ff509add
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 2 13:35:12 2012 -0700

    Adding basic stem intro to sphinx index

commit bad7ae65902f648c3127de66a35a95223b6582a6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 2 11:41:54 2012 -0700

    Configuring sphinx documentation attributes
    
    Setting attributes and toggling attributes for the autogenerated sphinx
    documentation. This seems like a pretty slick tool, though I'm still puzzling
    out how to make it index all of stem. It seems to only register modules that I
    explicitely tell it about - I'm probably missing something...
    
    Also still deciding between the default and haiku themes. Ruled out the others.

commit 5c9579626a131e8bd17c283e0904525e30fe4488
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jun 2 11:06:42 2012 -0700

    Adding module metadata (author, version, etc)

commit ff5257f2e848ed84fa3e8e23b60149d464262c0e
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu May 31 10:01:30 2012 -0700

    Ignoring sphinx build output
    
    Build artifacts from sphinx are placed in 'docs/_build/*', so having git ignore
    it.

commit feeac963d1383b9eb5a08251dbd329ff64f1b846
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu May 31 09:57:51 2012 -0700

    Initial sphinx autogenerated content
    
    Results right after running 'sphinx-quickstart'

commit 2a21ec5848df9d1d5a520abbb6d5927a804acd09
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 28 17:07:57 2012 -0700

    Moving ControlMessage and ControlLine into stem.response
    
    The ControlMessage and the ControlLine instances in it are the... well,
    messages that we get from tor. It belongs in stem.response so moving it there.
    
    Functionally this is fine, and I'm happy with the tests in a functional fashion
    as well. However, all this splitting and refactoring has made the tests a mess
    in terms of what the test functions belong to (some test.unit.response tests
    are checking stem.connection functionality, for instance). More cleanup work to
    do there...

commit d3fe6a1ed751a60b395f80aa7b4663ab6f78d230
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 28 16:38:30 2012 -0700

    Cleaning up response classes
    
    General refactoring of the response classes.

commit be06640765510b42912f074ebad6acb7cf4d0bac
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 28 15:52:18 2012 -0700

    Renaming getinfo 'values' attribute to 'entries'
    
    The name 'values' has bugged me for a while because it's a dictionary, so users
    might end up typing things like "my_getinfo_response.values.values()". Renaming
    it to something that's still generic but without these issues.

commit 6015799baf9c0e6d3a618c948c32c27b2d7e1fc3
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 28 15:48:32 2012 -0700

    Being more anal about 'OK' status checking
    
    Refusing to parse messages that lack a 250 response on all lines, and checking
    for a 'OK' status on the last line (and only the last line).

commit 152059f7d687c93326e039c954bbfa72a9356323
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 28 15:43:35 2012 -0700

    Adding an is_ok() method to controller responses
    
    Method added by Ravi during his safecookie work. Stealing it a little early
    since I want it for refactoring that I'm doing.

commit 72aed261dceb560d05563fab1af4e6f549e2bdfd
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 28 15:23:52 2012 -0700

    Single retry for socket connection
    
    I just had integ tests fail due to an interrupt while trying to connect to the
    tor control socket. This is the first time that I've seen it, so this isn't
    much of a concern, but connecting to a socket is idempotent so we can do with
    retrying it once if we fail.

commit 1d7ab654ec473a5e54ea3253f82abdcadc93158f
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 28 15:12:16 2012 -0700

    Moving tor response classes into their own module
    
    As we add more response classes it'll be messy to sprinkle them all about the
    codebase. Making a single 'stem.response' module that'll contain them all.
    These could probably do with some more love so I'll next see if I can make them
    any tidier.

commit 717c5aaecc8e61cee5ca8f1bae80b3eade5f6985
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 27 20:29:51 2012 -0700

    Implementing Controller.get_info
    
    Implementation and testing (both unit and a little integ) for GETINFO queries.
    There's still several todo notes to clean up, but the method itself is done.

commit 045a5ec8774074eb92066571dfc661c4098bd77a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 27 14:35:15 2012 -0700

    Adding Controller class
    
    Making a class for the general controller. As a todo comment mentiond, moving
    the from_port/from_socket_file helper functions there and revising the tests to
    reflect that.

commit 683078f749ca6627f8ff65c4562697b6e1034689
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri May 25 09:27:07 2012 -0700

    Minor error in logged message
    
    s/server/descriptor

commit 40a971105ff162d0caa9e0a1c5fa3c0f7d3c2b72
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri May 25 08:59:03 2012 -0700

    Validate that signing key hash matches fingerprint
    
    When parsing server descriptors checking that their fingerprints match a hash
    of their signing key as part of validation. This requires the rsa module which
    is neither built in, nor is it packaged for debian distros. Installation of it
    is easy, but requires pip or easy-install. Instructions are available at...
    http://stuvel.eu/files/python-rsa-doc/installation.html
    
    It looks like python's builtin crypto module might be capable of doing this as
    well...
    http://stackoverflow.com/questions/5000434/python-rsa-library
    https://www.dlitz.net/software/pycrypto/api/current/Crypto.PublicKey.RSA._RSAobj-class.html#exportKey
    
    However, those instructions include usage of a private member and the function
    that they suggest dosn't exist on my system (python 2.7.1), so I feel pretty
    justified in saying "the pycrypto builtin is crap for this use case, and
    patches welcome if someone can figure out how to make it work".
    
    All credit for this patch goes to Beck, who's been diving into the descriptor
    crypto on...
    https://trac.torproject.org/projects/tor/ticket/5810

commit f7fb726cc3dea8bfd294833b151117858490802d
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 23 10:00:04 2012 -0700

    Removing expectation that bridge nicknames are scrubbed
    
    Karsten plans to stop scrubbing bridge descriptor nicknames, so removing that
    expectation from the 'is_scrubbed()' method.

commit e7e03d2f61d6dcc7bc5e5ad4dee91c37a814ee16
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 23 09:52:23 2012 -0700

    Supporting bridge descriptor's router-digest field
    
    Scrubbed bridge server and extra-info descriptors will soon have a
    'router-digest' line which contains the server digest. This is a value that we
    can't calculate for ourselves because, of course, the descriptor's contents
    have been modified.
    
    I'm flagging this as being a required field. This does not yet parse it in
    extra-info descriptors since it neither has a digest() method nor did I realize
    that bridge extra-info descriptors were special.

commit f1d58b21fd5fe7dcd8b5330627f7521ac86ffbc0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 20 15:24:22 2012 -0700

    Citing ticket 4817 in related pydocs
    
    The pydocs warn users of authenticate_cookie and authenticate_password that we
    can't reliably differentiate certain failures. Citing the related ticket.

commit ce07df9962bc464c338cbee0753e40169f801505
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 20 15:11:55 2012 -0700

    Stating runtime when launch_tor timeout test fails
    
    Just had a transient testing failure from test_launch_tor_with_timeout. There's
    nothing to go on since we don't even know how long the test took, so providing
    that information on failure. Might need to mess with the bounds a bit if this
    happens again.

commit 688f321287f5e9c53451dd91608dc3aa5088a621
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 20 15:05:10 2012 -0700

    Disabling launch_tor timeout when on windows
    
    On windows the python signal module lacks a SIGALRM attribute, breaking
    launch_tor. Disabling timeouts on windows to avoid this issue - ideally we
    should find an alternative method for timeing out the function on windows but
    that can wait.
    
    Caught by pythonirc101. Issue is discussed on...
    https://trac.torproject.org/projects/tor/ticket/5783

commit 4d6c8c8943d9540d5fcc6c39c503ca523e4683d3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 20 14:49:31 2012 -0700

    Adding separate funtion for launching tor with tmp torrc
    
    Like Vidalia, adding a function that creates a torrc, launches tor with it,
    then removes it from disk. This is preferable to using the commandline
    arguments since the those shouldn't be used for substantial data nor the hashed
    password.
    
    This replaces launch_tor()'s options argument, though I added a more genral
    args replacement that can be used to do the same if the user wants.

commit 301401360337e4c02a5fd5e4e8520cc1ecf88633
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri May 18 10:25:49 2012 -0700

    Utilising metrics descriptor type annotations
    
    Using the @type annotations added by Karsten to the metrics descriptors as per
    ticket 5651.

commit aed91aa634b9f960e87296b9e631a2c6aa05f2c0
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri May 18 08:33:13 2012 -0700

    Removing GETINFO version queries from integ tests
    
    Sathyanarayanan spotted a couple points in our integ tests where we make
    'GETINFO version' queries, then compare against the parsed tor version. This
    es no bueno because the parsed version just contains what, according to the
    version-spec, is the version and the getinfo query includes extra data like
    the git commit id it came from.
    
    This is actually an issue we encountered before and was fixed. However, since
    then these queries slipped back into new tests since they seem like an easy
    way of exercising a socket. Oops. :)
    
    Tested by checking out the git commit that gsathya is using (ef0bc7f), then
    running stem's integ tests against that.
    
    This fixes ticket 5918 and might address 5917 too (I'm not sure why the later
    is hanging - that's not happening for me though the test did have an error).

commit 47c9ddd8e460488abad58afb8893a1524e37835c
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu May 17 09:59:24 2012 -0700

    Included last tor warning or error in launch_tor exceptions
    
    Adding the last tor waring or error message that's probably useful for figuring
    out the reason why tor failed to launch.

commit bca28c1e409a318bf9303a635530382c13c17115
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu May 17 09:29:04 2012 -0700

    Replacing enums with class instances for connect_* functions
    
    The connect_port and connect_socket_file convenience functions accept a
    controller type that we would like to return. I had been using an enum for this
    argument, but this means that we both (a) need to manually update it when
    adding controllers and (b) can't recognize controller the user makes.
    
    Replacing this with a class instance, like what TorCtl does. Tested by running
    against a script for printing BW events.

commit b69db7feba640967c2bb6f2740d0aac2bf32666f
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu May 17 09:19:28 2012 -0700

    Helper function to only run tests once
    
    Adding a helper function to skip tests if they've been run before. This is
    needed because some of our tests take a long time to run and are uneffected by
    testing targets, so there's no point in wasting time on running them
    repeatedly.

commit dd597ba4855197653c1d9cd543c3bd6c611432a5
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue May 15 08:36:56 2012 -0700

    Removing platform dependency note from process module
    
    Removing comment that sets the wrong expectations of the process module's
    capabilities. It needs to be platform independent since it's used to run our
    integ tests, and when it isn't that's a bug. That said, Windows is
    unfortunately having trouble with it at present...
    
    https://trac.torproject.org/5493
    https://trac.torproject.org/5783

commit e2974e85a6560d204b674e7179a13074a3494536
Merge: bfb3973 7d96e78
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 17:12:27 2012 -0700

    Merge extrainfo descriptor support

commit 7d96e78e90e455019f674eff2a611449107eb109
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 17:02:28 2012 -0700

    Unused constructor annotations arg
    
    Cached extrainfo descriptors don't have annotations, so excluding it from the
    constructor. Functionally it was already unused, just forgot to remove it from
    the args. Caught by pylint.

commit 5bbf40dbc9a7374f555b4ec7aaf838be6156c9fc
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 16:57:53 2012 -0700

    Missing util import
    
    This should have caused an error, but for some mysterious reason imports
    throughout stem have a wider scope than I'm intending (causing pylint to
    compain about lots of re-imports). Definitely an error - we don't want to rely
    on whatever buggy magic made it work.

commit 1b2ebb2d6624de2133f58386d0b66a0e0da6b962
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 16:53:01 2012 -0700

    Checking that cached extrainfo is all recognized
    
    Asserting that there aren't any unrecognized extrainfo descriptor lines in the
    cached file to help catch new additions.

commit 47acd940498dfe605324bd2e4f08826c7f8c5019
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 16:50:28 2012 -0700

    Parsing 'exit-*' extrainfo lines
    
    Handling the last extrainfo attributes...
      * exit-stats-end
      * exit-kibibytes-written
      * exit-kibibytes-read
      * exit-streams-opened

commit c7ea62037adef91fe11728caab75e19096c69d52
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 16:24:37 2012 -0700

    Parsing conn-bi-direct extrainfo lines
    
    Another special snowflake extrainfo attribute. Parsing and added a test.

commit 97fad7b6f7b6113fdb3e5ca8925224901748a98c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 15:46:58 2012 -0700

    Citing ticket about negative cell-processed-cells values
    
    Comment to explain why we're accepting negative values, and citing the related
    ticket.

commit 153a98eca741d3185e7d20c5e3566763a54384bd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 15:06:29 2012 -0700

    Parsing 'cell-circuits-per-decile' extrainfo lines
    
    Handling the last of the 'cell-*' attributes.

commit 8dd9d118977366ef8a17ad08a39809cbee735df4
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 14:47:40 2012 -0700

    Parsing some more cell-* extrainfo lines
    
    Handling the cell-processed-cells, cell-queued-cells, and cell-time-in-queue
    entries which are all numeric lists. I was a little surprised to find negative
    cell-processed-cells values in the wild.

commit 00e7ce52acc2ddb8f45ccae29469d3463ba39010
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 14:13:57 2012 -0700

    Parsing cell-stats-end lines
    
    Not much to this attribute. The next few are slightly different, though.

commit 86bce1e8db0799efbca86af72a4ef3c0a27f9674
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 13:46:49 2012 -0700

    Parsing entry-stats-end and entry-ips lines
    
    Handling a couple more extrainfo attributes. These were trivial thanks to
    earlier refactoring to make common value types easy to parse.

commit bbf0e80cf889cc4b0738acf93c698fc54720c767
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 13 13:35:33 2012 -0700

    Parsing dirreq-v*-direct-dl and dirreq-v*-tunneled-dl lines
    
    Parsing four extrainfo descriptor fields related to directory mirroring stat
    collection. These are similar to dirreq-v*-resp in that they're key=value
    mappings, with largely known key sets.

commit f7527756275e71b82d426394a74fc0f3ac9824b8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 12 20:10:12 2012 -0700

    Parsing 'dirreq-v2-resp' and 'dirreq-v3-resp' lines
    
    These lines were a bit more substantial to implement since callers would want
    enums for their statuses, but not all statuses are necessarily known. Looks
    like the 'dirreq-v*-direct-dl' and 'dirreq-v*-tunneled-dl' lines will be
    similar.

commit 63acd8169345270df0f1f6b545f67b5453b555fb
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 12 18:11:29 2012 -0700

    Renaming dir_v*_reqs to dir_v*_requests
    
    It's a little understandable that we'd want to squeeze out a few characters in
    the descriptors (er, or it would if we didn't compress them), but there's no
    point for us to do the same.

commit e0ad461440361736b8ce340786433fe6dfaf44f2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 12 15:39:15 2012 -0700

    Parsing dirreq-v2-share and dirreq-v3-share lines
    
    Couple more extrainfo descriptor fields. Ye gods there a lot.

commit b82dca6e2e2543cfa19e9d15be1d8ba89227407d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 12 15:15:55 2012 -0700

    Parsing 'dirreq-v2-reqs' and 'dirreq-v3-reqs' lines
    
    Adding support for a couple more extrainfo fields.

commit b9cd295c09afe662247cc6a89442b40860b9157b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 12 15:07:02 2012 -0700

    Reorganizing extrainfo fields by value type
    
    Grouping both parsing and unit tests by the value format to make it more
    maintainable and easier to add the dozens of other attributes. Also using a
    'dir_' prefix for all directory mirroring attributes rather than 'dirreq_' (I
    was previously being inconsistent on this).
    
    This expands the testing to cover more attributes, along with adding support
    for the "dirreq-v2-ips" and "dirreq-v3-ips" lines. Those lines revealed some
    misses in how I was parsing the maxmind locales. In particular I was missing
    '??' and numeric locales like 'a1' which it uses for an 'Anonymous Proxy'.

commit bff56e18ecde250eb9ad2eba7db5e29d1817bd8f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 12 13:57:53 2012 -0700

    Revising descriptor pydocs
    
    Revisited the ordering, defaults, and descriptions of descriptor attributes to
    be shorter and more user friendly.

commit 2ddffb3d961bcc01829ad58c96b8fa236a35cfff
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 12 12:42:25 2012 -0700

    Dropping measurement *_line attributes
    
    Removing attributes like read_history_line which contained the raw value for
    the read-history line. These are left over from when I made no effort to parse
    the line, and are useless since we have their parsed values.
    
    Minor changes to our pydocs so these related items are better labeled.

commit 6f40a5e87228ea8f0af76d2aeba859025ce468b2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 12 12:32:09 2012 -0700

    Parsing dirreq-stats-end extrainfo lines
    
    Parsing and test. This is the same format as bridge-stats-end so combining
    those tests.

commit c33c4a0f0e6b98898622d0929e531d12abdce72a
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu May 10 09:29:04 2012 -0700

    Parsing bridge specific extra-info params
    
    Parsing, validation, and tests for bridge specific lines...
    - geoip-start-time
    - geoip-client-origins
    - bridge-stats-end
    - bridge-stats-ips
    
    I haven't yet seen an actual bridge descriptor, and I'd like to add one as an
    integ test but that can come later.

commit 15a25f0b05eacfff21309715a4838d4ccd6f11f9
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu May 10 08:31:26 2012 -0700

    Parsing geoip-db-digest lines in extrainfo descriptors

commit d3f7c00fa6cf2573f30551589338e4d42d61465a
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu May 10 08:00:36 2012 -0700

    Accepting additional args on extra-info line
    
    Forgot that the spec says that extra arguments should be accepted and ignored.

commit 0737bada04f71434e9169bc964257dffce8f90c3
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 9 10:01:32 2012 -0700

    Parsing extrainfo dirreq-read/write-history lines
    
    Handling the dirreq-read-history and dirreq-write-history. These are almost
    identical to the read-history and write-history lines so tieing it into that
    parser. This is the last field I need to parse the metrics descriptor I'm using
    for an example in the integ tests.

commit 7178c5cb0a831d217c969ad38453dfdc85630983
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 9 09:27:16 2012 -0700

    Renaming read/write_history to read/write_history_line
    
    The 'read_history' variable name was a bit vague, and generally the attribute
    isn't useful unlike its read_history_* counterparts. Renaming the variables to
    make it clear that read_history_line is the raw content for the read-history
    line.

commit 9d24a9090a00a8686b6eae3258bb6af84cd06087
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 9 09:11:46 2012 -0700

    Only checking cached descriptor existance when needed
    
    Maybe once upon a time we had multiple tests that ran over the cached
    descriptors? I don't recall it, and we definitely don't right now so only
    checking for the existance of the cached-descriptor and cached-extrainfo files
    when we want to test against them.

commit 02451060cf8d5fcc8834d455e0c12a24b08b876b
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 9 09:02:51 2012 -0700

    Helper function for getting descriptor resource paths
    
    Making a common helper function for getting the paths of resources in the
    descriptor data directory rather than making everyone figure out a
    DESCRIPTOR_TEST_DATA constant.

commit 46fb6b272e45ab8c569e1439a53a985c5a8d827d
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 9 08:54:26 2012 -0700

    Letting runner's get_test_dir() provide resources
    
    The most common use of the integ runner's get_test_dir() method is to get a
    path for resources in that directory. Providing an optional argument to get
    that path rather than making the caller use os.join().

commit 406205c2549e855140723bc42af874248a6b3026
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 9 08:42:48 2012 -0700

    Extra-info integraion tests
    
    ExtraInfo counterparts for parsing a metrics descriptor and tor's cached file.
    There's some common bits with the server descriptor tests that I plan to move
    out.

commit fa6d81cf69d5590f6bcb0ad94cf34a7b313f02c2
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed May 9 08:09:00 2012 -0700

    Parsing read/write-history lines in extrainfo descriptors
    
    Copying the code for parsing from the server descriptors. I suspect that this
    isn't yet exercised by the tests - I'm next gonna improve our integ tests then
    check if this would be appropriate for a unit test too.

commit b8e08ce69d30d10cfa6879b7205f62e988fea0e1
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue May 8 09:53:13 2012 -0700

    Helper function for parsing timestamp/intervals
    
    Several extra-info descriptor lines are of the form...
    'YYYY-MM-DD HH:MM:SS (NSEC s) *'
    
    So making a helper function to parse it. Using it for the server descriptor's
    read/write-history entries and it improves code readability quite a bit...

commit 9042c172f8544912888d174ab18b4a19882db7e3
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 7 09:27:06 2012 -0700

    Reverting change to recognize 127.0.0.1 as bridges
    
    Reverting change 477d448 since Karsten reports that bridges with that scrubbing
    scheme should no longer exist.

commit 77a498ae31d0c95d1cdaa88752965fcea870990d
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 7 09:23:09 2012 -0700

    Unit tests for extrainfo descriptors
    
    Few basic unit tests for extrainfo descriptors. These are done in a similar
    fashion to the server descriptors. I'm highly tempted to refactor out some
    common bits, but at present that would probably hurt code readability more than
    it would help. This'll change if I keep using the same type of helpers for
    descriptor unit tests.

commit 7a91e7d85bbc24df130147cc8789e87a5c0f8911
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 7 08:39:34 2012 -0700

    Parsing extrainfo published and router-signature lines
    
    Parsing the remaining two mandatory fields for extrainfo descriptors. These
    fields are identical to what's in server descriptors.

commit bd1d3f345f25835a52813449f4bb45bdddf32eaf
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon May 7 08:37:34 2012 -0700

    Bumping runtime of async event test by a second
    
    Testing for asynchronous events relies on BW events happening every second by
    checking that we get two of them. This is not reliable since we don't have a
    hard assurance about its rate. Giving the test an extra second to make it more
    reliable.

commit 234a90a79624e5503c4bf65db01c14e54fb9b8de
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 6 20:52:21 2012 -0700

    Basic constraint checking for extra-info descriptors
    
    Parsing the 'extra-info' line and performing basic validation about extra-info
    descriptor fields (that we have required fields, most fields don't appear more
    than once, etc).
    
