<?xml version="1.0" encoding="ISO-8859-1"?>
<?xml-stylesheet type="text/xsl" href="clixdoc.xsl" ?>

<clix:documentation xmlns='http://www.w3.org/1999/xhtml' xmlns:clix='http://bknr.net/clixdoc'>
  <clix:title>Hunchentoot - The Common Lisp web server formerly known as TBNL</clix:title>
  <clix:short-description>
    A full-featured web server written in Common Lisp offering things
    like HTTP/1.1 chunking, persistent connections, and SSL.  Includes
    a framework for building dynamic websites interactively.
  </clix:short-description>

  <h2>
    <a href="http://www.htg1.de/hunchentoot/hunchentoot.html"
       title="Click here for the Hunchentoot logo"
       class="noborder">
      <img align="top" width="93" height="45" border="0" src="hunchentoot.gif" />
    </a>
    Hunchentoot - The Common Lisp web server formerly known as TBNL
  </h2>

  <blockquote>
    <clix:chapter name='abstract' title='Abstract'>
      <p>
        Hunchentoot is a web server written in Common Lisp and at the
        same time a toolkit for building dynamic websites.  As a
        stand-alone web server, Hunchentoot is capable of HTTP/1.1
        chunking (both directions), persistent connections
        (keep-alive), and SSL.
      </p>
      <p>
        Hunchentoot provides facilities like automatic session
        handling (with and without cookies), logging, customizable
        error handling, and easy access to GET and POST parameters
        sent by the client. It does <em>not</em> include functionality
        to programmatically generate HTML output. For this task you
        can use any library you like, e.g. (shameless self-plug)
        <a href="http://weitz.de/cl-who/">CL-WHO</a> or
        <a href="http://weitz.de/html-template/">HTML-TEMPLATE</a>.
      </p>
      <p>
        Hunchentoot talks with its front-end or with the client over
        TCP/IP sockets and optionally uses multiprocessing to handle
        several requests at the same time.  Therefore, it cannot be
        implemented completely in <a
        href="http://www.lispworks.com/documentation/HyperSpec/Front/index.htm">portable
        Common Lisp</a>.  It currently works with <a
        href="http://www.lispworks.com/">LispWorks</a> and all Lisps
        which are supported by the compatibility layers <a
        href="http://common-lisp.net/project/usocket/">usocket</a> and
        <a
        href="http://common-lisp.net/project/bordeaux-threads/">Bordeaux
        Threads</a>.
      </p>
      <p>
        Hunchentoot comes with a
        <a href="http://www.opensource.org/licenses/bsd-license.php">BSD-style
        license</a> so you can basically do with it whatever you want.
      </p>
      <p>
        Hunchentoot is (or was) for example used by
        <a href="http://quickhoney.com/">QuickHoney</a>,
        <a href="http://www.city-farming.de/">City Farming</a>,
        <a href="http://heikestephan.de/">Heike Stephan</a>.
      </p>
      <p>
        <font color="red">Download shortcut:</font>
        <a href="http://weitz.de/files/hunchentoot.tar.gz">http://weitz.de/files/hunchentoot.tar.gz</a>.
      </p>
    </clix:chapter>
  </blockquote>

  <clix:chapter name='contents' title='Contents'></clix:chapter>
  <clix:contents></clix:contents>

