commit 394876f258bf99132ba23c2e2bd07ee5ac6404cf
Author: Jason Kölker <jason@koelker.net>
Date:   Mon Feb 18 17:26:26 2013 -0600

    Make sure to install the oslo package as well.
    
    Each package under a namespace that should install that namespace
    package so the __init__.py for the namespace gets copied.
    
    Fixes Bug 1129587
    
    Change-Id: I064b05479c8a41d8f793886d2a5895dce166a85a

commit 36bbf72d2f32a1ae10c484259cf7f29fbbfcd78e
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Sun Feb 17 09:25:32 2013 +0000

    Add LICENSE file
    
    Change-Id: Icfbb242b0b5a1d494ed673d48cea07fdd25fbbd7

commit af4ab489df16f120bef53ddb5ca02a6e7e042676
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Sat Feb 16 14:31:38 2013 +0000

    Fix setup.py to install the correct package
    
    Change-Id: I87e3977ecd9078df0c8aa14ed18dc288fe49c596

commit 16c4ecbefc94941821f3ae7ca253fad7bc377c58
Author: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
Date:   Wed Feb 13 14:01:46 2013 +0900

    sort options to make --help output prettier
    
    explicitly sort options when adding them to argparse.
    it's a bit silly to print them in a dict iteration order.
    
    Change-Id: Id508331d7ee3b24e76be7fa958d27d29905bd3d2
    Signed-off-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>

commit b1730fcbc4f0852d9dd32fe6440e147fcd930abe
Author: Zhongyue Luo <zhongyue.nah@intel.com>
Date:   Wed Jan 30 22:27:51 2013 +0800

    Implements import_group
    
    Created import_group method in cfg.py
    Added testcases
    
    Fixes bug #1107843
    
    Change-Id: I427d4069dacbb6e586687370adbb08b5d50f7b63

commit a118969fb71c95bd816db05b3b1b3c6fc0902bb9
Author: Zhongyue Luo <zhongyue.nah@intel.com>
Date:   Mon Jan 28 14:35:49 2013 +0800

    Fixes "is not", "not in" syntax usage.
    
    Replaced "not ... is" to "is not"
    Replaced "not ... in" to "not in"
    Removed a redundant parenthesis
    
    Change-Id: I9564ab1207ccdcb32d7c2bb9e8f29658b2232ff9

commit 9805067bcf156dac354644b9b97ff8ceb4b5b3c6
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Sat Feb 9 15:36:08 2013 -0500

    Add sphinx documentation
    
    Add basic sphinx config and copy the theming from oslo-incubator which
    was originally copied from keystone.
    
    Change-Id: Ibb3b679ce6e160c157ff63f0943807bd82aa1a67

commit f24575c137abd204d5145ff59ae6e2fafc635a73
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Jan 28 09:54:01 2013 +0000

    Fix version to 2013.1

commit 7da69211e912a8000c6498e015522edd223edfe7
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Jan 28 09:28:30 2013 +0000

    Start using fixtures and testtools
    
    Sync these changes from oslo-incubator:
    
      60f70b0 Replaced direct usage of stubout with BaseTestCase.
      827547a Use testtools as test base class.
    
    Note: I've copied MoxStubout for now, but eventually I guess we'll
    have an oslo-testing library we can depend on.

commit c490e3515e8fee8c5743b944f60eeafb1b67864b
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Fri Jan 25 17:29:00 2013 +0000

    Sync latest setup code from oslo-incubator
    
    Changes include:
    
     6b3c544 setup: count revs for revno if there are no tags
     9c8685a Use revno and git sha for pre-release versioning.
     5f5ef7d Add env var version override for packagers.
     602aa9c trivial pep whitespace fix

commit 45f1f32751f2fb9a841f46255a8108b46ad5cabb
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Jan 21 14:13:59 2013 +0000

    Move logging config options into the log module
    
    We learned a lesson in Nova - it's best to declare and use config
    options within a single module if possible. Globally declared and
    use config options grow like weeds and it becomes harder to find out
    if, where and how individual options are used.
    
    Strangely, in cfg itself, we randomly declare a bunch of logging options
    which are only used within the openstack.common.log module - let's move
    the options there and remove the CommonConfigOpts class before they
    become part of the API we commit to when oslo-config is released.
    
    A minor detail in the patch - the logfile and logdir options are already
    deprecated in favour of log_file and log_dir, but we never got around to
    removing all other traces of the deprecated options.
    
    Change-Id: I3913ea54465658d93dc56e014dfe5d911b0541d6

commit 2ca3c749592c6b445b1ff27ccc40e0f644cf98a1
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Dec 17 23:02:47 2012 +0000

    Add setuptools magic

commit 8c6a4c7523910e0745ae06194fa9484e45feb190
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Jan 21 09:12:55 2013 +0000

    Add oslo-config project infrastructure

commit 27cc655c8046d5f3d59c72934ed11067e0e1e8f1
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Jan 21 09:11:41 2013 +0000

    Move files to new locations for oslo-config

commit 2b210f7ebb7d29adfb4c12179c55d0167d8eff77
Author: Zhongyue Luo <zhongyue.nah@intel.com>
Date:   Fri Jan 18 14:10:05 2013 +0800

    Fixes import order errors
    
    Change-Id: I3e35230dd2d96ab9f5a8c11b9ec1cd8d2d00e347

commit 54d834e485927d704a4c68c69135229ad6ff9826
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Jan 14 08:38:17 2013 +0000

    Make tox run doctests
    
    Use 'nosetests --with-doctests' to run any doctests found. We currently
    only use doctests in a handful of places, but we may as well run them
    to ensure they work.
    
    Make the cfg doctests avoid using the global CONF since we would need
    to reset its state between each doctest.
    
    Fix the cliutils doctests to actually pass.
    
    Use 'nosetests --exclude-dir=tests/testmods' to avoid loading the
    modules from this dir while discovering doctests. The cfg unit tests
    rely on these modules not having been previously loaded.
    