    I'll add some unit tests for this after I've added parsing for all of the
    mandatory fields.

commit 13944b062ff26a5efa178ff8d5552ff1d049574b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 6 20:08:53 2012 -0700

    Skeleton for extra-info descriptors
    
    Basic module for parsing extrainfo descriptors. This doesn't actually do any
    parsing yet, just turning the raw content into a ExtraInfoDescriptor instance.
    This abstracts code we'll need out of the server_descriptor module so we'll be
    able to use it.
    
    What code there is here is exercised via the runner tests.

commit 7930b74b6586220cd02ee816107a85fc1ba7e4ae
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun May 6 13:38:45 2012 -0700

    Missed a few more 'version 3' labels

commit bfb3973049c72a9bd6de951566417eca7e132e70
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri May 11 09:28:51 2012 -0700

    Integ tests for launch_tor
    
    I've been leaving launch_tor untested because it was exercised in order to run
    the integ tests and any tests involving it would take a long time, on the order
    of a dozen seconds! Hey, it's a long time when you run this as often as I do...
    
    It's an important function and has more options than what's exercised for
    running integ tests so adding some tests for it.

commit f3c5f334ce84621547139ae70eedb2c3fbfa2076
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri May 11 09:05:18 2012 -0700

    Expanding launch_tor to support commandline options
    
    When writing small scripts we often want to start a tor instance for a special
    purpose, for instance to test a new feature. When doing this we want a
    bare-bones tor instance with just a few specific options. Adding better support
    for this by making the launch_tor function accept command-line options and the
    ability to use a blank torrc.

commit 80809adcf7023d91c08f97ee9e7affcb3a3bc748
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 5 19:17:55 2012 -0700

    Testing for the controller's asynchronous event handling
    
    Adding a test that hammers the control socket with queries while also listening
    for BW events. This also tests for and fixes a bug where listeners wouldn't get
    all of the enqueued events if the controller was closed.

commit af691f5b54d1f571ee51d8c67010ea19e4c672fd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 5 17:48:29 2012 -0700

    Supporting controllers in the connection module
    
    Users will be using a BaseController subclass unless they really need to work
    at a low level, in which case they'll be using a ControlSocket. Making the
    connection module (which does authentication) support both.

commit 81f272b9e227b36fe93037cbd1d8ef2265403269
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 5 17:06:12 2012 -0700

    Making decriptor reader accept path or list of paths
    
    More often than not people will want to read descriptors from a single path
    rather than a list of paths. There's no need to make them wrap that path in a
    list before calling us.

commit 10cb30b546016f975559bdf1d8dd785ee516ae84
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 5 16:54:24 2012 -0700

    Making descriptor reader persistence more convenient
    
    Our current functions for loading/saving processed file listings are fine if
    you want error handling and a great deal of control. However, I suspect that
    most callers would prefer for this to be an attribute of the reader itself.
    
    Adding an argument that performs best-effort persistance of our processed files
    listing.

commit 477d448f5f19d9d5336e3297943b14567bb55c52
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 5 15:51:47 2012 -0700

    Parsing descriptors with a 127.0.0.1 address as bridges
    
    A prior scrubbing scheme for bridge descriptors scrubbed their addresses to be
    127.0.0.1 rather than 10.x.x.x. Parsing both as bridges so we're compatible
    with both schemes. Thanks to Karsten for pointing this out.

commit cffc3c1be1c4c2a00ce41325a37038d83c8a80a1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 5 15:47:06 2012 -0700

    Mislabeled server descriptors as being version 3
    
    We're using version 3 of the dir-spec, but server descriptors themselves are
    still using their initial version. Correcting the class names and documentation
    that labeled them as being version 3. Thanks to Karsten for the correction!

commit 3a5e9979e40a400b8f4861406bc75af32578c30c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat May 5 15:27:59 2012 -0700

    Separating bridge scrubbing validation into method
    
    Bridge descriptor scrubbing is a moving target and it's quite possible that
    we'll need to parse descriptors that conform to a newer or older scheme. Hence
    making scrubbing validation part of the constructor is inappropriate. Moving it
    to a separate method instead.
    
    This includes two methods, one to simply check if we think that the scrubbing
    is right, and another to get descriptions of the issues. This is of limited
    usefulness to callers, so we might need to switch to an exception hierchy
    later. However, we don't have any use cases that care to check the scrubbing
    yet so leaving that alone for now.

commit fcbbc58d700bb6a22a4d1e7bd7d5fcb22999c9f2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 22 21:14:33 2012 -0700

    Ensuring that we're closed when msg() raises SocketClosed
    
    The ControlSocket's recv() method cannot assure that we're closed when it
    raises a SocketClosed exception (as explained in 4f8be72), but the
    BaseController's msg() method can.
    
    Integ tests were inconsistently failing because I expected to have this
    assurance.

commit f921c569d3565a554f90a6ae72b52032eca86cb9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 22 20:57:31 2012 -0700

    Bug with socket tests when we had an emulated chroot
    
    Forgot to make one of the fixes for handling chroot tests in a socket use case.

commit 4f8be72088e40bb4a6b62e9e39a4c3d9121ec472
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 22 20:40:53 2012 -0700

    Concurrency is still hard
    
    It felt like the handle_close flag was hacky and now I know why. If recv()
    tried to call close() while a send() call was in flight and would also
    eventually call close() then we'd encounter a similer type of deadlock as
    commit d5162f4 tried to fix.
    
    This did not come up for control port connections, but it did frequently occure
    for control sockets (as seen by deadlock when running the RUN_SOCKET target).
    This did not always occure because the recv() sometimes acquired the send_lock
    before the next send() call, making everything proceed correctly.
    
    Fixing this by having recv() make a non-blocking attempt to get the send_lock.
    If it works then we know that we can safely close, and if we can't acquire the
    lock then we know that a send() or close() call is already in progress and can
    leave the closing to them.
    
    This does introduce a possible issue where the send() call succeeds, the recv()
    call fails with a SocketClosed, and we're left in a state of being alive.
    However, this is both a weird use case (how can the send() work if the socket
    is closed?) and also not strictly wrong (raising a SocketClosed does not mean
    that we've finished shutting ourselves down). It's a little tempting to add a
    dedicated close_lock to account for this, but after looking into that I
    realized that this would make the concurrency far more convoluted.

commit 7271f04a9cdaf6bb6eb633424891c17654b2807f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 22 18:38:16 2012 -0700

    Fixing testing thread leak and adding check for it
    
    Python does not gracefully handle lingering threads at shutdown, even if
    they're daemons (... which is dumb since that's the only point of the daemon
    thread). Having threads causes occasional nonsensical stacktraces with things
    like "'NoneType' object has no attribute 'socket'" for module references.
    
    Adding a check after running our integ tests that only the main thread remains,
    and erroring if that's not the case. This also fixes the thread leak that was
    probably the issue in...
    https://trac.torproject.org/projects/tor/ticket/5512

commit 636fd3704d6ddcba6a2a780119570fc2e25dd77f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 22 15:05:17 2012 -0700

    Handling chroot in stem.connection and integ tests
    
    On ticket 4896 [1] I proposed a couple ideas for how we could handle chroot
    setups, but on reflection neither of them were very good.
    
    At a low level we can't reliably expand paths, nor should we try. If the user
    makes a raw 'GETINFO config-file' query then we should simply return what tor
    gives us, not try to 'fix' it by expanding the path.
    
    Rather, we should correct for chroot prefixes at a higher level like the
    controller. My current plan is...
    
    * The Controller class will have an optional chroot_path constructor argument,
      and a get_chroot() method. All of the Controller's methods and those of
      subclasses should take it into account for tor resource paths.
    
    * The stem.connection functions now accept a chroot_path argument. We need this
      since they will construct Controller instances, and also do the initial
      authentication (we need to know about chroots for cookie authentication).
    
    [1] https://trac.torproject.org/projects/tor/ticket/4896

commit 0f4fc2ee2a3e442aa3cad3dd5bb28580771570a2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 22 13:55:04 2012 -0700

    Providing cookie path with CookieAuthFailed exceptions
    
    A common attribute that callers want when authenticate() raises a
    CookieAuthFailed is the path that we tried to read the cookie from. Since the
    authentcate() call does its own PROTOCOLINFO query the caller never sees it
    (except for the exception message). Simple thing to fix.

commit 6bb517d876243aafd5cf3ae0eef7ecbe45958389
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 20 12:25:18 2012 -0700

    Catching unexpected exception from descriptor reader
    
    The os.stat() call can raise an OSError, so catching that and notifying the
    skip listener.

commit 13f7ce34e735b2547f3ab669bed0ceb571624c0f
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Apr 19 19:14:42 2012 -0700

    Testing option for chroot setups
    
    Adding a new target to simulate a chroot setup by stripping the test directory
    from any control socket responses. As expected this is causing integ tests that
    validate this path to fail.
    
    Sathyanarayanan took the first several stabs at this, and this is just another
    potential option for...
    https://trac.torproject.org/projects/tor/ticket/4896

commit e1a3aba733313c842c24c40ac235c073a548030d
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Apr 19 09:23:27 2012 -0700

    Allowing for negative uptimes prior to tor 0.1.2.7
    
    Tor 0.1.2.7 fixed a bug which could result in negative uptime values. Making
    our descriptor parser accept negative uptimes for tor versions with the bug,
    and fail validation for later versions.
    
    This includes a test with one of the problematic descriptors, checking that
    we can both parse it, and will fail validation if it was for a later version.
    
    Thanks to Karsten for figuring out where these values were coming from!

commit 725beac5437d85e8da5baa8a6fce90ce76fa9f9f
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Apr 18 09:31:14 2012 -0700

    Adding test for contact info with carriage return
    
    Interesting use case spotted by Karsten on...
    https://trac.torproject.org/projects/tor/ticket/5637

commit 21228b096617abbaabc1d53ca964b8cad591bf88
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Apr 17 08:47:19 2012 -0700

    Retaining padding in the server descriptor digest() funciton
    
    Network status entries exclude base64 padding to save on space but there's no
    reason for us to do the same. Adding a comment explaining why they might
    mismatch. Thanks to Karsten for explaining it!

commit ebf4c0274f7e90da8946f2ba49c3e787fb367264
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Apr 17 08:46:21 2012 -0700

    Left out part of the whitespace rename change
    
    Oops! Forgot to do a 'git commit -a' to pick up the other changes from renaming
    to 'check_whitespace.py'.

commit e69cb95f93d6646e809bf0143ff595dae32a2b5e
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Apr 17 08:42:40 2012 -0700

    Checking if a file's a tarball could fail due to permissions
    
    I encountered a transient IOError from the tarfile.is_tarfile() call. I'm still
    not sure why it complained about '/vmlinuz.old' (and the test actually still
    passed), but it's inappropriate for the reader to raise an IOError there.
    Falling back to mime type.

commit 1a2c7559e31fea6fba6856594f982f6f0959cd5d
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Apr 16 09:22:25 2012 -0700

    Renaming whitespace_check.py to check_whitespace.py
    
    Previous naming was stupid, making it read better.

commit 06aa889eddea7b608fd5d776ff2c1a1cd90d3ab4
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 15 18:40:01 2012 -0700

    Only coloring test output if it's to a terminal
    
    Dropping the '--no-color' argument in favor of a builtin check for if our
    output is going to a tty teminal or not. This is the same trick that git does
    for its 'color.ui = true' setting.

commit 22965e5c209ecc7b924093757e7e75767ea82379
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 15 18:36:35 2012 -0700

    Correcting whitespace
    
    Fixing the mistakes caught by my neat new whitespace checker.

commit 25f65d01a10ab5e4993f48317927774c956e73ae
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 15 18:27:37 2012 -0700

    Checking whitespace as part of the tests
    
    A common mistake made by new contributors to stem (... or any project,
    actually) is incorrect whitespace. Checking for tabs, trailing whitespace,
    and incorrect indentation levels after running tests to make it easier for
    patch contributors to get this right.
    
    It's already showing a few mistakes in code that I wrote... oops.

commit 2237bce8cdf66f189bb6ab2e83435a9c3b06b694
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 15 16:47:25 2012 -0700

    Filling in missing __init__.py entries

commit a534c0607139ac9250243673d081c790866abc48
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 15 16:34:57 2012 -0700

    Adding missing funtion to hearder doc
    
    The ControlSocket was missing an entry in its header docs for __enter__ and
    __exit__.

commit a676e6f7d81700e670610db7e8769007d30284d3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 15 16:29:03 2012 -0700

    Removing incorrect comment about dead tor code
    
    Robert reports that getinfo_helper_listeners isn't dead code after all, so
    removing the incorrect comment.

commit 67dc38b9c7d9304caa1bffde05e912849f7dc5f1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 15 16:21:48 2012 -0700

    Changing None checks to be by identity
    
    The 'is' keyword is almost completely useless, but the one place where identity
    checks are preferred is when checking for None. It's faster and won't be
    tricked by buggy equality methods...
    http://jaredgrubb.blogspot.com/2009/04/python-is-none-vs-none.html
    
    This is the suggestion as per PEP 8...
    http://www.python.org/dev/peps/pep-0008/#programming-recommendations

commit ac3f6271e0896cbdb2e668576f43670b260bfa2b
Merge: 809b341 9f76969
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 14 19:12:01 2012 -0700

    Merge server descriptor changes suggested by Karsten

commit 9f76969739eccf740da0f77378bcabc5672a85bb
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 14 18:30:40 2012 -0700

    Handling descriptor files in a depth first fashion
    
    Directories enqueued all of the files that it contained prior to processing
    them which has a couple obvious disadvantages...
    
    - huge targets like the root directory or years worth of descriptors can
      consume lots of memory with the paths alone
    
    - this could easily cause us to have a huge startup time before we provided the
      caller any descriptors
    
    This was stupid, depth first parsing makes much more sense.

commit 707897bea765b3d403dea9eaf4a3a0788a455bd0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 14 17:23:03 2012 -0700

    Handling empty read/write-history value listings
    
    Bug where relay server descriptors with read/write-history lines but no values
    on them would fail validation. Added a unit test for this and fixed the bug.

commit 96678f167f210d10f65c14d977e9f764c4d70dbd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 14 17:09:37 2012 -0700

    Adding a test for descriptor content from 2005
    
    Version 2 relay server descriptors were first introduced in 2005 so adding a
    descriptor from back then. This in theory doesn't exercise anything that the
    unit tests don't already do, but it still provides some good use cases (mostly
    around deprecated attributes).
    
    I ran over the full December 2005 metrics tarball and all descriptors were
    happily parsed.

commit 04176384995a89555067ddecf4cf6a7fe882ce39
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 14 17:03:28 2012 -0700

    Making descriptor reader skip tests more resilient
    
    Couple problems with the skip tests for the descriptor reader:
    - A bug in the test for skipping unmodified files caused the test to rely on
      the order in which descriptors were read. It had a check that should have
      caught this, but rather than "assertEqual(1, len(skipped_files))" it had
      "assertTrue(1, len(skipped_files))".
    
      Fixed the check and narrowed the test to just the single descriptor file
      that it was supposed to test against.
    
    - The test which checks that we properly skip and report non-descriptor content
      would fail if other files are added to the data directory. This is as it
      should be, but the error message wasn't helpful and I've encountered this a
      lot due to vim '.swp' files.

commit 62f34c67bcaaa7205b09cc7cf6201d78447c0a39
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Apr 14 15:51:16 2012 -0700

    Expanding attributes for read/write-history
    
    Parsing and validating read-history and write-history attributes. These are
    depricated fields, but still found on archived and extra-info descriptors. This
    includes a unit test to exercise the additions.

commit b8f6fd2c1a601676be30dd2850211f8fedfb3c14
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 13 22:57:33 2012 -0700

    Removing integ test for descriptor digests
    
    Dropping the server descriptor integ test for digest(). It was a valiant
    effort, but is flaky because it'll always fail when the consensus or descriptor
    changes and are out of sync.

commit c64f663c6f3215eb4e492cfe60bf7be46f50a1d0
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 13 22:53:36 2012 -0700

    Removing restriction that descriptor socks port is zero
    
    The socks port of server descriptors is depricated and always zero, but this
    isn't necessarily true for archives.

commit ed0e8ac4d0dca0c58cc17e2314839351590038b8
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 13 22:39:33 2012 -0700

    Renaming calculate_digest() to just digest()

commit 8c1b5b6053e08bce55b9bc935daa492f727c68e8
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 13 22:19:33 2012 -0700

    Implementing calculate_digest() for server descriptors
    
    Implementation and test for a function to get the digest value for a relay
    server descriptor. This is the same value found in the network status entry.
    This was an addition suggested by Karsten.

commit b56bb55187d6baa00510b98c6ae1c91739476acf
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 13 20:36:33 2012 -0700

    Removing timeout from descriptor reader enqueue calls
    
    The reader had a timeout for enqueue operations so it could periodically check
    if we were stopped while waiting to insert a descriptor that we've read. As
    Karsten thought we were able to do better. Dropped the timeout by clearing
    our queue on shutdown (to unblock any enqueue calls in process), and skip
    further enqueuing while we're shutting down.
    