  <clix:chapter name="install" title="Download and installation">
    Hunchentoot depends on a couple of other Lisp libraries which you'll need
    to install first:
    <ul>
      <li>Pierre R. Mai's <a href="http://www.cliki.net/md5">MD5</a>,</li>
      <li>Kevin Rosenberg's <a href="http://www.cliki.net/cl-base64">CL-BASE64</a>,</li>
      <li>Janis Dzerins' <a href="http://common-lisp.net/project/rfc2388/">RFC2388</a>,</li>
      <li>Peter Seibel's <a href="http://weitz.de/cl-fad/">CL-FAD</a>,</li>
      <li>Gary King's <a href="http://common-lisp.net/project/trivial-backtrace/">trivial-backtrace</a>,</li>
      <li>Erik Huelsmann's <a href="http://common-lisp.net/project/usocket">usocket</a> (unless you're using LispWorks),</li>
      <li>Greg Pfeil's <a href="http://common-lisp.net/project/bordeaux-threads/">Bordeaux Threads</a> (unless you're using LispWorks),
      </li>
      <li>
        David Lichteblau's <a href="http://common-lisp.net/project/cl-plus-ssl/">CL+SSL</a>
        (unless you're using LispWorks),
      </li>
      <li>
        and my own <a href="http://weitz.de/flexi-streams/">FLEXI-STREAMS</a> (0.12.0 or higher),
        <a href="http://weitz.de/chunga/">Chunga</a> (1.0.0 or
        higher), and <a href="http://weitz.de/cl-ppcre/">
        CL-PPCRE</a> (plus
        <a href="http://weitz.de/cl-who/">CL-WHO</a> for the <a href="#teen-age">example code</a>
        and <a href="http://weitz.de/drakma/">Drakma</a> for the <a href="#testing">tests</a>).
      </li>
    </ul>

    Make sure to use the <em>newest</em> versions of all of these
    libraries (which might themselves depend on other libraries) - try
    the repository versions if you're in doubt.  Note: You can compile
    Hunchentoot without SSL support - and thus without the need to
    have CL+SSL - if you add <code>:HUNCHENTOOT-NO-SSL</code> to
    <a href="http://www.lispworks.com/documentation/HyperSpec/Body/v_featur.htm">
      <code>*FEATURES*</code></a> <em>before</em> you compile it.
    <p>
      Hunchentoot will only work with Lisps where
      the <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_c.htm#character_code">character
      codes</a> of
      all <a href="http://en.wikipedia.org/wiki/ISO/IEC_8859-1">Latin-1</a>
      characters coincide with their
      Unicode <a href="http://en.wikipedia.org/wiki/Code_point">code
      points</a> (which is the case for all current implementations I
      know).
    </p>
    <p>
      Hunchentoot itself together with this documentation can be
      downloaded from
      <a href="http://weitz.de/files/hunchentoot.tar.gz">http://weitz.de/files/hunchentoot.tar.gz</a>.
      The current version is <clix:library-version/>.
    </p>
    <p>
      The preferred method to compile and load Hunchentoot is via <a
      href="http://www.cliki.net/asdf">ASDF</a>.  If you want to avoid
      downloading and installing all the dependencies manually, give
      Zach Beane's excellent <a
      href="http://www.quicklisp.org/">Quicklisp</a> system a try.
    </p>
    <p>
      Hunchentoot and its dependencies can also be installed with <a
      href="http://common-lisp.net/project/clbuild/">clbuild</a>.
      There's also a port for <a
      href="http://www.gentoo.org/proj/en/lisp/common-lisp/index.xml">Gentoo
      Linux</a> thanks to Matthew Kennedy.
    </p>
    <p>
      The current development version of Hunchentoot can be found
      at <a href="https://github.com/edicl/hunchentoot">https://github.com/edicl/hunchentoot</a>.
      If you want to send patches, please fork the github repository and send pull requests.
    </p>

    <clix:subchapter name="port80" title="Running Hunchentoot on port 80">

      Hunchentoot does not come with code to help with running it on a
      privileged port (i.e. port 80 or 443) on Unix-like operating
      systems.  Modern Unix-like systems have specific, non-portable
      ways to allow non-root users to listen to privileged ports, so
      including such functionality in Hunchentoot was considered
      unnecessary.  Please refer to online resources for help.  At the
      time of this writing, the YAWS documentation has a <a
      href="http://yaws.hyber.org/privbind.yaws">comprehensive
      writeup</a> on the topic.
    </clix:subchapter>