    Change-Id: I19ad70767fa5c8352b994dc963b5d3a7c9d9eb95

commit 947bb9777bdfdcaa44ed4c22fab83a3d07b0a91f
Author: Davanum Srinivas <dims@linux.vnet.ibm.com>
Date:   Thu Dec 13 22:42:33 2012 -0500

    Verbose should not enable debug level logging
    
    Fixes LP #989269
    
    Currently setting --verbose in will still allow DEBUG level
    message to be logged to python logger object. we need to check
    for --debug first (set DEBUG level), then --verbose (set INFO
    level) and if neither is set then set default to WARNING
    
    DocImpact
    
    Change-Id: Ic9e3cb5979b2d7283552ad3a461870373f45a239

commit 049b3f8218eb8181cb4f3e3bf52af1d1b64c092a
Author: Monty Taylor <mordred@inaugust.com>
Date:   Fri Jan 11 12:01:42 2013 +0100

    Fix pep8 E125 errors.
    
    Caesar's wife must be above reproach.
    
    Change-Id: Iac85a57e71d403360f1567c07c8699057f0772fb

commit d3deabba42f3a5e7325392def03d985967d76827
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Tue Jan 8 21:48:11 2013 +0000

    Revert "Support lookup of value using "group.key""
    
    This reverts commit 525ac47.
    
    There are already two ways to reference an option in a group:
    
      CONF.group.key
      CONF[group].key
    
    Adding a third variant doesn't seem ideal. Also, for the specific case
    of LazyPluggable in Nova, we can easily just pass an optional config
    group name to the constructor.
    
    Change-Id: I1a29a18d90e8af3ce2563bd1b3eeb64422140016

commit d666430d5004ea56114a2b75ed4b9907e6a9e393
Author: Davanum Srinivas <dims@linux.vnet.ibm.com>
Date:   Wed Dec 26 22:50:35 2012 -0500

    Support lookup of value using "group.key"
    
    Let us check if the opt_name has a '.', if it does then split it
    into a group/key and try lookup using that combination. Since
    LazyPluggable uses "CONF[self.__pivot]" if we just add this
    capability to cfg, we get "LazyPluggable doesn't support
    option groups" for free.
    
    Fixes LP #1093043
    
    Change-Id: I9cedcf22014038e9fe4ed5e66ca5427aa99b5091

commit 8fa6c6232348754cf9f0586114732992df112dae
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Fri Dec 7 06:44:13 2012 +0000

    Add deprecated --logdir common opt
    
    --logfile and --logdir are aliases Nova has for the --log-file and
    --log-dir. If we're to support --logfile as a deprecated common
    option, we should do the same for --logdir.
    
    Change-Id: I16485a93070d9ad7789a287d5b035c6f270ffead

commit eb4e0f00ac927cee3767f5c25c7a5ec004af0917
Author: Dan Prince <dprince@redhat.com>
Date:   Thu Dec 6 10:40:36 2012 -0500

    Add deprecated --logfile common opt.
    
    This adds a deprecated common options for --logfile which
    is an alias for --log_file. This resolves some backwards compatability
    issues with the most recent oslo common code where --logfile
    was no longer a valid opt.
    
    Change-Id: I17b1277da94a2d81ae439d650a6d7321420dfe14

commit b749f1d3cfeb6937240c05507db8872c7041fa0b
Author: Davanum Srinivas <davanum@gmail.com>
Date:   Wed Dec 5 16:11:48 2012 -0500

    Allow nova and others to override some logging defaults
    
    - In log.py, indicate that logging module allows tweaking
    of just logging_context_format_string option
    - In cfg.py, add a method that can alter the default given
    the options and new default
    - add testcases for log.set_defaults and cfg.set_defaults
    
    Fixes LP #1083218
    
    Change-Id: Iefdbce8505eb7a07f2b59d4ed7564b0146f1b0cd

commit 1aba080308ec7e88a13533d0cfed30e640bb2530
Author: Michael Basnight <mbasnight@gmail.com>
Date:   Wed Dec 5 16:00:11 2012 -0600

    Fixing the trim for ListOp when reading from config file
    
    Fixes Bug 1087018
    
    Change-Id: I1c2d34166ae85add86daab6a7483b63297d00f66

commit 82de68129b2bcc611bb584d262291ec32c249a4c
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Nov 26 06:47:52 2012 +0000

    Fix set_default() with boolean CLI options
    
    Porting to argparse broke set_default() with boolean CLI options. The
    new test case shows this borkage.
    
    The issue is that, by default, argparse differs subtly from optparse in
    its handling of defaults for boolean options. Compare:
    
      >>> p = optparse.OptionParser()
      >>> p.add_option('--foo', action='store_true')
      >>> p.add_option('--bar', action='store_true', default=False)
      >>> p.parse_args([])
      (<Values at 0x7f28aba066c8: {'foo': None, 'bar': False}>, [])
    
    to:
    
      >>> p = argparse.ArgumentParser()
      >>> p.add_argument('--foo', action='store_true')
      >>> p.add_argument('--bar', action='store_true', default=False)
      >>> p.add_argument('--blaa', action='store_true', default=None)
      >>> p.parse_args([])
      Namespace(bar=False, blaa=None, foo=False)
    
    i.e. unless you specify a default for a boolean option, optparse
    defaults to None whereas argparse defaults to False. To get the
    same optparse behaviour with argparse, you need default=None.
    
    Change-Id: Ifc92a834c4ba59e939d80ac5de24d7051232f5b5

commit fd4cad82e84f8c67644bb9fcbd9bc5fe735ff604
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Fri Nov 23 15:50:04 2012 +0000

    Improve cfg's argparse sub-parsers support
    
    In order for sub-parsers to be useful, you need some way of knowing
    which sub-parser was chosen during argument parsing. It's pretty obvious
    from the current sub-parsers test case that we don't have a convenient
    interface for this.
    