    Both iterating and stopping are under a read lock so we don't need to worry
    about this changing the order in which descriptors are provided to callers.

commit 0d478c57866c868c63be50714887c1b1535fbd0f
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 13 20:11:37 2012 -0700

    Removing deprecation warnings for server descriptor fields
    
    The read/write-history and eventdns server descriptor fields are part of the
    spec but no longer used. I was logging a notice that they're deprecated when
    we encountered them, but this is neither useful nor does it make sense. We
    might be parsing old metrics archives where those attributes are perfectly
    valid.
    
    Integ tests include a check that those aren't seen in the wild, which is what I
    really care about.

commit 850b1cbbc07554ea369b44c89b3bbfdd1b99fa69
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Apr 12 10:53:28 2012 -0700

    Unit tests for bridge descriptors and related fixes
    
    Unit testing the new bridge descriptors. This mostly focused on parsing
    or-address entries and the validation that the class does for scrubbed values.
    Fixes included having string port values and not checking that or-address
    addresses were scrubbed.

commit 785b22f274c30c2e52703c97bee3f2fb5a23e247
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Apr 12 09:36:49 2012 -0700

    Moving relay/bridge differentation into parse_file_v3()
    
    It was confusing for parse_file_v3() to only work for relay server descriptors
    so moving the hack to differentiate relay from bridge descriptors there. This
    is also nice because it moves server descriptor logic out of the general
    __init__ module and into the one that's specifically for handling server
    descriptors.

commit c3fddad00ecad4ea5a94dafcfe6a0aaa10c4cfd0
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Apr 11 21:12:26 2012 -0700

    Differentiating read bridge and relay descriptors
    
    The last integ failure was due to our attempt to read a bridge descriptor file
    as a relay descriptor, which doesn't work because it is missing cryptographic
    entries. Adding a bit of logic to differentiate relay from bridge descriptors
    so we can parse them as their appropriate type. The cues the difference based
    on the 'Unnamed' nickname and '10.x.x.x' address scrubbing.

commit 19ba0506cec868e118ddfcf89cd30d44d3b43692
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Apr 11 20:59:56 2012 -0700

    Handler for bridge descriptor 'or-address' entries
    
    Parsing 'or-address' entries in bridge descriptors. This parsing logic should
    be moved to the ServerDescriptorV3 class if/when ipv6 support is available to
    normal relays.

commit dbd6a80ffc65f8d77d6d798b49f8c899e5ef09e9
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Apr 11 20:50:08 2012 -0700

    Utility for validating IPv6 addresses
    
    Function and testing to check a string is a valid ipv6 address.

commit 57ed21a26d440d32e01924cbc027d97665fe8dbc
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Apr 10 09:03:39 2012 -0700

    Server descriptor reformatting
    
    Cleaning up the recent changes to split apart relay and bridge descriptors,
    dropping an unnecessary parser class and rearranging things to be more
    readable.

commit b2d1d632696e5d47789ea9d1b4f817df8634e681
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Apr 10 08:25:13 2012 -0700

    Dropping duplicate constructors
    
    Bridge and relay server descriptor constructors are identical, moving to
    parent.

commit 0f0eac50b59e33d5a25b428ddb9b0d16be023a28
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Apr 9 09:11:13 2012 -0700

    Splitting relay and bridge server descritors
    
    Making separate classes for relay and bridge v3 server descriptors. This is the
    more proper object model and allows us to easily handle their differences in
    terms of parsing and validation.
    
    This includes the implementation and an integ test, but the implementation
    isn't yet complete (missing or-address parsing) and we still need unit tests.
    The current unit tests pass but the integ test fails due to or-address.

commit 57398f137882e22d5ca278a7944212ec62e4d5a1
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Mar 29 09:07:44 2012 -0700

    Stop tracking last-modified timestamps if file disappears
    
    Improvement suggested by Karsten. For long running processes with a rotating
    pool of descriptors we could accumulate a lot of 'path => last modified'
    entries, so only including the entries for files that still exist.

commit dd6b71c2b4fe27c33f82ba45e405be9c3fabd409
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Mar 29 08:54:22 2012 -0700

    Test for descriptor with non-ascii content
    
    Test suggestion by Karsten. This reads a descriptor with non-ascii content
    (hiragana and something else), checking that the results match expected values.

commit 809b34159067c25730d71b35460ba2e0fbf9a815
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 8 14:34:59 2012 -0700

    Removing stem.util.system.is_relative_path()
    
    Why did I make this function again? Python already has os.path.isabs()...

commit d9dc5babf533460975895cca1d6cde14dcea8e41
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 8 14:28:00 2012 -0700

    Making system unit tests more platform independent
    
    On Windows the alternate path separator caused the unit tests for expand_path()
    to fail. Switching the expand_path() and is_relative_path() to not make any
    effort to work on Windows and fixing the test to be platform independent. In
    the future we should make these funtions work on Windows but this isn't vital
    at the moment and a pita due to its '[A-Z]:\\' prefixes (I really don't wanna
    do a regex just to check if a path is absolute...).

commit 0354e9e656655f33c04fd75ebcb45ff0f70959c6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Apr 8 14:03:55 2012 -0700

    Replacing os.uname() calls with platform.system()
    
    I was doing os detection by calling 'os.uname()' which has the unfortunate
    disadvantage of not functioning in Windows. The 'platform.system()' apparently
    does the same for our purposes but... well, works.
    
    Caught thanks to reganeet on...
    https://trac.torproject.org/projects/tor/ticket/5493

commit 35f31d8c1527e106013ceb293302901dce8f9cf0
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Apr 6 09:26:50 2012 -0700

    Checking for windows style execs in is_available
    
    Stem tries to check that a command is available prior to running it, which
    choked on Windows due to the '.exe' extension. Issue caught by reganeet and
    fix suggested by him.

commit e427e6d7aff32d15ecf3d5ef3e0c78d2758deb24
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 30 08:14:33 2012 -0700

    Fixes for prior version parsing fixes
    
    I never cease to be amazed at how error prone something that sounds as simple
    as 'parse the tor version' can be. Guess that's why we're writing a library...
    
    This has a couple fixes for the prior set of fixes.
    
    a. The runner was misparsing 'GETINFO version' output for verisons without a
       space.
    
    b. Sinister bug (took me almost an hour to track down) where the results from
       our mocked "tor --version" call were cached and broke later tests. This only
       manifested if your tor version wasn't 0.2.2.35 and you ran both the unit and
       integ tests together.

commit c63124069422cbba45f4b08b223f2bcb6eb99ef6
Merge: 222d58a d5cd861
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Mar 29 18:54:23 2012 -0700

    Several fixes to tor version parsing
    
    Fixes by neena for the tor version parsing, discussed in...
    https://trac.torproject.org/projects/tor/ticket/5472

commit d5cd861c377d336a5f5f9edb4b85e2c2cccaabf3
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Mar 29 18:52:11 2012 -0700

    Trivial tweaks for version fixes
    
    Very minor changes to formatting and dropping a no-op line.

commit 9408f4a16a45d608a7d7711215496734902f5f12
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Thu Mar 29 23:04:17 2012 +0530

    Update code to reflect new version parser, all tests pass

commit 50a240d7ff83a435a75d54c90d52088cd723c450
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sun Mar 25 23:04:13 2012 +0530

    Treat versions with whitespace in them as invalid

commit d2f7e1925e61bf73c8d7a78ed48a01badb5399a0
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sun Mar 25 20:58:54 2012 +0530

    Neater string finding code

commit 27e3806d4f37f47965c3237083d885db147faee7
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Sun Mar 25 22:50:55 2012 +0530

    Fix Version to match . strictly

commit ec9101e44c74ea6e0144c970b0e675dddc3a9aee
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Fri Mar 23 04:55:48 2012 +0530

    Fix version parser to ignore git hashes
    
    get_system_tor_version strips the git hash
    add test for get_system_tor_version
    remove testcase which doesn't comply with the new version parsing code

commit 222d58a9bad7e1675f70002fb00b2a9b962cfe63
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 28 20:19:03 2012 -0700

    Typo corrections for descriptor changes
    
    Patch by karsten to fix numerous misspellings.

commit df2084b0e746e6c5036528b01884a168b3b513ce
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 28 20:09:08 2012 -0700

    Minor corrections for os.sysconf check
    
    The patch from beckbaibai fixed his issue but left CLOCK_TICKS undefined on
    Windows which would result in a NameError when using proc.get_stats (the
    function *should* cause an error, but not that one).

commit f5950d01e1578a532c03da55165030f8c3ba3919
Author: Beck <csybeck@gmail.com>
Date:   Tue Mar 27 01:11:30 2012 -0500

    fixed system module undefined in windows

commit bc0e578615d9a856ab9052acc5a116774c49d133
Merge: 989fd95 975d98d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 25 17:09:21 2012 -0700

    Merge branch 'metrics-lib'

commit 975d98d7337d4dd566aa5e9967e62afe2fdb0bc0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 25 16:12:32 2012 -0700

    Integer truncation could break reader's last modified check
    
    When saving the last modified timestamps they're saved as ints. However, the
    timestamp itself is a float, so the following check for an unchanging timestamp
    would fail.

commit cf0e790c809d4057eceb37550b33f6f5bf67b065
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 25 16:02:23 2012 -0700

    Fixing reader's pydoc example

commit 4502701f0ec8462c0848128f000ec18c14a94167
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 25 14:52:43 2012 -0700

    Unit tests for server descriptor parsing
    
    Borrowing heavily from Karsten's metrics-lib unit tests [1], plus a few of our
    own. This doesn't include everything, some are covered by util tests and others
    are effectively covered by parsing metrics and cached-descriptor files. That
    said, I wouldn't be against including more - my fingers are just getting
    numb...
    
    While looking through those tests I discovered these deltas...
    
    * When parsing the descriptor he validated that we had the correct first and
      last entry but we didn't. This was a bug, now fixed.
    
    * He checks that the 'protocols' line only contains numberic protocols but we
      accept any space separated content. The spec doesn't specifify what the
      protocols can be so I'm leaving this alone.
    
    * He had checks that a 'published' line with a year of '3012' or '1912' would
      fail. I can understand having a sanity check but the spec does not say that
      those dates are invalid so again leaving it as-is.
    
    * Metrics lib has validation for the exit policy but we don't. This is because
      we haven't yet implemented an ExitPolicy class - gsathya is currently working
      on that in...
    
      https://trac.torproject.org/projects/tor/ticket/5454
    
    * Karsten is parsing and validating read/write-history lines. We, on the other
      hand, log an INFO level warning when these appear and assert in the integ
      tests that they (along with 'eventdns') do not exist in our cached-descriptor
      file. My understanding is that these are deprecated entries and should not
      appear outside of the extra-info descriptors. Maybe I'm misunderstanding
      something here...
    
    On everything else we either match or are maybe slightly more strict about only
    allowing content conforming to the spec.
    
    [1] https://gitweb.torproject.org/metrics-lib.git/blob/HEAD:/test/org/torproject/descriptor/impl/ServerDescriptorImplTest.java

commit 896bb442fe9cc3039a98188fd68ddfcba4f0c1e5
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 25 12:39:35 2012 -0700

    Skipping auth message check when cookie missing
    
    When we first run our tests we don't have an authentication cookie unless we've
    run with the RUN_COOKIE target. We have special handling for this, but it did
    not account for running with an open connection (the default) in which case
    we'd encounter a second exception while trying to check the failure message.

commit 1b507eae6bd104e9eba68ad94ff96baa7044b267
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 25 12:31:40 2012 -0700

    Miswired 'ONLINE' and 'RELATIVE' targets
    
    The test runner still used the 'test.' prefixed configuration keys rather than
    'integ.'. This caused the 'ONLINE' and 'RELATIVE' testing targets to be
    ignored.

commit e1f7537e939a60e7c902e853b03daab9c28b3d33
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 25 12:20:43 2012 -0700

    Skipping integ test on cached descriptors if not available
    
    The first time we run our integration tests our data directory is not yet fully
    populated. This causes our test that runs over the cached descriptor file to
    fail. Adding a check that the cached descriptors exist before running it.

commit b3d357b9f12d9bce72ec411f2681fdf84dcb4139
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 25 12:15:15 2012 -0700

    Including deprecated server descriptor attributes
    
    I have mixed feelings about cluttering the server descriptor with dead fields,
    but since we're already this close to being perfectly spec conformant we might
    as well include this last bit. Adding the fields but logging an info level
    warning if we see it, and checking in our cached descriptor that they do not
    exist.

commit 092d36a17a9318c42f09199e66458e399af68fda
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 24 19:58:48 2012 -0700

    Integ test to validate our cached descriptor file
    
    Parsing all of the descriptors in our cached descriptors file to see if it has
    anything that'll make us choke (and by extension indicate either a tor or stem
    bug).
    
    This is an important yet lengthy test, taking around four second. Targets do
    not effect this test so only running it once for the test run.

commit d4fa1279f43c585d9551c6e1f94d3917e14eec99
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 24 19:45:06 2012 -0700

    Dropping attribute for block types
    
    We already have enough attributes, and parsing out the armor type is redundant
    since it's within the armor itself. Besides, it's not likely to change and I
    doubt callers will ever care about it.

commit 7dc7d82bd7b7f70f3dfae21dd607502b4eff6477
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 24 19:38:27 2012 -0700

    Missing version 3 attributes
    
    Adding the server descriptor attributes that are new as of version 3. The only
    attributes that I'm still excluding are eventdns (because it's dead) and
    read/write-history (since they're extra-info now).

commit e0f8bfcacca963e1cbc7811dd1f9e6e7e104c7f8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 24 18:50:20 2012 -0700

    Allowing for extra args on lines with multiple
    
    Well, crap. I've been developing against the version 2 spec on the assumption
    that was both the majority use case and what my test instance was running.
    However, that's evidently pretty ancient and I've been running against version
    3 all along.
    
    First change I'm spotting in the spec is that lines with multiple arguments
    should ignore extras so doing that.

commit d72c21de710ba78e16b6d0fa29949fb43a2e5eb6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 24 18:15:27 2012 -0700

    Server descriptor test for metrics content
    
    Test for parsing a single server descriptor and checking its content. This
    includes a few fixes and improvements for the ServerDescriptorV2 class, the
    most intersting of which is that declaring instance variables prior to init
    makes a single reference. Hence all server descriptors were using the same exit
    policy. Oops. :)

commit 9faad3255491642f87801926512c9cb6d334f6d8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 24 16:50:03 2012 -0700

    Integration test for reading non-descriptor files
    
    Integ test for reading non-descriptor plaintext and binary data.

commit 7e9d454ed4eafe6b5461c5ff5170ca5634899edf
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 24 13:41:11 2012 -0700

    Cleaning up prior descriptor implementations
    
    Adding header pydocs and made a variety of small fixes and naming improvements.

commit 8dc2d66b7dde16e0709dc5c5ff1ac0c320397f22
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 23 22:05:29 2012 -0700

    Server descriptor parsing fixes
    
    Resolving the variety of issues that were breaking the integ tests. We still
    don't have any tests to actually verify the server descriptor parsing, but
    at least we're now exercising it on some real data... and not dieing in a fire!
    Progress!

commit c91532362bd04d90c35adab22d497ba09ee7cc64
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 23 20:23:57 2012 -0700

    Resolving circular dependency with server_descriptor
    
    ... damn this was a pita to debug. A long explanation of the issue is:
    http://stackoverflow.com/questions/8991520/python-conditional-module-object-has-no-attribute-error- with-personal-package
    
    Short story is that the stem.descriptor module imported server_descriptor prior
    to being fully initialized which caused a server_descriptor use of
    stem.descriptor.Descriptor to fail.
    
    The confusing was that it was actually reporting failure in using
    "stem.descriptor" (the module itself) which made no sense because it was listed
    within sys.modules. It turns out that the python importer uses module
    attributes rather than sys.modules mappings to resolve statements (ie, it was
    doing getattr(stem, "descriptor")). However, a module's added as an attribute
    of its parent when its *finished* importing.
    
    This fixes the unit tests, but integ tests are still getting stuck when they
    use the new server_descriptor parsing.

commit ab8fe646817b585ceda7c6062c3294556f8f71cc
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 23 20:23:17 2012 -0700

    Cleaning up minor import issues

commit 55db7157d30e62af2fa610f1bb962e8fc8a29817
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 23 09:39:50 2012 -0700

    Merging descriptor.py with module init
    
    The syntax for using the descriptor module was stupid ("import
    stem.descriptor.descriptor"). Using the json module as an example for using
    __init__.py for base functions. Common ones will live in 'stem.descritor' but
    the contents, like 'stem.descriptor.reader', are also accessable.
    
    This is similar to how the os module works, though in that case they're using a
    lot more python magic to also support multiple operating systems.

commit 3b657b57b4cb4ce5304145a7a55051e758cf9cbd
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 23 09:28:55 2012 -0700

    Using server descriptor when parsing
    
    Parsing server descriptors via the new server descriptor class. This breaks the
    tests and will need quite a bit of work. :)

commit 245c85ecf35e7324ce008c351e1707b5ef96567c
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 23 08:41:43 2012 -0700

    Pylint corrections for server descriptor
    
    Corrections from running pylint over the untested server descriptor class (all
    basic problems that would keep it from running).

commit e20fd3efae6a0e3a7f647017da9f62ccc83aebcc
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 23 07:31:41 2012 -0700

    Parsing descriptor files and annotations
    
    Completely untested change to finish the implementation of server descriptors.
    This parses a cached descriptor file into individual descriptors and
    annotations.
    