    <clix:subchapter name="proxy" title="Hunchentoot behind a proxy">

      If you're feeling unsecure about exposing Hunchentoot to the wild,
      wild Internet or if your Lisp web application is part of a larger
      website, you can hide it behind a
      <a href="http://en.wikipedia.org/wiki/Proxy_server">proxy server</a>.
      One approach that I have used several times is to employ Apache's
      <a href="http://httpd.apache.org/docs/current/mod/mod_proxy.html">mod_proxy</a>
      module with a configuration that looks like this:

<pre><a href="http://httpd.apache.org/docs/current/mod/mod_proxy.html#proxypass" class="noborder">ProxyPass</a> /hunchentoot http://127.0.0.1:3000/hunchentoot
<a href="http://httpd.apache.org/docs/current/mod/mod_proxy.html#proxypassreverse" class="noborder">ProxyPassReverse</a> /hunchentoot http://127.0.0.1:3000/hunchentoot</pre>

      This will tunnel all requests where the URI path begins with
      <code>"/hunchentoot"</code> to a (Hunchentoot) server listening on
      port3000 on the same machine.

      <p>
        Of course, there are
        <a href="http://www.red-bean.com/pipermail/lispweb/2006-October/001342.html">several
          other</a> (more lightweight) web proxies that you could use
        instead of Apache.
      </p>
    </clix:subchapter>
  </clix:chapter>

  <clix:chapter name="support" title="Support">
    <p>
      The development version of Hunchentoot can be found <a
      href="https://github.com/edicl/hunchentoot" target="_new">on
      github</a>.  Please use the github issue tracking system to
      submit bug reports.  Patches are welcome, please use <a
      href="https://github.com/edicl/hunchentoot/pulls">GitHub pull
      requests</a>.  If you want to make a change, please <a
      href="http://weitz.de/patches.html" target="_new">read this
      first</a>.
    </p>
  </clix:chapter>

  <clix:chapter name="teen-age" title="Your own webserver (the easy teen-age New York version)">
    Starting your own web server is pretty easy.  Do something like this:
<pre>(hunchentoot:<a class="noborder" href="#teen-age">start</a> (make-instance 'hunchentoot:<a class="noborder" href="#acceptor">easy-acceptor</a> :port 4242))</pre>
    That's it.  Now you should be able to enter the address
    "<a href='http://127.0.0.1:4242/'><code>http://127.0.0.1:4242/</code></a>" in
    your browser and see something, albeit nothing very interesting
    for now.

    <p>
      By default, Hunchentoot serves the files from the
      <code><i>www/</i></code> directory in its source tree.  In the
      distribution, that directory contains a HTML version of the
      documentation as well as the error templates.  The location of
      the document root directory can be specified when creating a new
      <clix:ref>ACCEPTOR</clix:ref> instance by the way of the
      <clix:ref>ACCEPTOR-DOCUMENT-ROOT</clix:ref>.  Likewise, the
      location of the error template directory can be specified by the
      <clix:ref>ACCEPTOR-ERROR-TEMPLATE-DIRECTORY</clix:ref>.  Both
      <clix:ref>ACCEPTOR-DOCUMENT-ROOT</clix:ref> and
      <clix:ref>ACCEPTOR-ERROR-TEMPLATE-DIRECTORY</clix:ref> can be
      specified using a logical pathname, which will be translated
      once when the <clix:ref>ACCEPTOR</clix:ref> is instantiated.
    </p>

    <p>
      The <clix:ref>EASY-ACCEPTOR</clix:ref> class implements a
      framework for developing web applications.  Handlers are defined
      using the <clix:ref>DEFINE-EASY-HANDLER</clix:ref> macro.
      Request dispatching is performed according to the list of
      dispatch functions in <clix:ref>*DISPATCH-TABLE*</clix:ref>.
      Each of the functions on that list is called to determine
      whether it wants to handle the request, provided as single
      argument.  If a dispatcher function wants to handle the request,
      it returns another function to actually create the desired page.
    </p>