    One way of doing it is to use the 'dest' argument when adding
    sub-parsers:
    
      >>> subparsers = parser.add_subparsers(dest='cmd')
      >>> subparsers.add_parser('a')
      >>> subparsers.add_parser('b')
      >>> parser.parse_args(['a'])
      Namespace(cmd='a')
    
    The most sensible way to map this into cfg concepts is to register
    sub-parsers as an Opt. This way, we can make name and argument values
    of the sub-parser as an attribute on the ConfigOpts object:
    
      >>> def add_parsers(subparsers):
      ...     a = subparsers.add_parser('a')
      ...     a.add_argument('id')
      ...     b = subparsers.add_parser('b')
      ...
      >>> CONF.register_cli_opt(SubCommandOpt('cmd', handler=add_parsers))
      True
      >>> CONF(['a', '10'])
      >>> CONF.cmd.name, CONF.cmd.id
      ('a', '10')
    
    The handler method is a bit awkward, but each time cfg is to parse
    command line args it takes all the registered opts and creates a new
    argparse parser. So we need to be able to re-add the sub-parsers each
    time.
    
    Change-Id: I01bfd01bf8853cf57a9248b1663eb3da142366a4

commit 18065f5050b00e19c36781902c218c3703b2e50d
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Fri Nov 23 08:09:12 2012 +0000

    Fix regression with cfg CLI arguments
    
    Fixes bug #1082279
    
    Only options registered using register_cli_opt() should be available via
    the CLI, but since e42276a all options are added to the CLI.
    
    Also modify one of the existing unit tests to catch this problem.
    
    Change-Id: I742a4ae4e0fc17cd9ae5e4424c2edd38e2bc50a2

commit 3d4257fbd3dbfd7b64de22c2aae08a9e4cce9921
Author: Davanum Srinivas <dims@linux.vnet.ibm.com>
Date:   Sat Nov 24 08:54:20 2012 -0500

    Fix ListOpt to trim whitespace
    
    - throw in an extra strip() in the list parsing code
    - add a test case to verify that it works!
    
    Fixes LP #1079299
    
    Change-Id: I4f0864c72ecd2569d0461c301acda395c87a93e0

commit f1133ad6da81ed1f4f99c163ba3bd49b9ccd3e16
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Fri Nov 23 20:25:12 2012 +0000

    Add another duplicate opt test case
    
    This gets the code coverage of the tests back up to 100%.
    
    Change-Id: I737c1cfa52d10b3813237a9cb88b15211e0872c1

commit 877e625ca9fe1d93a924c8df6656fb7c56ccbb72
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Fri Nov 23 15:12:12 2012 +0000

    Hide the GroupAttr conf and group attributes
    
    There's no reason why an option group shouldn't have options called
    'group' or 'conf'.
    
    Add a test case which would have failed because the 'conf' attribute
    would have been a ConfigOpts instance and fix it by making those
    attributes private.
    
    Change-Id: Ic3e41a546c0d1b7b6aae04e1dbac2933ac661f57

commit 42376535a91460ea538b7c8c60efee30fcd28d21
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Fri Nov 23 11:45:04 2012 +0000

    Fix broken --help with CommonConfigOpts
    
    Since we switched to argparse, the way help strings are interpolated
    have changed and broken --help with the options registered by
    CommonConfigOpts.
    
    Fix and add a new test case which would catch the issue.
    
    Change-Id: I10e42efe4721e22ff41d0efbf390c805ccb9a6a0

commit 39eb58f6f010d08d00d06c596594b320de1fd8a9
Author: Joe Heck <heckj@mac.com>
Date:   Sun Nov 11 21:00:42 2012 +0000

    updating sphinx documentation
    
    * adding openstack theming (copied from keystone project theme)
    * updating .gitignore to ignore generated API docs
    * updated formatting in index.rst page
    * updaed openstack/common/processutils.py to match hacking docstring
    * updated docstrings to resolve sphinx warnings
    
    Change-Id: Ie89afe20eeab1efd2daf440fc65ccdf90f723c51

commit 6e3307be97ccba18793282582a6c1dd4509285fa
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Nov 12 16:26:08 2012 -0500

    Don't reference argparse._StoreAction
    
    This is a private implementation detail of argparse, so we don't want
    to rely on it. Just sub-class Action instead.
    
    Change-Id: Icfcc782cc334d1bc1d4940bec23af48ead692a9d

commit 33b6139fb2660fa230865d974f04156e3fe88760
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Nov 12 16:26:02 2012 -0500

    Fix minor coding style issue
    
    Use the same style of exception handling used everywhere else.
    
    Change-Id: I5436de1996f69ea6210f48c11ef231eb950ad21d

commit 45613f3aa7259c1192ea58919d144d1ce0c97f38
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Nov 12 16:26:00 2012 -0500

    Remove ConfigCliParser class
    
    This sub-class of ArgumentParser isn't really justified anymore.
    
    Change-Id: I705224b6e18e4609a8e2deba283767233b0bd578

commit cbe7cfa46ab56ea98f08489e29cbb3949bcf37d5
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Nov 12 16:25:59 2012 -0500

    Add support for positional arguments
    
    argparse makes it awkward to implement the current cfg API where we
    simply return the leftover arguments to the caller.
    
    Very few cfg users actually rely on this functionality and for those
    cases it probably makes more sense for them to explicitly register
    positional arguments or sub-parsers.
    
    Add support for positional arguments via a 'required' Opt attribute:
    
      opt = StrOpt('foo', positional=True)
      conf.register_cli_opt(opt)
      conf(['bar'])
      conf.foo == 'bar'
    
    Change-Id: Iea746d710237e1ea26c1ef4871643941d1df09bd

commit b5f84bf854b4ea45ca87ae8cb4124cf317162f77
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Nov 12 16:25:58 2012 -0500

    Use stock argparse behaviour for optional args
    
    optparse would print "Options" but argparse prints "optional arguments".
    
    The default argparse behaviour is fine, let's stick with that.
    
    Change-Id: Ib53a2581af9d776e9a7c1cd90eebe89b70034e57

commit 6d3ff9cc3c5c1b9b2ef6dcc7689553d19e3f85b3
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Nov 12 16:25:57 2012 -0500

    Use stock argparse --usage behaviour
    
    optparse would substitute the program name for %prog, but argparse
    requires %(prog)s. Also, optparse would print 'Usage:' whereas
    argparse prints 'usage:'.
    
    Neither optparse behaviour that's worth retaining, let's just use
    the default argparse behaviour.
    