    Next comes unit and integ tests to start exercising it in an automated way.

commit 989fd953c6cf59deeaeb65a2a6fbde88d3410136
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Mar 22 10:20:50 2012 -0700

    Fixing incomplete mocking for protocolinfo unit test
    
    The protocolinfo unit tests mock out system calls which should allow relative
    cookie paths to be expanded. However, this didn't account for the is_available
    checks which prevented those calls if the command was unavailable on the users
    system (for instance, OSX). Caught by gsathya.

commit 063ce0bd893bca006b13b8781740924b560cb79e
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 21 09:24:17 2012 -0700

    Removing bold formatting from test results with --no-color
    
    The --no-color argument should remove all ANSI escape formatting but it just
    removed color. As a result the test result text (such as "[SUCCESS]") was still
    bold. Caught and fixed by gsathya.

commit 838a3f37037b7533e010dbb0563191d228fb4947
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 20 09:00:58 2012 -0700

    Including exit policy among the attributes
    
    We'll later have a proper ExitPolicy class but that's a small project of its
    own. Until then this will just provide the list of strings that make up the
    exit policy.

commit 7e64119a665ecbfb3ee06d3ddcfe066f212f63f6
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 20 08:53:08 2012 -0700

    Making server descriptor validation optional
    
    Adding constructor pydocs and an argument to skip validation. This should have
    performance advantages and allow for best-effort parsing of malformed
    descriptor content.

commit 29fad9389f92557e349ffebb32d9cef58b2796c0
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 19 09:14:11 2012 -0700

    Handling the server descriptor's contact and family lines

commit 043f739405ec6012725ce4d86361fc01cb1a0c17
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 19 08:55:47 2012 -0700

    Handling the server descriptor's router-signature line

commit f6221839e8b012fa5848feb5b68eb3f390889ba2
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 19 08:38:40 2012 -0700

    Karsten already had a ticket for opt removal

commit 54c3e6b7c6ea41ff29230c4e6a476f6f7fbac19a
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 19 08:35:12 2012 -0700

    Generalizing PGP style block parsing
    
    Descriptors include blocks for both "RSA PUBLIC KEY" and "SIGNATURE" so
    generalizing that parser (the spec around how to handle these blocks is wrong
    and I'll fix that later). I'm also including the block type since callers will
    want that to be prepared for encryption migrations.

commit 5e7d6fadb520817e21436ecfea6437fa7ae26ab1
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 18 15:56:26 2012 -0700

    Handling the server descriptor's onion-key and signing-key lines

commit 68bb0d6621ddb4806285ccaab298d2fa166050a6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 18 15:49:33 2012 -0700

    Parsing descriptor signature blocks

commit be8e245400161fe05a2899da9a003eefa8110989
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 18 15:21:22 2012 -0700

    Handling the server descriptor's uptime line

commit da5fcc6ba9b170fce548f08848ae2b0b268bf547
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 18 15:18:56 2012 -0700

    Handling the server descriptor's hibernating line

commit 4ae18ed246b8f9f762df154ea493adb5cfbf1e80
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 18 15:08:57 2012 -0700

    Handling the server descriptor's fingerprint line

commit 47f21c97da50392d4d0be38d7292bca920d716df
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 18 14:39:13 2012 -0700

    Stripping off the 'opt ' keyline prefix

commit 0ae8839cd91e658a36e7a4c9e5eeec7a75146651
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 17 19:24:45 2012 -0700

    Handling the server descriptor's published line
    
    Ugggg, I'm always filled with hate when I need to touch the datetime module.
    I'm not entirely sure if I'm supposed to be associating timezone information
    with this datetime object or not - and if I am how to get the f*ing thing to
    realize that it's GMT.

commit ac81552c0b966b1a9bed97c812e5a574ffa6f3db
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 17 18:53:16 2012 -0700

    Handling the server descriptor's platform line

commit 9342515713edbf3c8a8a2db9ac89132a98eb859c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 17 18:44:28 2012 -0700

    Handling the server descriptor's bandwidth line

commit a2aa3c21ea66a38bedb58b561d081e5367ad3982
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 17 18:25:44 2012 -0700

    Validation for server descriptor router line
    
    Checking that there aren't any surprises from a server descriptor's router
    line.

commit 8fe000d4de00a073fba060b39f42a8fa4a423077
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 17 18:20:30 2012 -0700

    Function to check validity of a port

commit 28fb4b1d55952ec905762d4cb400d997e117a0ea
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 17 17:59:33 2012 -0700

    Funcions to check validity of nicknames and fingerprints
    
    Couple functions to check if a string is a valid nickname or fingerprint, plus
    tests. I needed to guess at the definition of a HEXDIG but otherwise nothing
    very interesting.

commit 31cd9b9aa5ea0b5139d3d32ee3a424d9af1e26ca
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 17 17:07:52 2012 -0700

    Implementing and testing is_valid_ip_address()
    
    Moving over an arm function for testing if a string is a valid IPv4 address.
    Also throwing in some unit tests and a fix for entries with leading zeros (for
    instance "1.2.3.01").

commit cb819ce76b011c4ee84fa15014c92166bca49577
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 17 16:19:09 2012 -0700

    Completely untested start for descriptor parsing
    
    Starting on server descriptor parsing. This is gonna need some other classes
    and helper functions (such as an ExitPolicy and is_valid_ip_address()) but
    otherwise should be pretty straight forward.

commit af1cdfcedf961b506b63c923fdd9b5a6a3edd1e5
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 17 15:54:15 2012 -0700

    Stub for identifying descriptor types
    
    The various tor descriptor types were not made to be readily identifyable, so
    we'll probably need to guess their type based on their first line. Ideally that
    first line would explicitly say the type and version, but oh well...

commit da4ccb0bedc122c4776048751c4de0b6cc0e010e
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 14 09:55:38 2012 -0700

    Descriptor module with common functionality
    
    Adding the descriptor module which has the Descriptor class (the common parent
    for all descriptor types) and the function for parsing descriptors from a file.
    This parser will later do type recognition to figure out how descriptors
    should be parsed. This is identical to how metrics-lib does this.
    
    Nice advantage is that we can now move all of the remaining 'TODO' notes out of
    the reader. It's done. :)

commit 5b7eacfce6ac4f370dfb4fb1baa293c4b934689f
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 13 09:26:01 2012 -0700

    Making _get_raw_tar_descriptors() thread safe
    
    At present integ tests are single threaded, but there's a project proposal to
    change that and spawn threads for each target. Fixing this helper function so
    it won't be sad when that day comes.

commit 149d1bd2932a093ffe7fb75298da9b442b3b64c7
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 13 09:19:48 2012 -0700

    Condensing basic example a tiny bit
    
    Combining the constructor and 'with' lines since that's probably the common
    case. The second example already shows a standalone constructor anyway.

commit 7faf52e5d122c9266baefec4a31e4d3a3b40ce9f
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 13 09:18:24 2012 -0700

    Inverting save_processed_files arg order
    
    Little nicer when the usually-static argument is first imho.

commit 985d1c473526748013bc7f09658c5d24df76ef5c
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Mar 13 09:13:35 2012 -0700

    Adding test for the DescriptorReader's buffer_size
    
    Simple test that we don't read ahead more than the DescriptorReader's buffer
    size.

commit 828d5dac1c0eda9db996438af45515c0dec0cef0
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 12 20:28:16 2012 -0700

    Adding tarball support to the DescriptorReader
    
    Adding support for reading directly from tarballs (which is how metrics are
    commonly fetched). This supports all forms of compression that the tarfile
    module does (gzip and bz2 among other). Including some tests and archives to
    read against.

commit 924e1bd3d3a9b7b0578ad9b9bb86a86d0278ff3b
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 12 19:25:06 2012 -0700

    Minor pylint corrections

commit adbc1991fdae353825b893f5e90aee6ee882e0a4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 12 19:17:33 2012 -0700

    Limiting the number of buffered descriptors
    
    Preventing the DescriptorReader from having unbounded memory usage by limiting
    the number of descriptors that we'll store before we wait for our caller to
    request some. This doesn't technically make our memory usage bounded since a
    single descriptor doesn't have a limited size, but if one descripter can
    trigger the OOM killer then we have a problem. :)
    
    This isn't yet tested because we only have a single descriptor in our test data
    (we need at least two before we can test this). Adding a todo note for now.

commit 63461450b7d7651c11d1f911f3443bf8a202f3f6
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 12 09:34:15 2012 -0700

    Minor tidying up of integ tests
    
    Reread the DescriptorReader integ tests and tidied up a few minor stylistic
    issues.

commit 34d14afca254eac8f0493f9c9b8aabc7c5400899
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 12 09:27:31 2012 -0700

    Notifying skip listeners of files that were already read
    
    Skip listeners are supposed to be notified of all files that we skip, but
    wasn't accounting for files that we skipped because their 'last modified'
    timestamp indicated that they'd already been read.
    
    Modifying the integ test for set_processed_files() to test this listener
    condition (more graceful that the prior test that it had been doing).

commit 5d61c479081b73eb241bf4a6c541e15ddb7ea713
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 12 08:55:46 2012 -0700

    Condensing no-op iterations over DescriptorReaders
    
    At multiple points in the tests we were iterating over all of a
    DescriptorReaders' items without doing anything. There's no need to do the
    iteration via a loop - coverting it into a list does the same thing.

commit ca837ddcb6f7cc6d118d4e1a774d2c4b22dbda79
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 12 08:52:17 2012 -0700

    Allowing the DescriptorReader to be run multiple times
    
    Improving the usability of the DescriptorReader class by making it so callers
    can reuse instance multiple times to get descriptor changes since the last run.

commit 16feb65cb0b8141137a14ab0cdde79ed79ed6f5a
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Mar 12 08:18:57 2012 -0700

    Being more careful to remove temporary test files
    
    When tests failed they could leave behind temporary files which, in the case of
    the 'secret_file' (which lacked read permissions) this would trip up future
    test runs.

commit 77ac27cbe057e9411c28d0a8be814046465d335b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 11 15:20:50 2012 -0700

    Integ test for files skipped due to permissions
    
    Adding an integration test that listens for a file which is skipped because
    the user lacks read permissions.

commit 827d3495084f762d4e2e0df3253fea00fc532707
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 11 15:13:57 2012 -0700

    Integ test for skipping files that don't exist
    
    Adding an integration test that listens for a file which is skipped because it
    doesn't exist.

commit 9f8683f182b1199ba1eb6955130e46db0bae5372
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 11 15:05:54 2012 -0700

    Integ test for skipping a file due to its mime type
    
    Adding an integration test that listens for a file which is skipped due to a
    type indicating that it doesn't contain descriptor data.

commit 3bf5098545db24eb178e6b6670463616f284ec96
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 11 14:39:01 2012 -0700

    Implementing skip listeners and exception hierarchy
    
    Notifying the skip listeners of files that we skip, and adding an exception
    hierchy for the reasons why. Also a little minor refactoring to move the file
    handlers out of the run() method.

commit ee5fa60dfb96b4ae0bc8f5a18aa244b4b5f296c8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 11 13:58:00 2012 -0700

    Integ test for the set_processed_files() method
    
    Testing that the DescriptorReader's set_processed_files() method properly
    causes us to skip priorly read files.

commit 664552e72719776f17cd654928dbf5d0a12e48d9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 11 13:41:48 2012 -0700

    Fixing sinister concurrency issue
    
    Replacing the _is_reading event flag with a 'FINISHED' entry in the
    _unreturned_descriptors queue. This is because python's queues stupidly have no
    notion of flushing, so there's no method for me to make a reliable check of 'if
    the reading thread is finished AND the queue is empty'. I may have called
    'put'. I may have a proveably not-empty queue. But can I make that check work?
    Nooooo. That is... frustrating. >:(

commit 2f8a3c838fd63a83a6087313b5a3f52160d2e4fb
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 11 13:17:09 2012 -0700

    Running test_basic_example() multiple times
    
    Rerunning the test_basic_example() test fifteen times to make it more sensetive
    to concurrency issues. It now fails consistently! (... yay?)

commit 1bfadbca0ae379b62c140193883a833519c4a887
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 11 13:07:12 2012 -0700

    Integ test for get_processed_files()
    
    Integration test for the get_processed_files() method of the DescriptorReader
    class. This and the test_basic_example tests are intermittendly failing right
    now which indicates a concurrency issue - though this test itself has passed
    and should be fine.

commit 0c021dc30ac9b4066e7c647f4a60bfda4511c9c3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 11 12:54:50 2012 -0700

    Only provide absolute paths with get_processed_files()
    
    The DescriptorReader's targets can be relative paths, and this is fine.
    However, the load and save functions for processed_files expect absolute paths
    so making get_processed_files() expand relative paths.
    
    I'm on the fencepost about if we should do this or make the load/save functions
    accept relative paths. Going with this until someone produces a use case where
    this makes them sad.

commit e5e0c3895c0cc9ce540e37da6eec9d88e14bdeda
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 11 12:50:56 2012 -0700

    Tidying up DescriptorReader attributes
    
    Bit of refactoring to limit the scope or improve readability of attributes.

commit c16df469c9d49f50ed07bdb8b6c823e8aa31f6ce
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 11 12:39:47 2012 -0700

    Inverting the 'finished_reading' flag
    
    Code is more readable when flags say what we're doing rather than what we're
    *not* doing.

commit 932ded67f961c67c515750bd6e1c9ddb735f8559
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 10 23:30:28 2012 -0800

    Fix and test for DescriptorReader stop() method
    
    Adding an integraion test and some fixes for the stop() method of the
    DescriptorReader class.

commit 0aba5ba4704acb620915ba5f2115b7d036c9477a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 10 22:55:50 2012 -0800

    Finishing basic DescriptorReader implementation
    
    Fixing the minor issue that was preventing the DescriptorReader from working
    and making its first integration test do automated verification, rather than
    needing a manual check.

commit 2d620558fa0885c532e8759c497881146a89ba0a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Mar 10 18:09:52 2012 -0800

    Implementation of DescriptorReader concurrency
    
    After much hair pulling figured out a relatively producer/consumer simple model
    for this class. It should be trivial to add stop() later, but making this
    re-runable would greatly complicate the class and probably isn't worth it.
    
    This isn't yet working, but this is a decent breaking point.

commit 50ac2381a069a9e72094ff210954f9db1bdad0cb
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 9 20:12:42 2012 -0800

    Integ test for basic usage example
    
    The header documentation contains two examples - the equivilant to 'hello
    world' with this class and an example of making use of the high water mark the
    DescriptorReader tracks. This adds an integ test for the first.
    
    At present this integ test dies in a fiery... well, fire. This is because the
    DescriptorReader isn't yet implemented. That's next...

commit 02d51e37e54aeefc77d906792d12e1bb711b89c5
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Mar 9 19:51:38 2012 -0800

    Rewriting reader documentation
    
    Moving all of the examples and substantial documentation up into the header,
    and minorly revising how I intend for the module to be used. Not all of the
    methods in the header have been added yet.

commit 86b90e84e26ac91e1f876ec96df4d983c5ab384f
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Mar 8 09:04:47 2012 -0800

    Integration tests for save_processed_files()
    
    Basic integ tests for the save function, checking that we can load persisted
    files and that it has some basic input validation.

commit c547ed5688da5015dbddbb091118e351fef31bdd
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Mar 8 08:27:51 2012 -0800

    Integration tests for load_processed_files()
    
    Basic set of integ tests exercising file reading when successful, non-existant,
    or lacking permission.

commit 5e6623ce92e0bee08df9af18aacdd30548d47ad7
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 7 20:28:44 2012 -0800

    Unit tests for the load_processed_files() function
    
    Unit tests for stem.descriptor.reader's load_processed_files() function. This
    mocks the open call to simulate file reads and exercise a variety of good and
    malformed contents. This does not yet test for use cases where the file doesn't
    exists or fails to be read due to permissions, but those will be integ tests.

commit 5f8064b3ba8f47162dc781a5804839c029599d14
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 7 20:20:09 2012 -0800

    Typo in import statement
    
    Basic syntax issue (import typo)

commit cbc0cb1a509911f89e8b14428efa65f0aa5b22b6
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Mar 7 20:17:29 2012 -0800

    Adding mocking support for builtin functions
    
    Builtin functions, like open, lack the __dict__ attribute among other things
    which prevents the current mocking scheme from working. Adding workarounds so
    we can accomidate them.
    
    This also adds a function for supporting the 'with' keyword on mock objects.

commit d93d894a3339216b7dc1791ea8c009cd280db7ff
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 4 11:24:24 2012 -0800

    Revising implementation details for processed files
    
    Like metrics-lib, the DescriptorReader will track the last modified timestamp
    for the descriptor files we have processed. Revising the pydocs for it and
    adding untested functions to save and load. Next step is to add integ tests.

commit 5a7bb2584ec976b64a91986e5aa8ca60d8ab60a0
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Mar 4 01:22:36 2012 -0800

    Skeletion for descriptor reader
    
    Initial skeleton for a class to read descriptor data, not including most of the
    functional bits. This will be similar to the metrics-lib DescriptorReader
    class, but with a slightly different API and the ability to handle tarball
    archives.

commit e447d3e24cc852795dce9b77da703c2ec77cd2d0
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Feb 21 15:36:14 2012 -0800

    Fixing an unexpected issue during test interrupts
    
    When we get a KeyboardInterrupt (ie, the user presses ctrl+c) we sometimes
    don't have a tor process to kill which results in an unexpected stacktrace.
    I'm not quite sure why, but if a process doesn't exist then... well... guess
    we don't need to kill it.

commit 9a7b383c18e42cbce093803e8cdd04d98aa900c5
Author: Ravi Chandra Padmala <neenaoffline@gmail.com>
Date:   Wed Feb 22 10:13:52 2012 +0530

    Stop running (all) integ tests on receiving a SIGINT. Fix #5199

commit be32e6a5017b220643ee4df4aace3d81a6acdd73
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Feb 19 16:33:42 2012 -0800

    Fixing another close() deadlock issue
    
    The previous fix narrowed the window where close() / recv() calls could trigger
    deadlock, but it didn't eliminate it. I'm adding another test that reliably
    triggered deadlock in that case and narrowing the window even more (which fixed
    the issue).
    
    I'm a little worried that this doesn't completely eliminate the issue since
    there is a theoretical race if recv() calls close after someone else calls
    close() but before they set the boolean flag. That said, I'm not sure if this
    is really an issue in practice.

commit d5162f4e369f5d2226f0e5262f2b63025f9a68ec
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 18 18:15:07 2012 -0800

    Fixing deadlock in BaseController
    
    Found two concurrency bugs which were causing deadlock issues, and adding a
    test that's more likely to trigger connect() and close() concurrency issues.
    
    The issues were...
      * The recv() method calls close if the socket is still flagged as being
        alive. Unfortunately this can cause deadlock if the closing thread joins
        on the recv thread.
    
      * For some reason using a Condition rather than an Event caused the event
        loop to sometimes miss the notice that caused the event thread to close,
        causing its join() call to get stuck.

commit fa2aeb917ed84433abc76ea77f5794ad695621d7
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 18 16:47:48 2012 -0800

    Async event handling for BaseController
    
    Implementation for the BaseController. This continually pulls from the control
    socket, providing three things...
    
    - asynchronous events are sent to a callback
    - msg() function which sends a message and provides back the response
    - functional is_alive() check (we need a continuous puller to know when the
      socket is closed)
    
    These are done in a similar fashion as the TorCtl class except that I'm aiming
    to provide thread safety. That said, I haven't got it right yet. Controller
    tests inconsistently fail with...
    
    - deadlock
    - seg faults (... not good, indicates a python bug)
    - occasional nonsensical stack trace on shutdown:
    
      Exception in thread Tor Listener (most likely raised during interpreter shutdown):
      Traceback (most recent call last):
        File "/usr/lib/python2.7/threading.py", line 552, in __bootstrap_inner
        File "/usr/lib/python2.7/threading.py", line 505, in run
        File "/home/atagar/Desktop/stem/stem/control.py", line 389, in _reader_loop
      <type 'exceptions.AttributeError'>: 'NoneType' object has no attribute 'socket'
    
    Needless to say, all of this needs to be fixed. However, I've been banging my
    head against these concurrency issues for days so I should finally commit what
    I have.

commit 4ff7efe297f2a076e1b4dc06a3686fdcbf260f8a
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Feb 13 08:59:11 2012 -0800

    Separating BaseController into standalone class
    
    The BaseController was previously a ControlSocket subclass because it was
    easier to use when several of its methods were accessable. However, from an
    implementation perspective the BaseController was a wrapper class, not a proper
    subclass.
    
    After experimenting a bit more I realized that I don't want the BaseController
    to provide *all* of the ControlSocket methods. In particular, it doesn't make
    sense for callers to use the send() and recv() when there will be a msg()
    method similar to TorCtl's sendAndRecv(). The wrapper/subclass mix was also
    just plain old confusing as an object-oriented design.
    
    I'm moving the notifications up to the controller so the ControlSocket is very
    similar to how it was a couple weeks ago. Also dropping the passthrough integ
    tests since most of them will break without send/recv - I'll need to add some
    more targeted tests for passthrough methods later.

commit 1b1d48c37ae9092fdb93a76a3bc6201dad3113e4
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Feb 9 10:32:43 2012 -0800

    ControlSocket listeners for state changes
    
    Adding functions to let users subscribe for state change notifications. This is
    something that I've found highly useful in the past, and will also be needed
    for the BaseController implementation.

commit 8f92a27705e78ec6751b6974f6f31d772137b026
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Feb 8 20:24:10 2012 -0800

    Putting all locks under a 'with' clause
    
    At first I was dubious of the usefulness of 'with' keyword gsathya showed me.
    However, now that I've discovered that locking can be done under it I take that
    all back - it's a wonderful, wonderful thing and I don't know how I got by with
    manual locking/releasing before.
    
    ... and then they ate Sir Robin's minstrels and there was much rejoicing.

commit db0e9ce6d969674a6a46d24d250fe8acd6245af3
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Feb 8 19:35:36 2012 -0800

    Raising NotImplementedError for _make_socket()
    
    ControlSocket is an abstract class without an implementation for
    _make_socket(). In python the proper type of exception to raise for this is a
    NotImplementedError.

commit 81fc285ff9c6e257313115d9b1e489966e172fc8
Merge: a3cddb0 e3b78e8
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Feb 8 19:17:32 2012 -0800

    Merging gsathya's stem.util.conf save() improvements
    
    Work is being tracked in...
    https://trac.torproject.org/projects/tor/ticket/4913

commit e3b78e898dcfd4d33356c7e791fb92e6c7128484
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Feb 8 19:15:05 2012 -0800

    Cleanup for stem.util.conf's save() method
    
    Minor changes including:
    - fixing a 'self.path' assignment to be 'self._path'
    - making the path optional for load() as well
    - raising a ValueError for both load() and save() if we use an undefined path
    - whitespace fixes
    - added argument to save() pydoc
    - replacing integ test with one that explicitely tests the three types of
      values it might handle

commit dd29415f27eb509ca496bc2159258a3a6b5d7774
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Thu Feb 9 01:16:51 2012 +0530

    Preserve structure of config file while saving
    
    Previously the single line config entries were
    converted to a multi line structure and then
    saved. Now, the structure of the config file is
    preserved.
    
    Fixed by atagar

commit 3f51e6920c1dc6dc28d7f70115cd30ed8c020b29
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Thu Feb 9 00:20:58 2012 +0530

    Added integ tests for config.save()
    
    Tests check save() with single and multi line
    configuration files.

commit d48aadd31678e6a4dd670a4e23832a3d563c4072
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Thu Feb 9 00:05:19 2012 +0530

    Updated config.save()
    
    - Added Pydocs
    - Added handling of multi-line entries
    - Added path argument
    - Changed single letter variable 'f' to 'output_file'
    - Instead of calling sort() on config.keys() and
      iterate over it, we pass config.keys() to sorted().

commit a3cddb03bd4ccd1dc3824bf2ec76695bde3b9322
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Feb 7 09:47:10 2012 -0800

    Replacing send/recv conditionals with RLock
    
    I'm not sure if threading.Conditional is re-entreant, but RLock definitely is
    (that's what it's there for). Quite possable we had a concurrency issue due
    to this, though if so I've never ran into it.

commit 398cbfb93a4982961e1ce27fc05d82265408e78a
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Feb 6 08:16:59 2012 -0800

    Helper constructors for BaseController instances
    
    A controller is a wrapper around a ControlSocket instance so adding convenience
    methods to construct both the socket and controller at the same time. These
    will belong to the Controller class later.

commit a500dbc23d7b3f012a2040d104c3d009b36b3694
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 4 20:23:02 2012 -0800

    Pass-through BaseController class
    
    Controller will be a ControlSocket subclass so making it a pass-through to
    the socket they're constructed from to start with. Testing this turned out to
    be a pita for a couple reasons...
    
    - The mock() function can't handle class methods, nor is it possable to add
      method handling to it.
    
    - Ideally the controller would dynamically pick its parent class based on the
      socket it's constructed from. However, this is impossable (actually, I don't
      know of any language where instances can dynamically define their class
      hierarchy). This mostly just becomes an issue for isinstance checks.
    
    Dropping the old, scrap controller implementation for now. Some of its useful
    bits will come back later.

commit 5c01e29839a084018185ce005c9ec90788e13fce
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 4 18:05:31 2012 -0800

    Stop tracking reverted mocks
    
    When we reverted function mocking we didn't clear our registration of it. This
    caused the revert_mocking() function to attempt to revert things that weren't
    still mocked. This didn't cause any problems since doing this was a no-op, but
    still good to fix.

commit 2076447cb6b86e02bf9bb0cfc666f7568a8f4d00
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Feb 4 18:03:27 2012 -0800

    Integration tests for the ControlSocket class
    
    The ControlSocket subclasses were indirectly tested via the ControlMessage
    tests, but we missed several interesting use cases. Adding more testing for the
    ControlSocket instances and correcting some of its behavior with respect to
    reporting socket closer while sending messages.

commit 2ae0298e816ccb70bf9ab52996a4c78e26f897f0
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 31 09:23:31 2012 -0800

    Exercising is_alive in ControlMessage integ tests
    
    Adding a basic check for the is_alive method. The method is a little troubesome
    because if you aren't continually pulling from the socket then you won't detect
    a disconnect until the following usage.
    
    This was already documented and iirc it won't be an issue in practice for
    controllers since they continually pull from the socket. But still, for low
    level controller interaction it's kinda confusing.

commit 06498d8b61ad5f01c04c7011c70892a9846d4c27
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 31 07:46:53 2012 -0800

    Adding ControlLine peek_key method
    
    Method to allow us to switch on the key/value mapping while parsing a line.
    This isn't needed yet (the only parsing done at this point is PROTOCOLINFO
    responses and that only uses positional entries). However, it's crossed my
    mind a few times that this will be needed later.

commit c58177c50fcdba85e5eac7a929dbf6d87c666852
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 30 00:34:38 2012 -0800

    Previous socket use could break get_protocolinfo
    
    If prior use of the control socket had caused a disconnect and already consumed
    the 'Authentication required.' response then future calls of get_protocolinfo
    would break.

commit a31aa4ad506ccf4e529182e31d0a84966e38d6e1
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 30 00:02:55 2012 -0800

    Accounting for big-endian architectures in proc
    
    The proc utils were assuming that encoded ip addresses were little-endian. This
    was fixed in the project these utils are based on (psutil) and I'm adopting the
    fix...
    https://code.google.com/p/psutil/issues/detail?id=201
    https://trac.torproject.org/projects/tor/ticket/4777
    
    This evidently occures on OpenWRT (ar71xx), thanks to swalker for the catch!
    
    This is the same as arm commit 8ec7095d79ecad9d3432193a2b1f9fdefab8d7f3

commit f7d18618e20b440b0efaccb6f4a5786842700f9c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 29 17:35:28 2012 -0800

    Removing CALL_MOCKING from the system module
    
    While writing the unit and integ tests I realized that the tests could be
    greatly simplified by adding an override function called 'CALL_MOCKING'
    that would let the tests modify the call function's behavior.
    
    This was a mistake. Production code should be designed to make testing easy
    because this leads to better production code, but adding hacks that are only
    useful for testing is bad. Now that we have a nice mocking module the
    CALL_MOCKING hack is no longer necessary so removing it.

commit 5e81a0b546413f6ef21266cb74b81e9be4c6a27c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 29 17:31:53 2012 -0800

    Refactoring system integ tests
    
    This is the last of the integration tests to be cleaned up. It's also the last
    user of the system module's CALL_MOCKING so now we can get rid of it - hazaa!

commit 0f913df18423f84650b27144f2a89fb9997349ab
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 29 15:05:05 2012 -0800

    Refactoring conf integ tests

commit f38ad0811290b51b99475de28628895a663e7dec
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 29 14:40:27 2012 -0800

    Refactoring ControlMessage integ tests
    
    Besides the normal cleanup I'm pushing most of the skipTest checks into the
    runner module. The advantage of this is that it simplifies tests a little more
    and allows a standardized skip message rather than having each module provide
    their own for the same thing.

commit 39672342bf81a5de15f40fe78a3170673f4035de
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 29 13:36:27 2012 -0800

    Renaming version requirements to be hierarchical
    
    Including a prefix with all version requirements for the option's catagory.
    Currently this includes GETINFO_* and TORRC_*.

commit 51e21cf88643e63953e897d0cd05875d00c26991
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 29 00:49:29 2012 -0800

    Refactoring protocolinfo integ tests
    
    Cleaning up the protocolinfo integ tests, joining a couple that had a similar
    prupose and other miscellaneous refactoring.

commit 39d93f75533243948d3222f800707dcb320b1a76
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 28 23:42:14 2012 -0800

    Refactoring connection integ tests
    
    More tidying up of the integ tests. The connection integ used a common
    port/socket tester function but the common code was tiny enough that it's
    clearer without it.

commit a73be509009942b4b2711b49a3b30e3c3f6e609b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 28 17:06:36 2012 -0800

    Rewriting authentication integ tests
    
    Excessive testing helper functions are bad. They hurt test readability and
    makes the code a pita to trace through. This cleans up the authentication
    integration tests to collapse the helper functions into just the few that make
    life better. Much more maintainable now. \o/

commit dbf8663cf9b5ca1e31f516109b019045ec63a04c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 28 13:14:28 2012 -0800

    Integ check for value of get_system_tor_version()
    
    We had basic verification tests for get_system_tor_version() but it didn't
    check the value. This was a refactoring mistake, iirc it once checked the value
    but since we can run with an arbitrary tor binary I removed that, forgetting
    that get_system_tor_version() could optionally take that binary path.
    
    Making this a separate test since it relies on the tor instance being
    accessible (which isn't the case with all targets).

commit 6f70bd329ab215754241ea56479a1c9490ab2dce
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 28 12:55:26 2012 -0800

    Tidying the stem.util.system unit tests
    
    Improving the mock call function, clarifying usage in its pydocs and bundling
    it with the partial call. I'm kinda impressed how much this simple change helps
    the readability of these tests...

commit 9b5e2295a2217081cf772ce3b0f24f88eb3dac34
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 28 12:06:29 2012 -0800

    Providing higher test runtime accuracy if low
    
    It looks kinda odd to say that the unit testing runtime is "0 seconds", so
    raising the accuracy when the runtime is less than a second.

commit db9d695f16dc2f2ec409be40aa69d7d13cf24826
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 28 11:22:38 2012 -0800

    Enum constructor function for capitalized values
    
    On occasion we want enumerations where the values are capitalized (for instance
    logging runlevels and integ targets). This could be done in a succinct fasion
    via a bit of python hackery but it wasn't pretty. Providing a function to do
    this instead.

commit 7fb7946939ed4801cfe79adbd6f9a0702f4464b2
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 28 11:04:11 2012 -0800

    Renaming assert_message_parses to be private
    
    Very minor correction to the ControlMessage unit tests.

commit 92cf5eb531ab7bfda4361f15b8c5eba92b2050f8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 28 10:54:08 2012 -0800

    Removing pydocs from test classes
    
    The documentation in the headers and classes for unit and integ tests were
    redundant. Joining the two and removing pydoc blocks from the tet classes.

commit 9a2b50138356f317220819b2351f666bd45564e1
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 27 18:53:39 2012 -0800

    Using mocking module for protocolinfo unit tests
    
    Expanding the mocking module to provide a ControlMessage (very frequently
    needed in tests) and using the module to simplify the protocolinfo unit tests.

commit 39e9f0cfdf8efd6cd8445e33feaeb9dd4a44d92f
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 27 09:24:59 2012 -0800

    Rewrite of authentication unit tests
    
    The unit tests for stem.connection.authenticate* functions were the first to
    use the decent style of mocking that was later used as the basis of the mocking
    module. Using the module to clean up the unit test and making a stab at making
    the tests less confusing.

commit 55c3fa6e6bfa90e47f116029a08999fe5cc4ac2a
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 25 09:54:07 2012 -0800

    Target accidently included with early test output
    
    When printing errors an earlier change included the target with the errors
    listed in the footer. This also changed the errors listed earlier when running
    the tests which was a mistake.

commit 030579292593a9e8262edfc713e1d86b379dabf5
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 25 08:48:48 2012 -0800

    Dropping Config.sync() in favor of config_dict()
    
    The Config class' sync method was a step in the right direction, but the name
    was confusing and the usage was suboptimal. In the vast majority of cases the
    caller simply wants a dictionary that stays in sync with the configuration. The
    config_dict() function is essentially the same as sync but with more succinct
    calls.

commit 9e3834d80b0ae10f8ab5076e00186ddd346b39f5
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 25 07:36:11 2012 -0800

    Renaming the Config update method to synchronize()
    
    Calling this method 'update()' has confused a couple people. It changes a
    dictionary to match the key/values in our current configuration. Calling
    this synchronize() makes more sense though now I should change the sync()
    method - that's next.

commit 2ae0452fb59ef5781e0803919b02cc0f3fa5212a
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 24 09:47:51 2012 -0800

    Integ testing for multi-line configs

commit 1f5b05766ce0c2533d8bd2861d142cc6523e4410
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 24 09:26:59 2012 -0800

    Config support for multi-line entries
    
    Adding support for multi-line values in the configuration files. I've wanted
    this for a while, mostly for user facing messages. This still needs an integ
    test.

commit bda94c4dffa97f9cc915cddb45d90aab51e8e5b5
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 23 19:11:07 2012 -0800

    Including the test type or target with errors
    
    When we have a failed test run with numerous targets we know at the end that an
    error occured and what it is, but not what target it occured under. It's a pita
    to scroll back looking for a stacktrace so including this with the footer's
    error information.

commit 1b71039087a0420e4138e4eb61291f6b6a752dd6
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 23 09:48:05 2012 -0800

    Integ RUN_NONE target wasn't runnable
    
    The RUN_NONE target wasn't being included among the runnable integration
    targets because it had an empty torrc which failed a check that should have
    been for the attribute's existance instead.
    
    Also correcting the case of the 'TARGETS' enum. While it is a contant, I've
    been going with class case contentions for enumerations.

commit 59764266b9dbce91accd6a55a371f24516b257a7
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 23 09:25:46 2012 -0800

    Unit testing for stem.util.conf
    
    I needed unit tests for the new listener functions so wrote long overdue tests
    for everything within the utilitity too. Caught a couple bugs in the process.

commit cfb2c383fc487f4a129345c5724b6a73d96c712d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 22 23:03:41 2012 -0800

    Implementing testing --no-color arg
    
    An option for suppressing color escape sequences is important for when test
    output is piped to a file (the escape characters then just look like
    gibberish). Still... output is sooooo ugly without color. :(
    
    Routing most of the testing stdout writes through the output module, and
    either respecting or discarding formatting there based on if we had a
    '--no-color' argument.

commit 92f97544455aee94a229daee20f099f14f92829e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 22 22:24:53 2012 -0800

    Eliminating the 'test.' config prefix
    
    It's redundant for testing configurations to have a 'test.' prefix.
    Configurations are already segregated by their handle so they would be only
    accessed via the 'test' config anyway.
    
    My heart isn't set on these config hierarchies but this will do until someone
    proposes something better.

commit 0a5ff78074309be31ce6f71fde2b0eca86a8c3e6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 22 22:15:52 2012 -0800

    Synchronizing runner config
    
    Using the shiny new sync() method for the runner config to make the world a
    better place.

commit e7cb3b59298088b4a8adf35364ff6d775559d27a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 22 19:10:41 2012 -0800

    Overhaul of run_tests.py
    
    Several substantial changes to the run_tests.py script to improve readability:
    - splitting arg parsing from the rest of the main function
    - adding a config sync method to keep config dictinaries in sync with the main
      configuration (this will be especially important for arm since it allows for
      proper runtime configuration editing)
    - moving remaining print functions into test/output.py
    - lots of general cleanup
    
    Remaining todo items from this...
    - still need to add testing for the config listeners
    - we should note the module that failed with the failures at the end
    - we still need multi-line config entries
    - the --no-color option was added but not yet implemented
    - the RUN_NONE target looks to be broken

commit a0be1932e96e9e52a12a4bbc49e72a059c005f78
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 22 13:16:01 2012 -0800

    Expanding Config class' set() method
    
    Users of the set method would usually expect it to overwrite our current
    configuration value rather than append to it (thanks to Sathyanarayanan for the
    catch). Also expanding it to handle list or tuple values.

commit d465a9162c58915cbf41c544844cd3520329e3e3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 22 11:53:35 2012 -0800

    Renaming CONN_* targets to RUN_*
    
    The former CONN_* targets each represent a run of the integration tests. These
    were originally called CONN_* because they defined torrc parameters for
    connecting to tor, but as time went on they became more general. RUN_* (for
    instance RUN_ALL) is far more intuitive to users.
    
    Also dropping the 'integ' prefix from target configurations. It's accurate, but
    pointless to have. Target options are used frequently enough that we shouldn't
    make them more verbose without a reason.

commit fa934a0c842b86788203199e7dff57a85d7c5387
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 21 23:07:04 2012 -0800

    Minor runner refactoring
    
    Messing with the order, dropping the get_torrc function, and using STEM_BASE
    rather than figuring this out again in 'run_tests.py'.

commit 6c275ae49fae5ca45c8e596b6119841458a95fb6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 21 22:37:24 2012 -0800

    Renaming get_connection_options() to get_options()
    
    The Runner.get_connection_options() method is no longer really connection torrc
    options, but rather all custom options for our instance. This is frequently
    used so having a shorter name is nice to have, too.

commit d7fa1891aaac5f5907cbff05a8fd2043155b475f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 21 22:14:25 2012 -0800

    Replacing runner OPT_* with enumeration
    
    Replacing the test.runner.OPT_* constants with an enumeration. This is a little
    more elegant and lets us get rid of the cluncky RUNNER_OPT_MAPPING dictionary.

commit 6c3fb340a83865f75399c60b4e9201810f60c8ee
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Jan 21 15:34:43 2012 -0800

    Moving test target attributes to config
    
    Moving the attributes for integration testing targets from a big dictionary in
    the source to a separate configuration file. Progress!

commit 34da3669909c5e89957bcf77c92f2a00714f921c
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 20 10:20:07 2012 -0800

    Overhaul of integ testing targets
    
    The integration testing targets had grown organically and were partly defined
    in mulitple places. Now that I've played with them a bit I've realized that
    testing targets are purely a mapping between user input and runner attributes.
    Hence I'm removing all notion of testing targets from runner.py, and instead
    having the run_tests.py handle all conversion between test targets and runtime
    attributes.
    