    Change-Id: Ied2acb37c366f1a45aed72b6b76f11e2de23828e

commit 17e3addb2b81ae0238d23e59133c2357d409144c
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Nov 12 16:25:55 2012 -0500

    Use stock argparse --version behaviour
    
    optparse prints the version to stdout, argparse prints the version to
    stderr.
    
    There's no need to preserve the old optparse behaviour, let's just stick
    with argparse behaviour.
    
    Change-Id: Ie141c72112a63149d098afa9db55a95a309e79d7

commit fb800a211f85b20301a273e39e9288edf23d0de8
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Nov 12 16:25:54 2012 -0500

    Remove add_option() method
    
    argparse just has an add_argument() method, so there's no reason for us
    to keep add_option() around.
    
    Change-Id: I6f4be089ceaf0fd8c4c99565af392b445916172e

commit 1ed027d07c150ad0cd1f9b6304e7a99c163243b9
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Nov 12 16:25:52 2012 -0500

    Completely remove cfg's disable_interspersed_args()
    
    The use case for disable_interspersed_args() is that nova-manage
    needs to be able to parse sub-commands. We now have a
    add_cli_subparsers() method which better supports this use case.
    
    Change-Id: I1fcd15889745fe4ddff0ac4bacf385004f9b61af

commit a4fcc4cb5916f0123d47ccdfbf4f1fee1c629401
Author: Laurence Miao <laurence.miao@gmail.com>
Date:   Sat Oct 6 21:08:14 2012 +0800

    argparse support for cfg
    
    * openstack/common/cfg.py
      Optparse is fading out since python 2.7, this
      patch will help openstack/common work on more
      advanded version of python(argparse).
      Now, disable_interspersed_args() has no effect.
      Added new method add_cli_subparsers, return
      argparse  subparser, for  usages  such  as
      subcommand.
    
    * tests/unit/test_cfg.py
      SubcommandTestCase added.
      Disabled test_disable_interspersed_args test entry
      for happiness of tox, temporarily.
      Modified test_help for port of argparse.
    
    * tools/pip-requires
      include argparse module for python 2.6
    
    Change-Id: Ie5cbde1a92a92786d64dea0ddfcfbf288c29f960

commit 12c7d781471566f57c1884ce4c4978565913741b
Author: David Ripton <dripton@redhat.com>
Date:   Wed Oct 31 13:14:52 2012 -0400

    Add a missing comma in a docstring.
    
    Change-Id: I1f75c7da1ab1543637198ecbb80a81b39ad35fde

commit 85851dbec62a11458204a5b32c60d63637b4aa0e
Author: Julien Danjou <julien@danjou.info>
Date:   Fri Oct 26 16:55:17 2012 +0200

    cfg: fix required if option has a dash
    
    If an option has a dash in it and is required, the check fails because it
    tries to self._get() on the name (with dash) rather than the dest (with
    underscore).
    
    Change-Id: I6448019f70f98bc2e58a325d0cf9ce88b8bb085b
    Signed-off-by: Julien Danjou <julien@danjou.info>

commit f4cf739349ed66d6944fd02b8f1c6c50bc23b4a9
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Aug 13 11:35:25 2012 +0100

    cfg: clean up None value handling
    
    Remove the need for an internal NoneValue class by making the existence
    of the 'default' or 'override' keys signify whether a default or
    override is set.
    
    Change-Id: Iacf49553df5ba8414307904a3ee334c7b8c55758

commit a88773ac6dcea042dc7e946316de013538969f0f
Author: Vishvananda Ishaya <vishvananda@gmail.com>
Date:   Fri Aug 10 14:28:59 2012 -0700

    Allow set_default and set_override to use None
    
    The current implementation interprets set_default('foo', None) and
    set_override('foo', None) as 'clear the existing default or override',
    which makes it impossible to override a value with None.
    
    This patch adds support for overriding with a None value by adding
    a special internal class. set_override('foo', None) will now override
    the existing value with None. This is a slight change to the existing
    behavior, so this patch adds two calls for the old functionality of
    clearing defaults and overrides. Example syntax for the new calls
    are shown below:
    
    conf.clear_default('foo')
    conf.clear_override('foo')
    
    The patch also updates the tests to reflect the change in functionality
    and adds new tests to verify the new functionality.
    
    Fixes bug 1035478
    
    Change-Id: Iee5e20e44da9bef6b86e0483ab0b48b625fe503c

commit ae93f9f03b0e0c03d3606c4fdbcfc4e2d4cdd55b
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Sat Aug 11 12:21:51 2012 +0100

    Tilde expansion for --config-file and --config-dir
    
    Fixes bug #1012671
    
    Allow a filename starting with ~ or ~user to be passed for
    --config-file or --config-dir.
    
    Change-Id: I67705401ed1c35c0cc2161095e36616552740aba

commit bf799b73eabb104804a120f174c5fcaa9ef9d3bb
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Tue Jul 31 12:16:28 2012 +0100

    Add import_opt() method to ConfigOpts
    
    Related to blueprint cfg-global-object
    
    When using the global config object pattern, you often have modules
    which define options that are referenced in other options.
    
    So, for example if module A defined option 'foo' and module be needed
    to reference that option, you might do:
    
      import A
      print CONF.foo
    
    However, this makes it entirely unclear to the casual reader why
    module A was imported.
    
    Nova has a flags.DECLARE() function that helps with this problem
    by allowing you to do:
    
      flags.DECLARE('foo', 'A')
    
    The function simply imports module A and checks that the 'foo'
    option is now defined in the global config object.
    
    This is fine, but it is also implicit that this function applies
    to the global config object. Instead, let's do the following:
    
      CONF.import_opt('foo', 'A')
    
    Change-Id: I7b98f5be71068bbde70cc0eab991eaebb577de52

commit c8f7c3c93948a8db22d6d9cfef75241210be3e1a
Author: Giampaolo Lauria <lauria@us.ibm.com>
Date:   Fri Jul 20 16:41:45 2012 -0400

    Modifies _is_opt_registered fcn to check for duplicate opts
    
    This change fixes bug 999307
    
    Currently, the check for duplicate options is done by checking
    whether they are the same object. The proposed fix is to check whether
    all the object fields have the same value.
    