    All target attributes are now generalized and stored in a large dictionary.
    Good news is that this makes code readability and maintainability far easier.
    Need a new target? Just add an enum and the attributes to the hash. Down side
    is that this is more static configuration than code, and run_tests.py would
    be far nicer without it.
    
    I'll next look into moving this to a test_settings.cfg to abstract code from
    testing configuration but references to constants (test.runner.OPT_* and
    stem.version.Requirements) will take a little thought.

commit 6846ce359ede2c697688a2e143a3b3625f8a540b
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 16 10:09:46 2012 -0800

    Removing test.runner.get_connection_type()
    
    Integration tests used the get_connection_type() method to either query or
    infer attributes about the connection it was running against. This was stupid -
    they should query the attributes directly and make test assertions based on
    that. The get_connection_type() value was our testing target which should
    simply be a user friendly tag for a set of testing attributes. Next I'll be
    moving testing targets completely from the runner so they only exist in
    'run_tests.py'.

commit 94479aadebcaeee26d57b91e4900dad56edaeb91
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 15 22:13:32 2012 -0800

    Minor fixes for conf.save()
    
    Sorting the keys prior to saving, handling multi-value keys, and trivial
    changes (whitespace, reordering, and putting the strip call in the try block
    which is the only place it's needed).

commit 451e13eb9552ab448072b863b180a338a4baeb24
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Sun Jan 15 23:16:06 2012 +0530

    Add stem.util.conf.save()
    
    This saves all the contents in the config file.
    It overwrites the contents of the original
    config file. The comments of the original file
    will be lost.

commit 6c5f0205682877c76d6633abce487673787f5d9b
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Sat Jan 14 21:32:31 2012 +0530

    Minor update to stem/util/conf.py
    
    - More pythonic way of doing File I/O
    - Use try and catch the exception while
    parsing the config file instead of using
    an if-else logic.
    - Log messages when the (key, value) are
    not provided in the proper format in the
    config file.
    - Typos

commit 19491fb8a411a5737ca438b9dc943a13f8eb44b8
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 13 09:15:49 2012 -0800

    Integ target for running with ptrace
    
    Integraion target that runs with 'DisableDebuggerAttachment 0'. This allows us
    to exercise use cases that require ptrace with newer tor versions. If
    'DisableDebuggerAttachment' isn't supported then we skip the target.

commit 4a0e1a755e43802340020b663f6982536bd57ecf
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 13 08:46:19 2012 -0800

    Skipping integ that need ptrace when inaccessable
    
    When the tor connection is inaccessable I'm skipping the system integ tests
    that need ptrace support, since we can't check if it's enabled or not.

commit 0071fadd8b69eda81cdd9826719037fafdebbb74
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 13 08:28:32 2012 -0800

    Making explicit checks for tor accessibility
    
    Several 'test/runner.py' functions query against our tor instance. However, in
    some integ tests we don't have any method for doing that. Rather than returning
    a default value I'm raising in those cases, and adding an is_accessible method
    to make our tests more readable.

commit b01a790b902bfc8a0bb368d6b5bc72c52ff528af
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Jan 13 07:43:15 2012 -0800

    Placing integ control socket in tor owned dir
    
    Tor's new version (0.2.3.10) checks that the control socket is in a directory
    that's only accessable by the tor user, and refuses to start up otherwise.
    Accounting for this in stem's integ test startup.

commit 27e84652f1bd97747a1360f7632783923cf762b2
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 12 09:51:40 2012 -0800

    Making integ account for DisableDebuggerAttachment
    
    Stem's system integration tests failed with tor's new version because
    DisableDebuggerAttachment prevents cwd and port lookups. Skipping those tests
    when that option is set.

commit 479a620ee8d3351e07d1bc86bec3f1107c04cc69
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 12 09:17:20 2012 -0800

    Making lsof cwd checks account for permissions
    
    With the anti-debugger tor change the /proc/<pid>/cwd symlink becomes only
    readable by root. This causes lsof queries for a process' cwd to return the
    proc path with a message saying that permission is denied, and the sys tools
    misparsed that as meaning that the cwd _was_ the proc path.

commit 13c3975bc3585847dd204d074a66394786bf2007
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 11 10:20:30 2012 -0800

    Testing argument to run with a tor binary
    
    Adding a '--tor PATH' argument so the caller can select the tor binary they
    want to run against.

commit 11cbf61c2dca7cde628e5eec4f3dd4da7867860f
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 10 09:49:26 2012 -0800

    Mocking utility functions
    
    As the tests grow I've found that I'm repeating a lot of mocking logic. Making
    a nice, general purpose utility to make mocking easier and less error prone.
    
    A benefit of this (and the original goal) is that we'll be able to remove
    stem.util.system.CALL_MOCKING. Testing code in the library is icky, and once we
    swap over the system integ tests we'll be able to remove this override
    function.

commit 608a5371370796a3d6ace5deb54bc019783cf2c9
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 5 22:42:33 2012 -0800

    Moving log handlers to log util
    
    Adding a couple basic log handlers to the log util...
    log_to_stdout - sends further events to stdout (mostly intended for debugging
      at the interpretor)
    LogBuffer - simple handler used by the testing to buffer logged events so we
      can run over them later

commit 1f13b0b665ca22fe11febeeb3d610f5852a3ba49
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 5 21:23:12 2012 -0800

    Replacing split("\n") with splitlines()
    
    Oops, didn't spot that there was a str method for splitting on line breaks.
    This behaves slightly differently with respect to trailing newlines and
    preserving '\r\n' entries, but is something I should have been using...

commit 21e8f65ad3e981f0060eb0fc71c2c432f32194b0
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 5 18:39:41 2012 -0800

    Improving system TRACE logging
    
    Showing request/reply with a debug level message for system calls with runtime
    is redundant. In case of a call failure we just need to give a debug, and in
    success we show the debug followed by an improved trace message.

commit ed68e97e2d6a27ee979bf75b1e1b7f3df2d6c8d0
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 5 10:39:09 2012 -0800

    Skipping unreliable lsof test on osx
    
    On OSX the Quicklook process sometimes claims '/tmp' and possibly other common
    paths too, making the test unreliable there. Skipping is better than a false
    positive to avoiding the test when running on OSX.
    
    Caught and patch submitted by gsathya

commit b7e9f5c9314159679c73c707e3a6a15f5ff03fdc
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Jan 5 10:34:09 2012 -0800

    Implementing TRACE level logging
    
    Adding request/reply logging for tor and system calls. I'm also messing with
    the runlevels and messages of other logging to make the integ test output more
    readable.

commit ad5483352456ab3556068baa43db4fa97c008760
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 4 17:41:10 2012 -0800

    Indenting multi-line logging output
    
    When sending logging to stdout indenting the lines in multi-line log entries to
    make them far more readable.

commit 228a431b06b2a89182335216b4d9cfce166c6ce9
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Jan 4 17:34:50 2012 -0800

    Registering TRACE and NOTICE level names
    
    Registering the custom TRACE and NOTICE runlevels with the logging module so
    their names will appear when we log a runlevel.

commit 4788095929ab2a3bfde04ade0b2cdf52fe82dc82
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Jan 3 18:52:14 2012 -0800

    Tidying up lower runlevel logging
    
    Cleaning up the log output for DEBUG and INFO runlevels we encounter when
    running the unit tests to be more helpful and user friendly.

commit 4cd8ac450fd0f3c80a74c4f7835bee1fd2644ef5
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 2 20:39:50 2012 -0800

    Making integ test ps query just match 'tor'
    
    The fix in c421b54 checked for endings of 'tor' rather than ' tor', causing
    things like 'monitor' to be picked up.

commit c421b54e52d181e14a6677752fe2c07d2b56a796
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Tue Dec 20 17:13:53 2011 +0530

    Remove awk and grep from ps call
    
    Instead of piping the output from the ps call
    to grep and awk, use python itself to parse
    the result.

commit 8c8ad75bd1e3dc5541273336b03d6300f689b66e
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Mon Dec 19 22:23:41 2011 +0530

    Replace pgrep with ps in the BSD systems
    
    Now, the integ test checks if the os is a BSD system with a
    call to stem.util.system.is_bsd(). If it is BSD, then it
    uses ps to check if an extra Tor instance is running, instead
    of using pgrep which does not exist on BSD.

commit 29652d978ce7465f288575061c819e77a7e67013
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 2 16:42:14 2012 -0800

    Dropping get_system_tor_version() usage from tests
    
    At one point our integ tests was using the get_system_tor_version() function to
    determine our test instance's version. This is fine for now since we run
    against whatever tor binary is first in our path, but in the future tor devs
    will want to be able to select the binary so that won't work.
    
    Adding a test runner method for querying our version from tor instead.

commit a08c09a21743225181569a9ba08b3b15a7c584e2
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 2 16:27:48 2012 -0800

    Integ tests for stem.version
    
    I didn't have any integ tests for versions? Seriously? *sigh*
    
    Adding some basic verification tests that we can handle the version of our
    system and test instances. Earlier boerni was running into a bug where another
    integ tests inadvertantly failed due to an uncaught exception from
    get_system_tor_version(). It was inappropriate for that test to fail (it wasn't
    testing versions) but we should have some test for the function so this is it.

commit b4f86854437d0e81cc1adb18de66ca46c49dabb4
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 2 14:49:12 2012 -0800

    Integ for cookie auth failed due to missing cookie
    
    If we're running the integ tests for the first time then it would fail because
    it attempts cookie auth but the cookie file doesn't exist yet. This went
    uncaught for quite a while because after the first test run we leave the data
    directory (to speed up future test runs). Caught thanks to boerni.

commit ea511db40259a74d57d87a40339b41e66fe34d9a
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Jan 2 14:32:13 2012 -0800

    Uncaught IOError when checking version in integ
    
    If our attempt to look up 'tor --version' failed then we'd trigger an uncaught
    IOError. We're only checking the version there to determine if we should skip
    the test due to tor being out of date, so making that a best-effort lookup
    instead. Caught by boerni.

commit 4a5586afb64d50c627652c1ead9314596cb0b36c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 1 10:24:38 2012 -0800

    Deduplicating unknown auth type messages
    
    There's quite a few log messages that really don't make sense to repeatedly
    log. Adding optinal logging deduplication and applying it to the INFO level
    message about getting an auth method we don't recognize.

commit ad77d82535a5d86cb9b059d2365ff70cfa260c53
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 1 10:12:18 2012 -0800

    Reducing socket/proctol error logging to INFO
    
    The base socket and protocol errors are too low level to be of interest to
    users. They're interested in the symptoms and what it means for them not that,
    say, a control message was malformed.
    
    Also escaping the logged message content - much easier to make sense of that
    way.

commit 5a988db9a78161240a9f3146936a57dfc0a6734e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Jan 1 09:56:26 2012 -0800

    Downgrading protocolinfo version mismatch to INFO
    
    The response for PROTOCOLINFO queries do not necessarily need to be what we
    requested. We make a best effort to parse any response as a v1 response, and
    warned if the response was for something besides v1. Downgrading the warning
    from WARN to INFO since this is a valid tor response and, if we're still able
    to parse it, the user doesn't really need to be alerted.
    
    If the version mismatch can't be handled like a v1 response then that most
    likely *will* cause warnings or errors that we'll alert the user of.

commit 94e009c8fcba1569b0e2439c626f0031d62a43b6
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 31 13:51:18 2011 -0800

    Using log util runlevels for test runner
    
    Replacing hardcoded logging values with util abstraction.

commit 249035cf3338a10ba1e7e311cb650f10fcbbcb60
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 31 13:34:46 2011 -0800

    Moving null-handler logging hack into log util
    
    Without a handler the builtin logging class emits a warning. To get around this
    I added a no-op handler if none already existed before any logging was done via
    a module import hack. However, now that we're using a log util we can put the
    hack in that instead.
    
    I tried to repro the warning but didn't have any success... maybe the issue
    only exists for older python versions? Reguardless, this workaround doesn't do
    any harm so keeping it around.

commit af9cdc40e9aa3c8ae625dd6941fb63815ad361f3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 31 13:23:01 2011 -0800

    Clarifying that jail ids only exist on FreeBSD
    
    Sebastian and rransom confirmed that jail ids do not exist on OSX and OpenBSD
    so clarifying that the function is only known to apply for FreeBSD.

commit a3f9647832dab69c5eda8047748d0de534fac073
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 31 12:54:49 2011 -0800

    Routing all logging through stem.util.log
    
    Back and forth I go - first I wrote my own logging then used the builtin python
    logging module. Now I'm going to a hybrid of the two. The builtin logging
    module is rich but somewhat hard to customize without a wrapper.
    
    Changes we want on the logging module are aliased commands, additional
    runlevels ('NOTICE' and 'TRACE'), and logging sends complaints to stdout if
    there isn't a listener (very, very dumb of its author and I've been working
    around this via a __init__ module hack).
    
    Functionally this does not change anything. I'll be changing the modules to
    take advantage of these new runlevels later.

commit d84376a1ff3c459de603878763e4fddff019fa43
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 31 12:13:45 2011 -0800

    Making the testing log runlevel customizable
    
    I'm gonna be making stem include two runlevels that it doesn't natively
    support: TRACE and NOTICE. The later is so we have runlevels to match tor, and
    the former is so we can have request/reply logging without cluttering the DEBUG
    logs.

commit b56b9c627332208913b91357c02267c472644fe9
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 30 10:12:57 2011 -0800

    Only issuing WARN for get_bsd_jail_id on FreeBSD
    
    The stem.util.system.get_bsd_jail_id() function is only available on FreeBSD
    (and maybe OSX) so only issuing a warning for failures on that platform.
    Everywhere else this now gives a DEBUG level message to warn the developer that
    the function's unsupported.

commit 08f3061dc7c03a735a27024ad417fb2e4026a7c5
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 30 09:49:28 2011 -0800

    Testing argument to include logging output
    
    Adding a '-l' or '--log' argument that includes DEBUG level logging output with
    the test results. Hopefully by making the log output more visible it'll become
    more helpful.

commit a5023f160e184bb0b800d4789a96c47d411fef2c
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Dec 28 17:51:50 2011 -0800

    Expanding success/failure label and other fixes
    
    Multiple minor fixes and display improvements around how test output is
    displayed.

commit 8795873a50c722419700dec1fff56d289799a1c0
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Dec 28 10:24:58 2011 -0800

    Making test results right aligned and bold
    
    Instead of showing test results as...
    test.foo ... ok
    test.hello_world ... ok
    
    I'm now showing them like...
    test.foo           [OK]
    test.hello_world   [OK]
    
    with the result text bolded. Thanks to Josh Barr for the idea.

commit 285bffdab1c818ee675390701a0539510f41b4f7
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Dec 28 10:01:23 2011 -0800

    Moving testing module from results to header
    
    Dropping the custom header titles and putting the module name there instead.
    This means that the module name is redundant with the test results, which was
    cluttering the output (bad for readability). Thanks to Josh Barr for the
    suggestion.

commit d02aa3e9128e92982e6f2760594523556bd45e85
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Dec 28 09:24:35 2011 -0800

    Moving python testing output filters to module
    
    Rewriting the hacks I was applying to test output for better readability as a
    proper module. Nicer code and now much easier to add more filters.

commit 839df4492c61777eaf660665a2aeb0d4267251bb
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 27 10:24:40 2011 -0800

    Connection pydoc spelling corrections
    
    Just some quick corrections now that this module has reached maturity. I'll
    check the other modules later.

commit ad7d822afb59692686b9fe3c97b39f5a7b5a5d21
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 27 10:18:02 2011 -0800

    Expanding connection header pydocs
    
    Explaining what the module is most commonly used for and moving the example to
    the top.

commit a9a240c553cc154f8019c6332ece686202398c34
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 27 09:58:00 2011 -0800

    Implementing and testing connect_* functions
    
    Similiar to TorCtl.connect(), stem's connect_port and connect_socket_file are
    convenience functions for trivially getting an authenticated connection. This
    isn't ideal for applications since it hijacks stdin/stdout and lacks
    exceptions, however for CLI apps and the interactive interpretor it's very nice
    to have.

commit b7b9ff1f214b87fc84425cb26882ec4f96362736
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 27 05:20:10 2011 -0800

    Reordering header pydocs so exceptions are last
    
    The ordering of the header documentation was based on the order of things in
    the file. This placed exceptions at the start which isn't what readers usually
    want.

commit a7772f8c62957ea46db7c3aa012c0ba6b018a3d0
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 26 16:53:20 2011 -0800

    Defining when auth success should occur
    
    Specifying in the general auth unit tests when authentication should succeed or
    fail. I'd wanted to also check for the highest priority exception but after
    messing with this quite a bit I'm waiving the white flag. This is decent
    enough...

commit e642cb669ebc01055ac1873c709ef57585efe99f
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 26 13:52:17 2011 -0800

    Unit test to exercise all general auth use cases
    
    I'm aiming for both expansive and deep unit testing for the general
    authenticate method and this does the former. It runs the authenticate()
    function under all combinations of valid input and having authenticate_*
    functions either succeed or raise one of its exceptions. At present this just
    checks that the authenticate() function always either succeeds or raises a
    AuthenticationFailure - an important test, but I'll next try to make it a
    deeper validity check.

commit 39528a51489598b970d1f64cd37a3a5de97fc4cb
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Dec 24 17:37:39 2011 -0800

    General auth unit test for protocolinfo fetching
    
    Unit tests for the general authentication function will be exercising it in
    various error conditions by mocking out the functions it uses. This first test
    exercises all of the use cases when the function needs to make a PROTOCOLINFO
    query and, more importantly, lays the groundwork the rest of these unit tests
    will use.

commit 22e64d083ee9a7f15e3f7333477c9c9dde7609cc
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Dec 21 09:37:52 2011 -0800

    Integ test for general auth's password arg
    
    Testing the various use cases for a password argument (missing argument,
    incorrect password, and correct password). When we're running connection tests
    with something other than password auth this is a very dull test (since we
    succeed without raising a password exception), but when we just have a
    HashedControlPassword this is a decent test.

commit 22481b03901f84e9bc9d1951cc0f7343fbc6eb84
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Dec 21 09:00:58 2011 -0800

    Replacing runner authenticate with shiny new auth
    
    The test runner authenticated tor sockets manually for our integ tests.
    Replacing that with our nice, new general auth function.

commit 6d30fe9c9ca67a167b0085306ee5590a26f61d4b
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 20 10:01:02 2011 -0800

    Basic verification integ test for general auth
    
    Simple test that checks that the general authentication function can
    authenticate to all of our tor instances.

commit b4ca09871c7650ad6469c3b998ebc174605f4a7d
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 20 09:41:29 2011 -0800

    Transparently allowing multiple PROTOCOLINFO calls
    
    Tor hangs up on the socket for multiple reasons prior to authentication. One
    of them is if multiple PROTOCOLINFO queries are made. This is annoying and does
    not make sense to controller callers, so transparently re-establishing those
    socket connections.

commit 1483606df372bebbd5c2b6707f777bfb898f21b8
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 20 09:18:16 2011 -0800

    Integ test for general auth example
    
    Testing to make sure that the example in the authenticate() function's pydoc
    works.

commit f1a41ecc8534327772f9eca3430e373a2b8a54b8
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 19 09:52:03 2011 -0800

    Strip whitespaces from is_running ps results
    
    Fix and unit testing for whitespace in the ps results. Also fixing the unit
    test so we make sure that we get False rather than None when we confirmed that
    the command wasn't running.

commit 292a90b6aac0574453f0187f3705e84aee59f808
Author: Sathyanarayanan Gunasekaran <gsathya.ceg@gmail.com>
Date:   Mon Dec 19 22:20:01 2011 +0530

    Fixing the reg exp to accept git SHA1 in Version
    
    Originally the reg exp checked didn't take into account
    the git commit [0.2.2.23-alpha (git-7dcd105be34a4f44)],
    which produced an Error. Now, the status tag consumes the
    git SHA1 as well.
    
    Appropriate test cases have been added for integ testing.

commit def66d45d3cf4c82786fa9a8526a18839fcbf816
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Dec 18 17:25:46 2011 -0800

    Implementing general authentication
    
    Finally finished the fabled general authentication function, which uses
    PROTOCOLINFO information for authentication. The vast majority of the effort
    behind this function was getting the exception hierarchy right.
    
    This is totally untested. Unit and integ testing comes next.

commit 297a41b085a92eb1c5b5c92464d3bc9cd27cef94
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Dec 13 10:06:53 2011 -0800

    Reconnecting socket after auth failures
    
    Tor disconnects the control socket after Failed AUTHENTICATE calls in an effort
    to mitigate the issue discussed in...
    http://archives.seul.org/or/announce/Sep-2007/msg00000.html
    
    This is unintuitive to stem users so I'm making a best effort attempt to
    reconnect the socket in those cases. This isn't guaranteed to succeed, but
    checking is_alive() is far nicer for callers than a try/connect/except block.

commit 4d0d5598eec4936f103293802c217377ada8751f
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 12 18:51:03 2011 -0800

    Joining get_protocolinfo_* functions
    
    Having a joined socket constructor and PROTOCOLINFO query doesn't make sense.
    It allowed the user to make a PROTOCOLINFO query in a single line without a
    leftover socket but that use case is so rare that it's fine for it to take a
    couple more lines. This api is simpler and better for the common case.

commit 8164af5ba0e571dfcd735274e1c4079c79a6fd8d
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 12 18:24:44 2011 -0800

    Using input version string for __str__
    
    Originally we reconstructed version strings from its components since that's
    all the Version class had (a separate factory translated strings into Version
    objects). However, that's no longer the case and much simpler if we just have
    the __str__ method provide the constructing string rather than try to recreate
    it.
    
    patch by gsathya

commit a584686dd01f12c8209ea7bc55019fec752eeaa2
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Dec 12 18:05:23 2011 -0800

    Defaulting ControlSocket constructors to connect
    
    ControlSocket users almost always want the socket to be connected initially to
    defaulting it that way.

commit 98ad681bc05d91b596ceebb8f46e04f1f8598b84
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 9 10:01:27 2011 -0800

    Removing tag from version comparisons
    
    As Sebastian pointed out, the version-spec says "The status tag is purely
    informational". Removing it from comparison operations so versions like
    "1.2.3.4-foo" and "1.2.3.4-bar" will evaluate to being equal.

commit bd51ff4d69decfc866366a5381bd1ba0dc7ee866
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 9 09:47:17 2011 -0800

    Testing and fix for invalid auth cookies
    
    Adding a test and fix for when the authentication cookie is the right size but
    has the wrong contents.

commit 76be3997e31d5a56e335a111eb1f3c823dbc9787
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 9 09:06:26 2011 -0800

    Adding an authentication exception hierarchy
    
    For arm a substantial torctl pain point has been poor exception raising from
    authentication functions. If everything raises a ValueError then the caller
    needs to parse message responses or re-implement the code to get finer error
    granularity.
    
    Providing a hierarchy of authentication exceptions that can both make life
    simple for callers, letting them catch the exceptions they want then fall
    back to a catch-all for the AuthenticationFailure base.
    
    One mildly icky point is that this relies on the error message provided from
    tor to differentiate between when an authentication value is rejected verses
    the type. That said...
    - this issue is noted in the pydocs
    - I have integ testing to check for error message changes
    - callers are encouraged to use a general authenticate method which doesn't
      have this flaw since it uses the PROTOCOLINFO response (barring edge cases
      like a buggy PROTOCOLINFO response or the auth type changing in the split
      second between PROTOCOLINFO/AUTHENTICATE)
    - even if the message changes direct callers will still be getting the right
      category of exception
    
    Pydocs reference a general authenticate method but it's not included yet (first
    draft is written but needs some work and testing).

commit 3b89feb04f2d9389e387f7d2e1af5f69158d52e9
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Dec 2 07:47:49 2011 -0800

    Issuing IOError for cookie read failures
    
    Replacing OSError for read failures with IOError (more appropriate for this).

commit 8fd556572a170d06359458a37848d947313984d2
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Dec 1 10:10:43 2011 -0800

    Function and testing for cookie authentication
    
    Adding a function for password authentication. This includes checks for the
    file's existance and that the size is valid (for 4303).

commit 2c91e342a58dd389d5cb777cb1f4f59bfc198f71
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Nov 30 19:02:32 2011 -0800

    Skipping get_pid_by_name test when doomed
    
    The get_pid_by_name function fails when there's multiple instances of a process
    with that name. As a result the integ test for the function fails if there's
    extra tor instances running on the system.
    
    Using pgrep to check for other instances and skip those tests if they'd be
    doomed to failure.

commit 7f760f86414ee0bfbd050480e1753555c66e9a5b
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Nov 29 10:02:24 2011 -0800

    Function and testing for password authentication
    
    Adding a function for password authentication. This included escaping quotes
    but otherwise is trivial - most of the effort was refactoring the
    authentication integ tests.

commit 312423c9d0b68cea486c3269ddcaf95fcb3314cc
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 28 10:08:20 2011 -0800

    Function and testing for null authentication
    
    Function for authenticating to open connections and integration testing for it.
    The tests both check the happy case and responses we get in a variety of
    'authentication needed' scenarios.

commit c033f19b081149f49a65cb6ab79e836af5352a2e
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 28 09:39:14 2011 -0800

    Using ControlSocket for integ tests
    
    Replacing the send_message and recv_message calls via raw sockets with the
    ControlSocket class. Neither of these integ tests are for testing those methods
    and the higher level objects make the tests much more readable.

commit 206e4de3ffab8764aefbc02d969092cc47dfc37a
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 28 07:00:45 2011 -0800

    Exercising the get_socket arg with integ testing
    
    The get_socket (previously keep_alive) argument wasn't being exercised so
    adding that to the test for fetching a protocolinfo response via the control
    socket.

commit 04975ff9d149d8a881edd8ce26925d785150eda1
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 28 06:56:38 2011 -0800

    Splitting socket attribute from protocolinfo
    
    Bundling the requesting socket with the protocolinfo response was kinda clunky.
    I thought that it owuld make the api a little nicer, but in retrospect it's
    just weird so going to a more conventional tuple response instead.

commit 7bae33db31f26440810596b0b702d7f85bbfb1cd
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 28 06:18:03 2011 -0800

    Specialized subclasses for ControlSocket
    
    Adding a ControlSocket subclass for control ports and control sockets. This
    allows for a connect() method which we'll need when trying multiple connection
    types since the socket becomes detached after a failed authentication attempt.
    This is also gonna be a bit nicer for callers since it bundles the connection
    information (the port/path we're using) with the socket.

commit ec5c082e570123b011e0d9cb416d5ede1eba5c95
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 26 10:23:48 2011 -0800

    Moving scratch control connection into a module
    
    Later there will be a stem.control for the general controller code (ie, most of
    what TorCtl encompasses on its surface). Moving the first draft at that out of
    stem.connection, which it didn't really belong in anyway. Now none of the
    modules except control contain untested, scrap code.

commit 5b505ec579ab7edcaf55d1e35a2ccd3b9aee6672
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 26 10:10:21 2011 -0800

    Fixing relative cookie expansion test
    
    A couple protocolinfo tests filtered system calls so that pid lookups by
    process name would fall and we'd fall back on looking it up by the control port
    or socket file (to exercise alternative code paths). However, I'd forgotten
    that this would also filter out the get_cwd lookup calls, causing those tests
    to fail.
    
    The relative cookie expansion by socket file wasn't being exercised at all
    because I didn't have a integ test configuration where we had both a control
    socket and authentication cookie. I've added this test now and fixed this issue
    with the socket test too.

commit fd90d6c50011b8764b1c82c448fe39eab76c9c20
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 25 22:55:57 2011 -0800

    Using space or newline div when logging
    
    When logging a multi-line message using a newline divider with the "Sending:"
    or "Receiving:" prefix, otherwise using a space (minor bug had the space always
    included previously).

commit 8f8257d4ddcf6830b2413f53890a1fab4414ab45
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 25 22:42:35 2011 -0800

    Only respecting open default when undefined
    
    When no target is defined we should have a test.runner.TorConnection.OPEN
    default for integraiton tests. However, if we have an alternative connection
    target then this should be overwritten.

commit e313bb35c827fdb75a382d7c5e837692baeead73
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 25 22:36:57 2011 -0800

    Replacing protocolinfo lookups with ControlSocket
    
    Replacing raw socket use in the protocolinfo lookup functions with the
    ControlSocket class, and attaching it to the responses instead.

commit 62ecde1d4432eb96a3346289c8257561baa4faff
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 25 22:13:56 2011 -0800

    Refactoring version functions into module
    
    Moving the last of the types.py contents and a related function from process.py
    into a module specifically for handling tor versions and requirements (the
    later part will grow as the library matures).

commit 9a06ff17de582e834e2791c096bd9a51834a6d66
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 25 17:54:59 2011 -0800

    Moving control message handling into stem.socket
    
    Making a module for all low-level message handling with control sockets (ie,
    pretty much all of the library work done so far). This includes most of the
    code from the grab bag 'stem.types' module and the addition of a ControlSocket
    class. The socket wrapper should greatly simplify upcoming parts of the
    library.

commit e3d4311962b7e95b882a9a50b2402b17f2f2d3d9
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Nov 23 10:05:15 2011 -0800

    Showing shorter single-line controller messages
    
    When controller messages are on a single line logging them that way too, making
    the output a little more readable. I should probably file send/recv at a trace
    runlevel or with a separate logger...

commit 8dc796d142bebb6d370188f7e5be6bf65e402743
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Nov 23 09:55:16 2011 -0800

    Function for writting to control sockets
    
    Writing directly to the socket file isn't hard (it's just a write and flush).
    However, this is nicer since it wrap the control formatting, logging, and
    exception quirks. Functions still need unit tests and I might just wrap the
    socket object completely...

commit f98822f01e00197140ccea9dc88690d6ece004df
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Nov 23 09:23:42 2011 -0800

    Defaulting open connection integ target
    
    All connection targets were being defaulted to false, causing plain "run_tests
    --integ" runs to be no-ops. Hacking in the default values. I should probably
    use the more conventional dict/update pattern later.

commit 1c2337a71f9f19e262cd222679bf66eb046a9e73
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Nov 23 06:47:11 2011 -0800

    Nicer logging when receiving controller data
    
    Changing the logged controller messages a bit. This is a bit less faithful to
    the raw controller response but it's more readable.

commit b3afd554bad16bcde8b5ed5a70ed2a34c7e84762
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Nov 22 10:13:39 2011 -0800

    Better exercising cookie expansion in integ tests
    
    The get_protocolinfo_by_* functions weren't exercising cookie path expansion by
    port or socket file because lookups by process name would succeed and bypass
    this logic. Added a filter to the integ tests so we exercise both.
    
    When running with both the 'RELATIVE' and 'CONN_COOKIE' targets this reveals a
    bug with the stem.util.system.get_pid_by_port function that I'll address next.
    The test using the socket file passes.

commit 03d2e84894822825fc59507cdc93599f81cd8e36
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Nov 22 10:06:48 2011 -0800

    Noting at the end of testing if it passed/failed
    
    As testing output has gotten longer its become less clear at the end if all
    tests passed or not. Adding a note at the end saying if they all passed and, if
    there were failures, what they were.

commit 9fb52210494558a63d494d5ceab7a01282e8f88b
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Nov 22 09:53:30 2011 -0800

    More selective options for running integ tests
    
    Providing targets for all of the tor connection configurations so the user can
    opt for any combination of targets. Previously you needed to run the
    'CONNECTION' target which exercised them all and took around forty seconds to
    run (kinda a pita if you just want to test cookie auth).

commit d3b8f0aed7d715721724a7a94e96a188f1ee4e79
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 21 10:13:10 2011 -0800

    Integ target for running with a relative data dir
    
    Having a relative path for our data directory can cause headaches since tor
    then provides relative paths for the data it gives (for instance, for the
    authentication cookie location). Adding an integration testing target to have a
    relative data directory, to better exercise the path expansion code.

commit 56c2ea8c33e63456dc1316f3936439478e12cc03
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 21 09:21:21 2011 -0800

    Function to query PROTOCOLINFO via control socket
    
    Same as before, implementation and integ sanity check for making a PROTOCOLINFO
    query via a control socket. This has the common bits between it, the control
    port function, and a bit of the PROTOCOLINFO response parsing delegated to
    helper functions.

commit 7f89ff58d2ad6ecbde7f8225e7b2c02ec400be1b
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 21 08:36:20 2011 -0800

    Function to query PROTOCOLINFO via a control port
    
    Implementation and integ testing to query a PROTOCOLINFO response via a control
    port. Next is to do the same for control sockets.

commit e17bfdd0c44b004468bd3121fd5408e2a29074fc
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 21 07:51:00 2011 -0800

    Resetting system call mocking after test
    
    The protocolinfo test mocked system calls but didn't reset the mock when it was
    done. This didn't cause any errors but that was only luck (the system unit
    tests probably ran afterward and cleared the mock when it was done). Oops, this
    is gonna be an easy testing bug to introduce... :/

commit 708e082fb7bdddf2261a44fb290d20d768bc180c
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 21 07:42:14 2011 -0800

    Moving PROTOCOLINFO tests to be grouped by subject
    
    All the PROTOCOLINFO related tests might as well be together. Shuffling them
    around so all the tests can reside in a test/*/protocolinfo.py rather than have
    separate protocolinfo_response.py, protocolinfo_query.py, etc.

commit 07fd7f7d23f5312d4b4ee4ef64d63fbe97a911ba
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 21 04:29:10 2011 -0800

    Expanding coverage of PROTOCOLINFO integ test
    
    Providing the assertions for all connection configurations in the integ test.

commit 22667eaa6a02e6585ad502ea164fce15cd96cc1f
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Nov 21 03:59:27 2011 -0800

    Unit test for comparing Version class with others
    
    Had a pesky bug where comparison of Version instances with other classes would
    raise an exception and had forgotten to add a unit test - fixing that.

commit 3d14e1bd1076bdfd6fd109be0b76f71b12fbebaa
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 20 15:51:29 2011 -0800

    Running integ tests multiple connections methods
    
    Adding a 'CONNECTION' target that, if set, will run integration tests with
    multiple connection and authentication methods...
    - no control connection
    - control port with no auth
    - control port with an authentication cookie
    - control port with a password
    - control port with both an authtication cookie and password
    - control socket
    
    This means running through the integ tests six times which currently results in
    a runtime of arond fourty sectons, so this isn't the default.
    
    The primary purpose for doing this is to exercise the PROTOCOLINFO parsing and
    upcoming connecion methods with all of these tor configurations. The
    ProtocolInfoResponse integ test doesn't yet actually test all of these - fixing
    that is next.

commit e62502661059e8a3c93babf3e15df8a0a07d529d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 20 14:22:07 2011 -0800

    Adding a set method for config instances

commit da937730bb3aaa303e584eb3e23494703923c46b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 20 14:20:13 2011 -0800

    Using enum.__iter__ for values rather than keys
    
    Enumeration keys are of very limited use. Iteration over an enumeration should
    give the values instead so swapping values() and __iter__() to be keys() and
    __iter__().

commit a34316c59efbce56b82214bc257d8ddfa11e489c
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 20 00:55:15 2011 -0800

    Runner startup argument for tor connection type
    
    Integration tests are about to get an option for exercising multiple connection
    methods, so adding a runner initialization argument for starting with a torrc
    for all of the connection methods that we care about. This also includes a
    minor fix where we'd get a stacktrace when the torrc had an empty line.

commit 7b302e2eac72d9c3b4f09b705fd6c3adcb755f84
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 19 23:58:26 2011 -0800

    Integration test for general PROTOCOLINFO parsing
    
    Integ test for parsing a PROTOCOLINFO reply from our general integraion test
    instance. We'll need a separate target for testing multiple connection methods
    (password auth, cookie auth, and control socket).
    
    This also includes a fix for the Version class (equality checks with
    non-Version instaces would raise an exception - didn't expect __cmp__ to be
    used for that...).

commit 2c4686adcecbe73c32250b6201a3970284a1ca8a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 19 23:16:26 2011 -0800

    Unit test for relative cookie expansion
    
    Now that we have system call mocking we can have a unit test for expanding
    relative cookie paths. It kinda bugged me that testing wasn't complaining when
    we had a system api change. :)

commit 5847301f6011df37d517041c00ff23419e107316
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 19 22:25:21 2011 -0800

    Using new system api for protocolinfo queries
    
    The protocolinfo uses system utils for expansion of relative cookie paths.
    Making it use the new api (the breakup of the get_pid_by_* functions was
    largely for this class).

commit 314804dfce5069a13f07b79e5640e2e1b57e924f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 19 22:05:52 2011 -0800

    Dropping log inference from conf util
    
    The config's get method inference for logging runlevels no longer makes sense
    since the log util has been removed. Dropping this inference entirely rather
    than trying to make it work with logging - those config options have always
    been unused anyway.

commit ff5895f821370fe34609388fa44ebfb6cb1ff923
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 19 15:37:11 2011 -0800

    Color for testing output
    
    Applying color to the unittest output: green for success, blue for skips, red
    for failure. Bit easier on the eyes and makes issues easier to spot (at least
    on my terminal).

commit 70ab8ff1c2ccb858abe462dd1f9c3e81c1733659
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 19 14:18:13 2011 -0800

    System test bugs caught by pylint
    
    Couple copy-paste bugs with the system integ tests.

commit f6560b1210a20bdee7ccb68567b44b7f877a3e38
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 18 23:23:28 2011 -0800

    Rewrite and testing for system utilities
    
    Spent most of this week improving the implementation, api, documentation, and
    most importantly testing for the system functions. They now have almost
    complete code coverage by both unit and integ tests. Besides the obvious, this
    will help cross-platform compatability in the future since I'll have a sampling
    of input for platforms I don't have.
    