    Change-Id: I2b72d630a0c8821df1d81e25d316d8d9195be492

commit dcce372a4e8d29cb3a0c0a8e468a777a71de415f
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Tue Jul 17 05:52:51 2012 +0100

    cfg: allow empty config values
    
    Fixes bug #1025522
    
    Commit 83044a7 caused this to stop working in Quantum:
    
      api_extensions_path =
    
    and could only be worked around with:
    
      api_extensions_path = ""
    
    Change-Id: I8c1a57225a2c135e6baf567b8e71d61e974da4e2

commit b16b92a83abdaaf3e3007e71533f23703e959f86
Author: Vincent Untz <vuntz@suse.com>
Date:   Thu Jul 5 14:51:26 2012 +0200

    cfg: Fix typo in documentation
    
    with with -> with
    
    Change-Id: I7a524c024b05639ec7ab4d57b6a52f70a95d2235

commit 0ddee649f3053ba738209920c310315f5d2d38ac
Author: Gary Kotton <gkotton@redhat.com>
Date:   Sun Jun 17 04:05:37 2012 -0400

    Update common code to support pep 1.3.
    bug 1014216
    
    Change-Id: I3f8fa2e11c9d3f3d34fb20f65ce886bb9c94463d

commit 9d2119aa388a20f3f758e013382e0e8224178608
Author: Johannes Erdfelt <johannes.erdfelt@rackspace.com>
Date:   Fri Jun 8 17:08:14 2012 +0000

    Use 'is not None' instead of '!= None'
    
    Fixes bug 1010570
    
    pep8 suggests the former over the latter
    
    Change-Id: Ice3a3b1cc2eea9228fffb4ee40fc360ff79054a3

commit c103186111bfbefb5cf8dd84f79cfb63273d18c4
Author: Russell Bryant <rbryant@redhat.com>
Date:   Wed Jun 6 21:57:14 2012 -0400

    Fix a pep8 error.
    
    Change-Id: Iab7e703254a354764a5667425ef887b0afc89115

commit 38105a582eaa954cf306f4d4dfe1ca09df2419db
Author: Vishvananda Ishaya <vishvananda@gmail.com>
Date:   Wed Jun 6 11:26:18 2012 -0700

    Adds support for bol and eol spaces to ini files
    
      * Fixes bug 1009639
      * Adds tests
    
    Change-Id: Id00563dfcc6f143c3e86ec380d66cffc967b8c48

commit c377d8fafc2001b85152bb3688dd54758a1070ab
Author: Joe Gordon <jogo@cloudscaling.com>
Date:   Mon May 21 18:17:35 2012 -0700

    Add support to include config aliases
    
    Implements blueprint config-aliases
    
    * Supports loading deprecated aliased options from a config file
    * Supports using deprecated aliased CLI options
    * For MultiStrOpt Can use mix of name and alias
    
    Change-Id: I04678880bc8ee1f85335f5656367bd1437245c6e

commit 072addef6f2740fe46f2aba63dd4f95a0a1299df
Author: Kevin L. Mitchell <kevin.mitchell@rackspace.com>
Date:   Mon Jun 4 10:27:36 2012 -0500

    Fix pep8 errors.
    
    Fixes a couple of pep8 errors that appeared due to a pep8 tool update.
    
    Change-Id: Ida70b1fb962529d3a157f44dcf2e71af773a4431

commit 5b5a5ab240a4b82ddc128c3763b59e0ba51dd022
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Tue May 29 08:27:05 2012 +0100

    cfg: add a global CONF object
    
    Implements blueprint cfg-global-object
    
    Add an instance of the CommonConfigOpts class to the cfg module's
    global namespace.
    
    The usage pattern is:
    
      from openstack.common import cfg
    
      opts = [
          cfg.StrOpt('foo', default='blaa'),
          cfg.StrOpt('bar', default='blaa'),
          ]
    
      CONF = cfg.CONF
      CONF.register_opts(opts)
    
      def do_something_later():
          print CONF.foo, CONF.bar
    
      def main():
          CONF(project='pulsar')
    
    Change-Id: I77e87b1e186c243b2638a4b1c202f865249dafce

commit a4b2511dae14ea1cce7017c51d2730c870995cb2
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Tue May 29 08:27:05 2012 +0100

    cfg: add generators for iterating over all options
    
    We have a few places now where we do:
    
      for opt in self.opts:
          foo(opt)
    
      for group in self.groups:
          for opt in group.opts:
              foo(opt, group)
    
    Use generators to turn this into simply:
    
      for opt, group in self.all_opts():
          foo(opt, group)
    
    Change-Id: I7a32779c20caeb1bacb85528d7e36c3c18c6c16a

commit 75b581e93b41cec82da982d3278af97be6bc773f
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Tue May 29 08:27:05 2012 +0100

    cfg: move constructor args to __call__() args
    
    In order to effectively use a global ConfigOpts object, you
    need to be able to initialize the global object with none of
    the information we currently require at construction.
    
    By moving those constructor args to the __call__() method,
    we enable the global object usage model but also make the API
    generally more flexible. For example, you can now reset the
    object and re-use it for parsing a different set of config
    files with the same options.
    
    There are a couple of other minor behavior changes as a
    result:
    
      - print_usage() and print_help() no longer works before
        the object has been called to parse options
    
      - registration of duplicate short options are no longer
        detected until the options are parsed
    
      - the --config-file and --config-dir options aren't
        registered until just before parsing the options since
        the default set of config files can be specified at
        that time
    
      - find_file() can't be used until after the options have
        been parsed, again because of the late registration
        of --config-file and --config-dir
    
    Finally, an unregister_opt() method is added to support
    the re-registeration of the --config-file and --config-dir
    options.
    
    Change-Id: I650d8e299e92cbc5d10da47a7ce1b73ca8066bd0

commit c7e16af82215234f8d8c8f5617202f2e77a3e005
Author: Russell Bryant <rbryant@redhat.com>
Date:   Wed May 16 11:34:29 2012 -0400

    Run pep8 on tests.
    
    I noticed that pep8 wasn't running on the tests.  This patch fixes that,
    as well as a couple of pep8 errors in test_cfg.
    
    Change-Id: I4429bfe6813a2e9394efb1753cbebbadb9f23833

commit 09327ad97206a9856ced348cdd6a54cc66dcc5a8
Author: Joe Gordon <jogo@cloudscaling.com>
Date:   Mon May 14 13:36:42 2012 -0700

    Alphabetize imports in openstack/common/cfg.py
    
    In preparation for enabling alphabetized import checking in Nova
    
    Change-Id: I709fca6a121ba44df193757e5ad838de710c2f15

commit 6c9c437f29de055a600ecace43ef782c16ebf877
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Sat May 12 11:52:53 2012 +0100

    cfg: make reset() clear defaults and overrides
    
    Fixes bug #998396
    
    Both Nova and Keystone need to clear the overrides on their config
    object between test runs. It's reasonable to expect the reset()
    method would do this, so let's make it so.
    
    Also add a clear() method with the old behaviour.
    
    Change-Id: I192c5bb07e81f0fb844fa2fd429dc2e7133800de

commit cd5456bc78917751fe511a66b4115844a6e94976
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Thu May 10 14:25:19 2012 +0100

    cfg: automatically create option groups
    
    Implements blueprint cfg-auto-create-groups
    
    Remove the restriction that groups must be explicitly created. Often
    you only need a group to have a name (not e.g. a title or help string)
    so we can easily just auto-create groups for that case.
    
    Change-Id: I150ab3900e3aad0068b93487c8d396d21d26cfea

commit 358c9aa23b226d8f4ae8da3add1a7595126aeb32
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Thu May 10 14:25:19 2012 +0100

    cfg: allow options to be marked as required
    
    Implements blueprint cfg-required-options
    
    Add a 'required' flag to option schemas:
    
      StrOpt('foo', required=True)
    
    which causes a RequiredOptError exception to be raised if the
    user fails to supply a value for the option on the CLI or in
    a config file.
    
    Change-Id: Ied7bb25f0c1582c4991d0f212f4871b9358b73fb

commit 4e67d716e9ee64468fbccd32d40f33e805227aac
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Thu May 10 14:25:19 2012 +0100

    cfg: use a list comprehension instead of map()
    
    Change-Id: Iaccb71d83d957aae77fa0f6bc71952b899d3a159

commit ecdd78bf99d8f8fcf06c15a9a80ddb8e54364e36
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Tue May 1 08:59:18 2012 +0100

    New ConfigOpts.find_file() for locating conf files
    
    Most services have the need to locate files like api-paste.ini or
    policy.json.
    
    This new method attempts to find these files by looking alongside
    the config files already parsed by ConfigOpts and, failing that,
    falls back to a standard set of directories.
    
    Change-Id: I95897816485b88f78854df194cab7872d7c5452a

commit 59610964dd35b730582cbdb08671ae3bdadf2141
Author: Eoghan Glynn <eglynn@redhat.com>
Date:   Mon Apr 23 21:06:56 2012 +0100

    Support for directory source of config files
    
    Implements bp cfg-config-dir
    
    Allow multiple config files to be pulled in from a config directory,
    as opposed to individual config files being explicitly enumerated.
    
    This logic is enabled using the --config-dir=/path/to/config CLI option,
    causing config to be retrived from all matching /path/to/config/*.conf
    files.
    
    Sections may be re-opened across config files, and all config items
    must reside in an explicitly specified section (i.e. it does not default
    to [DEFAULT]). This behavior is unchanged.
    
    Change-Id: Ia29dffe82dfb4742dcf3e8d36b376d906a2492cf

commit 122dc40da12c86de673ea7a7d73f4b35f3748cf1
Author: Brian Waldon <bcwaldon@gmail.com>
Date:   Wed Apr 25 16:11:45 2012 -0700

    Provide file extension when when looking for files
    
    * Allow an extension to be passed to find_config files, defaulting to '.conf'
    
    Change-Id: I022a3b28d9067158e9ed0da741a5e72cb73af167

commit cb4acbfa481831c5e3e0494f1ef762c416603b19
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Tue Apr 24 14:56:07 2012 +0100

    Some refactoring of the cfg cache
    
    A fairly misc bunch of changes:
    
     - init cache before registering config-file and just let
       register_cli_opt() clear the empty cache
    
     - use @__clear_cache on set_default() and set_override() since
       these are just used by the unit tests and doing so allows us
       to kill _remove_from_cache()
    
     - use @__clear_cache on reset() too
    
     - remove recursion from _get() and the substitute param
    
     - just use (group_name, opt_name) as the cache key
    
    Change-Id: I66934e748eca9ec03e44d7f80a7e10d96a77d8eb

commit ad859c16310bce0ca38b6d57df7000e58792d45e
Author: Yuriy Taraday <yorik.sar@gmail.com>
Date:   Sat Apr 14 01:16:35 2012 +0400

    Add caching to openstack.common.cfg
    
    Speedup of 'nova list' benchmark by up to 40%, eliminates 3 lines in
    top-10 cProfile methods.
    
    Change-Id: I2d4636f94d88b4a7e38d1565fdd4d1b8a89e560e

commit aff6ff33fdeae4f3f95258c1eb8f5363eaefb9d3
Author: Rick Harris <rconradharris@gmail.com>
Date:   Thu Mar 29 04:51:09 2012 +0000

    Typofix, OptionGroup should be OptGroup.
    
    Change-Id: I67473bb847759ce719876e08f8a894e000f11bb3

commit fbdf1acb681df10e2723fa2b1dcc8626b311bc78
Author: Rick Harris <rconradharris@gmail.com>
Date:   Wed Mar 28 18:37:16 2012 +0000

    Use absolute import for iniparser.
    
    Fixes bug 967400
    
    Change-Id: I0c028f6b5285cd641dedbcea3132224e404b004e

commit 4b81c673c5f3334753c8bb9ecafa24edbfc0ba90
Author: Johannes Erdfelt <johannes.erdfelt@rackspace.com>
Date:   Wed Mar 14 22:24:14 2012 +0000

    Finish implementing MultiStrOpt
    
    Fixes bug 955308
    
    Previously only multiple string options from the CLI were supported.
    This change adds support for config files too and merges the results
    from both CLI and config files.
    