    Generated real output for all commands except sockstat (I only have access to
    linux and mac, not free/openbsd). I'll probably contact Fabian for help with
    this one.

commit dbe041ac1bbb775f71f34f0d705d4dd654d74858
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 13 16:21:09 2011 -0800

    Header documentation for the proc and system utils

commit 382bc5627b8d4f29645207628ca224f08f7b6883
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 13 16:13:33 2011 -0800

    Basic integ test for stem.util.conf
    
    Adding an integ test for the example given by the conf utility. There's a whole
    lot more that could be tested in that class (especially parsing and type
    inferences) but this doesn't seem too worth while so just adding this basic
    test for now. I might expand it later.

commit b34f8990b6604996d7d60d139267ae43bac1de02
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 13 15:09:51 2011 -0800

    Shuffling integ tests to match unit tests
    
    Unit tests are nicely categorized under their respective modules, so reordering
    integration tests to match.

commit d30e229c325bc243697d1d015a7bda03082c9756
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 13 15:02:47 2011 -0800

    Unit tests for enum

commit db7635abc7878d873a989c35e7aea413ecde2f52
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 13 14:38:56 2011 -0800

    Revising enum docs and methods
    
    Minor changes including...
    - standard header documentation
    - replacing the keys() method with making enums iterable (functionally the
      same, but a little nicer for callers)
    - dropping the alternative LEnum - I've never used it

commit f9aeefb2805e2ef9cd63f236c7a27308718575e9
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 13 14:01:55 2011 -0800

    Rearranging conf util to improve usability
    
    Adding the standard configuration header (... sooo helpful) and moving the rest
    of the util's contents around to be more intuitive.

commit fc562144fd8a333ad517575976328bc2dc3cec5d
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 13 13:43:59 2011 -0800

    Including stem prefix for paths in stem comments

commit 2b797d0ba9310a015bef5bc2605020ebe8ffafdd
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 13 13:15:49 2011 -0800

    Unit tests for PROTOCOLINFO responses

commit 5756f9940ff0bbd55330a700d648af347dec713a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 13 02:19:26 2011 -0800

    Parsing and class for PROTOCOLINFO responses
    
    Finally have enough plumbing in place to write the parsing for the PROTOCOLINFO
    queries. I'm pretty happy with how it turned out - next is testing for the
    class, then moving on to functions for issuing the PROTOCOLINFO queries.

commit 5d200b1cdbebcd60cadf6455a9b1baf1d9431fb3
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 12 16:19:34 2011 -0800

    Avoid adding duplicate null logging handlers
    
    Checking if logging already has a handler before adding a NullHandler, to
    avoid having _every_ import trigger an extra no-op handler.

commit efe28987ffd27e69a2c6c566bdad2efc5f16e93e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 12 15:55:09 2011 -0800

    Expanding contorller exception types
    
    Exception types pretty much mirror what TorCtl has (protocol error, socket
    error, and controller closed) with a base type users can catch instead. One
    difference though is that if stem functions raise a socket.error (without
    documenting that they do) then that's a bug - those errors should cuase a
    stem.types.SocketError instead.

commit d1c3c8db724eb17c754c26f66016315d0524a74a
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 12 15:46:13 2011 -0800

    Standardizing on suppress_exc args for sys util
    
    The system functions are mostly best-effort, platform specific attempts to
    retrieve data and usually don't have meaningful information to provide back
    to the caller via exceptions. I'm defaulting to have them return None, with
    an optional arg for having them raise IOError exceptions on failure instead.
    
    I've gone back and forth on the fencepost here a few times, and I'm still not
    positive if this is the right choice. Might change this later if it doesn't
    work out.

commit fab872725d791e8a9ca20f28f6eb5c5358aff87b
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Nov 10 09:30:06 2011 -0800

    Function for querying a process' pwd
    
    Implementing and testing a stem.util.system.get_pwd(pid) function. This will be
    needed for correcting the relative cookie paths mentioned in...
    https://trac.torproject.org/projects/tor/ticket/1101

commit 7c7bd6adeca59e593635d6eb21c6a9a0e7cf585e
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Nov 9 10:23:07 2011 -0800

    Fully qualifying util paths
    
    Changing the conventional usage of util imports from "from stem.util import X"
    to simple imports (not pulling them into our namespace). There's some
    exceptions to fully qualified util usage where it hurts readability (often the
    case with stem.util.term), but explite paths will be the more common case.

commit 713eec7b70c06ab3361f22fb49f6eb6a8b9e82d3
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Nov 9 10:13:55 2011 -0800

    Replacing custom logging util with python builtin
    
    Finally abandoning my custom logging implementation for the python builtin
    logging module. It's far more customizable, standardized, and what users would
    expect from a library like this. The only disadvantage is that it doesn't
    buffer past events so we lose anything prior to adding a handler.

commit 78de4e4cf1f442b50ec1b30733127f9b83a97fce
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Nov 8 09:04:16 2011 -0800

    Util for reading proc contents
    
    Rewrite of arm's utility for parsing proc contents, changes including better
    error handling and refacotring for stem's coding conventions and commenting.
    
    While writing the PROTOCOLINFO response handler I needed a system util for
    querying tor's pwd, which works best with proc information, so pulling it all
    in. For now most of this isn't being exercised, but will be later.

commit eb5b522430bc32acb0f8f18c1e92e8ae7782df62
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Nov 6 11:24:21 2011 -0800

    Adding optional key arg to is_next_mapping checks
    
    Option to simplify testing for the next key being parsed.

commit 40b7ab1b36dd4bd0ce455714ea8a9dfd4d3bb76b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 5 17:18:10 2011 -0700

    Unit testing for stem.types.ControlLine
    
    General unit tests to exercise the ControlLine class with PROTOCOLINFO output.
    This also has a minor fix so we throw an IndexError rather than ValueError when
    pop_mapping() is called while empty.

commit c0525d58a7e98e2c4b3fa7e2130e69e85aee5247
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 5 15:49:48 2011 -0700

    Unit testing stem.types.ControlLine.pop() examples
    
    Adding unit tests and correcting the pydocs for the pop method examples.

commit 78d44b0995fb607715d189051a6aec13f512ad35
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 5 14:15:33 2011 -0700

    Moving testrc.sample into testing module
    
    Tired of having 'test' autocompletion trip up on the sample rc file, so moving
    it into the test directory. It's probably not of much interest to general
    library users anyway.

commit c4c998f6201dae0123c068c3e081a26d6af8a99b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 5 14:10:07 2011 -0700

    Better organizing unit tests for stem.types

commit aba675c3f565aee7f1855215e90b1d7af26a1ce8
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Nov 5 13:56:13 2011 -0700

    Header documentation for ControlLine class

commit 12064b7b9fc4799b50ff83cffbab55829c0e1046
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Nov 4 10:16:28 2011 -0700

    Replacing get_entry with ControlLine class
    
    Making a string subclass to help with the parsing of controller output. Most
    entries are space separated lists of elements, which this class has functions
    to easily parse. For controller messages that don't follow this pattern we can
    still treat it as a normal string.
    
    Next is to add header documentation and tests.

commit 45d3c819b21dd29aec70178f66777a365200a617
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Nov 3 07:07:11 2011 -0700

    Unit tests for types.get_entry examples
    
    Unit tests for the examples in the pydocs of the function and fixes for the
    issues they uncovered.

commit a090373e5384026295b31daee8ef622a589a913d
Author: Damian Johnson <atagar@torproject.org>
Date:   Tue Nov 1 18:49:55 2011 -0700

    Utility function for parsing control messages
    
    Trying out a utility function to pop the first item off a control response,
    with the caller providing attributes it should have.

commit 263dfd53cec592c10d9dbe40211ea56915ac7552
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 30 00:51:47 2011 -0700

    Config argument for the test runner
    
    Integration tests can have custom behavior via a testrc, which is simple for
    now, and will grow as tests become more complex. Previously I was loading a
    static settings.cfg but that was stupid. The user's config file should neither
    be hardcoded nor under version control. This change also includes the config
    loading in the status output to stdout.

commit 63a9a5fa524e352c94bab7beabd5052e4972c6c4
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 27 20:31:43 2011 -0700

    Merging version constructor and parser
    
    The version constructor wasn't really useful so merging it with the get_version
    function to make its usage more intuitive.

commit 52be689448931b1eca7de3add345f9aa2d79319f
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 27 19:43:45 2011 -0700

    Putting stem under the LGPL v3

commit f07c6cc77f4ad77b2390d03f9aad5f32aecd245e
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 27 09:39:41 2011 -0700

    Expanding header documentation
    
    Adding a summary of a files functions/classes/methods at the top of the file.
    This makes using a library much nicer.

commit 11b2a71d3a6ae24365ebd1e9a30c323a6a3ce275
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 27 09:17:26 2011 -0700

    Caching for stem.process.get_version()
    
    Caching results of the stem.process.get_version() function to avoid unnecessary
    lookups.

commit 7766cb79fff29020b72b5fe45fc43d184abada23
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 27 07:16:06 2011 -0700

    Requirement check for the config-text test
    
    An integration test requires 'GETINFO config-text' requires tor version 0.2.2.7
    so adding a check for it.

commit fbfa73a099d9645f18d9846420cbf0145065b11d
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Oct 26 20:13:34 2011 -0700

    Rewrite of integration test runner
    
    Several imporvements for the integration tests, most notably including...
    - Test configurability via a 'test/settings.cfg' file
    
    - Thread safety for runner usage
    
    - Vastly better startup time for how integration tests run by default...
      - Reusing data directory so we don't need to request as much from authorities
        when starting (faster startup and less burden on them). Users can opt for a
        fresh temporary directory instead by setting 'test.integ.test_directory' to
        a blank value.
    
      - Starting tests when bootstraping reaches 5%. This is enough for tests that
        don't require network activity to run, and we can explicitly run those
        tests by setting the 'test.integ.run.online' option. This change also means
        that we can now run integration tests while offline.

commit 51d2ded44fe48f54c63cf4e5937a2ff91fa99e95
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Oct 21 08:10:57 2011 -0700

    Adding function to query tor's version
    
    Command to issue a 'tor --version' query with a stem.types.Version response.

commit 52768d36f85083923c4f09a1b0956ee146ae0b16
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Oct 21 07:33:40 2011 -0700

    Quietly stopping integ launch on ctrl+c
    
    Keyboard interrupts while launching tor would result in a stacktrace. Quietly
    exiting instead since this isn't really an issue.

commit 8293c518d02f879110d45e0cff074e7d31d2fbfb
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Oct 21 07:28:27 2011 -0700

    Making the tor launching timeout an arg
    
    Moving the timeout for launching a tor process from being a constant to being
    an argument of the function.

commit 0a083c8b73a663d38f473e0180b0009d28e1b5d9
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 20 10:09:36 2011 -0700

    Accidently suppressing integ asserts
    
    A previous commit to suppress errors from the close() method in python 2.7 also
    suppressed the following asserts. These asserts still work with 2.7 so removing
    them from the try block.

commit 3a913b6b92796f2195c28e55ead532a0fef906e6
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 20 09:58:37 2011 -0700

    Moving tor launcher into stem function
    
    In writing the stem integ tests I needed a function for starting tor then
    blocking for its bootstraping to complete. Yesterday Jake mentioned that he
    could use a launch_tor() function in TorCtl so I'm generalizing this code and
    moving it into the stem lib.

commit 26374bafd5957d3039cf10596de0f7a021f1afb9
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 20 09:49:45 2011 -0700

    Message integ tests compatability issues
    
    The message integration tests had compatability issues with...
    - Newer python versions (2.6 -> 2.7) due to an unexpected failure when calling
      close() on the socket's file object. We're operating with a closed socket at
      that time so all bets are off about if this should/shouldn't raise so the
      difference in behavior just warranted a testing workaround.
    
    - Older tor versions don't support 'GETINFO config-text'. I've left this as a
      TODO note for now and will add a version check for that test later.

commit 1b44b967e75b08d6f702ffa1507d8ad8c4980bda
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 17 09:45:43 2011 -0700

    Integration tests / fixes for types.ControlMessage
    
    Adding integration tests for basic control port communication, exercising...
    - connection failure
    - bad commands
    - bad getinfo queries
    - general getinfo queries
    - setevent/basic event parsing
    
    This also includes fixes for a variety of issues found while testing.

commit 24d4881025c891c829a63de9b9b1b8849c8c4faf
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 16 19:30:57 2011 -0700

    Moving integ runner into testing base
    
    Moving the integ runner module from 'test/integ/runner.py' to 'test/runner.py'.
    The 'test/unit/*' and 'test/integ/*' are for test cases, and will later be
    kinda crowded. The runner is special since it doesn't test, but rather provides
    the runtime context for the integ tests so probably best to keep this separate
    from the rest.

commit aa01c0798960a520ac1da81e5a1a447c29dbb962
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 16 19:23:57 2011 -0700

    Adding get_pid and get_bsd_jail_id util functoins
    
    Stealing arm's getTorPid and getBsdJailId from the torTools util, generalizing
    both functions to be for arbitrary processes rather than just tor. This also
    adds unit tests for get_pid and a simple exercise of get_bsd_jail_id (I can't
    really test the later since I'm not on BSD).

commit 3731bba464e22dc5501b7b9ae62c29ebc34ad7ba
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 16 17:27:57 2011 -0700

    Adding BSD compatability to system.is_running
    
    Copying over an arm fix I made earlier this week from
    util.torTools.isTorRunning so that it would work on OSX/FreeBSD/OpenBSD.

commit 1a30fbd6526b243468a6ca92f0223eb84ff9b21b
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 16 17:19:01 2011 -0700

    Integration tests for stem.util.system functions
    
    Very, very basic tests that the tor process is running and available.

commit 0c03d4eb2a04f71942c53caf6314c25ce66be18e
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 16 15:19:03 2011 -0700

    Moving integration test init/shutdown into class
    
    Making a Runner class which will be available to integration tests. This both
    better modularizes the code and will allow for us to give runtime context to
    the tests.

commit bdbf352ad7f56aec17ce0156eaf53c9a6ac1fe68
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 16 01:17:38 2011 -0700

    Running tor instances for integ tests
    
    First draft for the startup/shutdown of integration tests. This...
    - makes a test directory
    - generates a torrc
    - starts a tor instance, waiting until its bootstrap completes (timing out if
      it gets stuck)
    - runs tests (not done yet - those are next)
    - shuts down the tor instance
    
    This'll certainly go through some refactoring to better modularize, but it's
    functional as-is.

commit a2fde9374e188725b9976c42f8a679814d87fac1
Author: Damian Johnson <atagar@torproject.org>
Date:   Fri Oct 14 20:33:12 2011 -0700

    Adding util for handling configuration files
    
    Integration tests and other things will be stored via configuration files
    (hardcoding data in source is a horrible thing and you have no excuse unless
    you're writing in LISP). This is being stolen from the arm codebase, with the
    code refactored for this project's conventions, some fixes to better generalize
    the util, and vastly improved documentation.
    
    Tom has been using the config util a bit for his torperf rewrite and having
    difficulties so hopefully this will make its usage easier.

commit c74ac0f7f0c75d429a7d32f7d516b3518fc90ff0
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Oct 12 09:37:49 2011 -0700

    Tesing and fix for disconnected socket reads
    
    Adding a unit test and fix for when the input file is derived from a socket
    that's never been connected. I'm suspicious that this won't catch disconnects
    that accur a little while into the socket's use, but I'll need to implement
    integration tests for that. Guess that's next...

commit 05b106bb575f669487fc376a29b7710d2addb2b0
Author: Damian Johnson <atagar@torproject.org>
Date:   Wed Oct 12 09:17:59 2011 -0700

    Protocol error tests and fix
    
    Expanding the ControlMessage unit tests to cover causes of protocol errors, and
    fixing an issue that this revealed where it was possable for a malformed line
    prefix to go undetected (if, say, a dropped character caused a valid line
    divider to fall into that place).

commit 5b1a7e8f2d375db5ad4d3c212fbae93ed4724277
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 10 21:24:58 2011 -0700

    Basic unit tests for ControlMessage
    
    Just some sanity checks that ControlMessage parses common GETINFO responses.
    I'll expand on these including some error cases next.

commit 92f78d459497dd3c0d688c040e06d6cc6668c43c
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 10 19:54:30 2011 -0700

    Moving ControlMessage and parsing into types
    
    The ControlMessage is a standalone class so moving it into types. This is both
    appropriate for the types module and will make testing easier.

commit 5d7688e7233a5a1f8a4bb4a1c939639031a71aa0
Author: Damian Johnson <atagar@torproject.org>
Date:   Mon Oct 10 10:05:14 2011 -0700

    First draft controller connection
    
    This is a functional rewrite of the sendAndRecv TorCtl functionality. It
    follows a similar pattern, having a couple threads to continually pull the
    socket and provide event notifications. This still needs testing, better
    exception handling, and some more thought about the ControlMessage api.

commit 1aff6b875213d70a05d0cb5d86a5ff9914ce6e41
Author: Damian Johnson <atagar@torproject.org>
Date:   Sun Oct 9 03:24:44 2011 -0700

    General system utilities
    
    Adding some simple utility functions that I'll need later. These are rewrites
    of their arm counterparts, which I wrote before discovering the subprocess
    module.

commit 00739ee4ebf3c74d1451225c026a90ab52eed711
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 8 23:23:08 2011 -0700

    Adding logging utility
    
    Borrowing the logger from arm, with some refactoring to remove unneeded
    functions and make it conform with this project's coding conventions.

commit 88921da126612d90a2dbf002f37ba236da047cd4
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 8 16:15:48 2011 -0700

    Adding gitignore for pyc and vim swap files

commit 3dd9ea0d9222ddf22e63c7f5dcf8e7032e6fb494
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 8 16:04:08 2011 -0700

    Running tests based on input arguments
    
    Making the test runner accept arguments for the type of tests to be ran. The
    integration tests especially will take a while when they're implemented so
    letting the user specify the use cases for those.
    
    This included copying and refactoring some basic utilities from arm for
    enumerations and terminal text attributes.

commit 39e10eed903c1920a50a175803726035b8448712
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 8 14:03:08 2011 -0700

    Using Pascal Case styling (unscores) for variables
    
    According to PEP8 [1] both functions and variable names should use the
    underscore naming convention (ie, 'my_var') rather than camel case ('myVar').
    This is a little weird for me and python standard libraries use both, but I see
    a readability advantage to this for functions and my previous approach of using
    camel case just for variables is kinda weird. Hence switching to the 'right'
    convention while the codebase is still tiny.
    
    As an added bonus this conforms with torctl...
    
    [1] http://www.python.org/dev/peps/pep-0008/

commit d38b8830804e78cd0921333c39c5e9a2ee8ad42f
Author: Damian Johnson <atagar@torproject.org>
Date:   Sat Oct 8 13:10:38 2011 -0700

    Subdirectories for integ and unit tests
    
    Unit and integration tests will be separate subdirectories of test.

commit e1c31c060c0246a62e911643a383372889214300
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 6 09:57:24 2011 -0700

    Removing checked in binaries
    
    Forgot to establish a gitignore before the prior commit. An extra copy of
    run_tests.py also snuck in (haven't a clue how).

commit 69083da95c5ab536adcd23f19a424ed9395015bb
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 6 09:52:04 2011 -0700

    Class and function for Tor Versions
    
    Starting with a simple (but non-trivial) class that will be needed for handling
    PROTOCOLINFO responses. This is partly to establish conventions for
    documentation and unit tests.

commit 211c52e6f24440909bf243e0b8c11e3158c7fcee
Author: Damian Johnson <atagar@torproject.org>
Date:   Thu Oct 6 09:33:34 2011 -0700

    Initial commit
    
    Simply filling in an empty file to initialize a root for the repo.