    Change-Id: I642408c03ed295fac050105fd4380940e876f228

commit 0e4f86ec0998779b1ef4a1ae72a985d823886ff4
Author: Eoghan Glynn <eglynn@redhat.com>
Date:   Thu Mar 22 16:54:26 2012 +0000

    Avoid leaking secrets into config logging.
    
    Implements bp cfg-password-options
    
    Allow options to be declared secret so that their value is
    obfuscated before logging.
    
    Change-Id: Ie2168d218b029d9c12fa5b48342cd5b17b2cc77a

commit dd018d72b9b38992b8ebb8645beb11958e527b47
Author: Joe Gordon <jogo@cloudscaling.com>
Date:   Tue Mar 13 17:25:19 2012 -0700

    Fix bug 954488
    
    Change-Id: I99b764310c575e70aff4a6790e8ba8f55e43deeb

commit ef808088f6e5f3b04b65e0b5b2bd18902f6a235b
Author: Doug Hellmann <doug.hellmann@dreamhost.com>
Date:   Fri Mar 9 11:11:16 2012 +0000

    fix restructuredtext formatting in docstrings
    
    blueprint sphinx-doc-cleanup
    bug 94516
    
    - Correct parameter declarations, list formatting, cross-references, etc.
    - We don't need "let" in generate_autodoc_index.sh since we aren't doing math.
    - Change conf.py to not prefix class and function names with full namespace
      in generated output to save width on the screen.
    
    Change-Id: I9adc8681951913fd291d03e7142146e9d46841df

commit bdebe01b1dc38cc6b0db598b070d97017eb39779
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Wed Feb 22 16:29:59 2012 +0000

    Add ConfigOpts.print_help()
    
    Keystone uses this optparse method.
    
    Change-Id: Ic840b2fb2234a12cd94ca671a8d90cd2affe3a5e

commit f8e620d53f00194d6bd137963e7ea3c160566773
Author: Zhongyue Luo <lzyeval@gmail.com>
Date:   Fri Jan 20 01:10:58 2012 -0500

    cfg: fix a small comment typo
    
    Change-Id: I2646d7e674ef3d1759558e820f051cc5e7f3b4ae

commit 8c589d5634c947a07c68bdfe69e6c063bda49528
Author: Zhongyue Luo <lzyeval@gmail.com>
Date:   Sun Feb 12 16:04:21 2012 +0800

    cfg: unneeded multiple inheritance
    
    Fixed bug #927650
    
    In python=<2.6, collections.Mapping inherits from
    collections.Sized, collections.Iterable, and collections.Container
    which are also subclasses of object.
    
    Change-Id: I6238c683324127abd9fb637748a10b6bdb2961e0

commit b54d839ba9cdc12778368c5037a8a2f0217f0969
Author: Zhongyue Luo <lzyeval@gmail.com>
Date:   Sat Feb 11 20:18:08 2012 +0800

    PEP8 cleanup (openstack-common)
    
    Fixes bug #930625
    
    Remove backslash continuations in openstack-common.
    Fix type checking taboos.
    
    Change-Id: I49ddb9ff5fa5af760dcfccb52cb4793b71e02f19

commit 76a2802013b29c2f5313b04325a3e9f075fff09b
Author: Zhongyue Luo <lzyeval@gmail.com>
Date:   Fri Feb 10 17:21:32 2012 +0000

    Backslash continuations (misc.)
    
    Fixes bug #925166
    
    This patch for packages which have few backslash continuations.
    
    Follow up patches will be for packages network, scheduler, virt,
    db/sqlalchemy, tests, and api/openstack.
    
    Change-Id: I4200010b47b33fa8b9115b5d379b543200f6668d

commit 93e99f7bb50b7b5f1b908cb8f633a63a85fdf50d
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Fri Feb 10 17:14:52 2012 +0000

    Disable ConfigParser interpolation (lp#930270)
    
    This breaks e.g.
    
      volume_name_template=volume-%08x
      instance_name_template=instance-%08x
    
    and is not part of the API contract anyway. We use $opt based value
    interpolation.
    
    Change-Id: I7ba566ae7c9a77322b52c67c5e1ffbffb760f0fc

commit 75b173c0b567ae7e975747b74cd983806027f8ac
Author: Anthony Young <sleepsonthefloor@gmail.com>
Date:   Thu Feb 2 23:28:24 2012 +0000

    Implements blueprint separate-nova-volumeapi
    
    [...]
     ** Removes flag osapi_extension and replaces with osapi_compute_extension and osapi_volume_extension
    [...]
    
    Change-Id: I4c2e57c3cafd4e1a9e2ff3ce201c8cf28326afcd

commit 93d043c77c51804f69bb9b6e30359d5713b6e671
Author: Vishvananda Ishaya <vishvananda@gmail.com>
Date:   Thu Feb 2 23:25:53 2012 +0000

    Makes common/cfg.py raise AttributeError
    
      * fixes bug 915039
      * includes test
    
    Change-Id: I67b886be3b5af3763f52fffe54085975d61d61eb

commit 6f986bdcd603777541e38c8f5e4acb7991f3a2f3
Author: lzyeval <lzyeval@gmail.com>
Date:   Thu Feb 2 23:24:46 2012 +0000

    PEP8 type comparison cleanup
    
    Fixes bug #910295
    
    The None, True, and False values are singletons.
    
    All variable *comparisons* to singletons should use 'is' or 'is not'.
    All variable *evaluations* to boolean should use 'if' or 'if not'.
    
    "== None", "== True", "== False", and "!= None" comparisons in sqlalchemy's
    where(), or_(), filter(), and_(), and select() functions should not be changed.
    
    Incorrect comparisons or evaluations in comments were not changed.
    
    Change-Id: I087f0883bf115b5fe714ccfda86a794b9b2a87f7

commit b0a32ca9e4e49103d8b6c615fddb2febfe53c595
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Fri Jan 27 20:01:39 2012 +0000

    Add the Mapping interface to cfg.ConfigOpts
    
    Implements blueprint cfg-mapping interface
    
    With cfg, option values are accessed via attributes on ConfigOpts
    objects e.g.
    
      conf = ConfigOpts()
      conf.register_opt(StrOpt('foo'))
      conf()
      print conf.foo
    
    One use case that isn't easily supported with option values represented
    this way is iterating over all the registered options. Standard
    interfaces for listing attributes on an object aren't suitable because
    they will list more than just the options.
    
    For this use case alone, it's worth having ConfigOpts implement the
    mapping interface. That way we can do e.g.
    
      for opt, value in conf.items():
          print "Option %s = %s" % (opt, value)
    
    It's interesting to compare argparse's approach to this problem - option
    values are attributes on a Namespace object which has no attributes or
    methods to pollute the namespace of option names. This is a nice
    approach, but would mean that we would be passing around both a
    ConfigOpts object and a Namespace object. That's a bit too much overhead,
    and the mapping interface provides a usable workaround where there is a
    conflict.
    
    Change-Id: Ic113919a20291048f962999229c76884ebdd5ad8

commit 23e2286135e1454a7dafdea7ff239abf594c07ce
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Jan 23 15:21:50 2012 +0000

    Add cfg test case for recursive substitution
    
    i.e. test that if blaa='blaa', foo='$blaa' and bar='$foo' that
    the value of bar after substitutions is 'blaa'
    
    Change-Id: I01d370832a871603b7cb47bfb3546f6aaad8c34d

commit ac1c384499049fedec1bc934697d406f6854f0c4
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Mon Jan 23 10:42:07 2012 +0000

    Add support to cfg for disabling interspersed args
    
    Implements blueprint cfg-disable-interspersed-args
    
    Nova currently relies on cfg being implemented with optparse because it
    uses optparse's disable_interspersed_args()
    
    The use case for this is if you do:
    
      $> nova-manage --verbose create --project foo --user bar
    
    you want invoking ConfigOpts() to return:
    
      ['create', '--project', 'foo', '--user', 'bar']
    
    as the "extra" args rather than aborting when it doesn't recognize the
    --project arg.
    
    This is a reasonable use case for cfg to support and it should just have
    {disable,enable}_interspersed_args() methods.
    
    If we ever switch from optparse to argparse, we'll do something like this:
    
      parser.add_argument('--verbose')
      ...
      parser.add_argument(
        'extra_args',
        nargs=argparse.REMAINDER if disable_interspersed_args else '*')
      ...
      ns = parser.parse_args(...)
      extra_args = ns.extra_args
    
    i.e. we will need an 'extra_args' multi-value positional argument in any
    case and we'll just pass nargs=REMAINDER if we want trailing options to
    be included in the extra args.
    
    Change-Id: I3ecb7dc18230327cf5aaaa7d832224e64aafa40c

commit a877ff9d785a4ae8df5f52d8b3bd79670367cd8e
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Thu Jan 12 07:04:39 2012 +0000

    Get cfg test cases to 100%
    
     * Test cfg.find_config_files() sys.argv usage
     * Test boolean values in cfg config files
     * Finish off incomplete cfg bad value test case
     * Test register_opts() and register_cli_opts()
     * Test the quiet ignoring of option/group re-registration
     * Test cfg print_usage()
     * Test explicit option group titles
    
    Change-Id: Icbe4b7c48d4785551f06873821d1be758adf942c

commit 9b4801f236d8613a336b976648c7a6c51c6e411e
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Thu Jan 12 06:55:25 2012 +0000

    Add cfg test case for exceptions' __str__ methods
    
    Increases coverage from 93% to 97%
    
    Change-Id: I6a41b31e29238831fe2a888d5d64dc0bffd770c0

commit b8ce5777af175bbe7c47896e80ecd378407c0393
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Thu Jan 12 06:33:00 2012 +0000

    Fix some cfg test case naming conflicts
    
    As pointed out by Vish, there are duplicates of:
    
      OverridesTestCase::test_default_override
      OverridesTestCase::test_override
      SadPathTestCase::test_conf_file_not_found
    
    Also, rename the ConfigFileOptsTestCase so it is obvious they
    don't clash with the tests by the same name in CliOptsTestCase.
    
    Change-Id: I1d650d05d32501623cfed8f0b6399858d101ae02

commit ab488789f514fc6584913bf7b81945b59a72dbf5
Author: Mark McLoughlin <markmc@redhat.com>
Date:   Tue Jan 10 20:51:49 2012 +0000

    Add new cfg module
    
    As described here:
    
      http://wiki.openstack.org/CommonConfigModule
    
    The module implements an API for defining configuration options and
    reading values for those options that a user may have set in a config
    file or on the command line.
    
    The module is successfully in use in both Nova and Glance. Some work
    remains in Nova to switch from using it under a gflags compatible
    shim layer, but Glance is using it fully.
    
    There doesn't appear to be any blockers to other projects moving over
    to it fairly easily. Swift would perhaps be the next project to tackle.
    
    Just to go through potential future compatibility concerns:
    
      - Nova (the scroundrel) hackily uses the private ConfigOpts::_oparser
        in order to disable interspersed args. This was just for nova-manage
        and can probably be resolved some other way. In any case, Nova
        shouldn't switch to openstack-common's cfg API until it removes this
        hack.
    
      - the CommonConfigOpts subclass set of logging related options is
        perhaps assuming too much about what configuration options should
        be common across all the projects. However, it seems a fairly sane
        set and the worst that can happen is that projects avoid using it.
    
      - the parameters to the Opt constructor fairly closely mirror
        optparse, but they're fairly generic and shouldn't prevent us from
        switching to e.g. argparse
    
      - stuff like %prog expansion in the ConfigOpt's usage ctor param is a
        similar concern, but it's a very minor concern.
    
      - find_config_files() search path is perhaps too much policy for
        openstack-common; however, it is probably as generic as it could
        be and projects which need a different policy can just not use
        the function
    
    On the whole, I think we're in good shape wrt future compatibility.
    
    Change-Id: I279a9db7806d80aff3b9b085b4a9e4fb193662f9