    <p>
      <clix:ref>DEFINE-EASY-HANDLER</clix:ref> is accompanied by a set
      of dispatcher creation functions that can be used to create
      dispatchers for standard tasks.  These are documented in the <a
      class="noborder" href="#easy-handlers">subchapter on easy
      handlers</a>
    </p>

    <p>
      Now be a bit more adventurous, try this
<pre>(hunchentoot:<a class="noborder" href="#define-easy-handler">define-easy-handler</a> (say-yo :uri "/yo") (name)
  (setf (hunchentoot:<a class="noborder" href="#content-type*">content-type*</a>) "text/plain")
  (format nil "Hey~@[ ~A~]!" name))</pre>
      and see what happens at "<a href='http://127.0.0.1:4242/yo'><code>http://127.0.0.1:4242/yo</code></a>" or
      "<a href='http://127.0.0.1:4242/yo?name=Dude'><code>http://127.0.0.1:4242/yo?name=Dude</code></a>" .
    </p>

    <p>
    Hunchentoot comes with a little example website which you can use
    to see if it works and which should also demonstrate a couple of
    the things you can do with Hunchentoot.  To start the example
    website, enter the following code into your listener:

<pre>(<a class="noborder" href="http://common-lisp.net/~mmommer/asdf-howto.shtml#sec11">asdf:oos</a> 'asdf:load-op :hunchentoot-test)</pre>

    Now go to "<a href='http://127.0.0.1:4242/hunchentoot/test'><code>http://127.0.0.1:4242/hunchentoot/test</code></a>" and play a bit.
    </p>
  </clix:chapter>

  <clix:chapter name="extras" title="Third party documentation and add-ons">
    <p>
      Adam Petersen has written a book called <a
      href="http://www.adampetersen.se/articles/lispweb.htm">"Lisp for
      the Web"</a> which explains how Hunchentoot and some other
      libraries can be used to build web sites.
    </p>
    <p>
      Here is some software which extends Hunchentoot or is based on it:
    </p>
    <ul>
      <li>
        <a href="http://weblocks-framework.info/">Weblocks</a> by
        Slava Akhmechet is a "continuations-based web framework" which
        is based on Hunchentoot.
      </li>
      <li>
        <a href="https://github.com/slyrus/hunchentoot-cgi">hunchentoot-cgi</a>
        (by Cyrus Harmon) provides
        <a href="http://en.wikipedia.org/wiki/Common_Gateway_Interface">CGI</a>
        handlers for Hunchentoot.
      </li>
      <li>
        <a href="http://weitz.de/cl-webdav/">CL-WEBDAV</a> is a <a href="http://webdav.org/">WebDAV</a>
        server based on Hunchentoot.
      </li>
      <li>
        <a href="http://restas.lisper.ru/">RESTAS</a> is a web
        framework based on Hunchentoot.
      </li>
    </ul>
  </clix:chapter>

  <clix:chapter name="reference" title="Function and variable reference">

    <clix:subchapter name="acceptors" title="Acceptors">

      If you want Hunchentoot to actually do something, you have to create and
      <a href="#teen-age">start</a> an <a href="#acceptor">acceptor</a>.
      You can also run several acceptors in one image, each one
      listening on a different different port.

      <clix:class name='acceptor'>
        <clix:description>
          To create a Hunchentoot webserver, you make an instance of
          this class or one of its subclasses and use the generic
          function <clix:ref>START</clix:ref> to start it (and
          <clix:ref>STOP</clix:ref> to stop it).  Use the
          <code>:port</code> initarg if you don&#039;t want to listen
          on the default http port 80.  If 0 is specified for the
          port, the system chooses a random port to listen on.  The
          port number choosen can be retrieved using the
          <clix:ref>ACCEPTOR-PORT</clix:ref> accessor.  The port
          number chosen is retained across stopping and starting the
          acceptor.
          <p>
            There are other initargs most of which you probably
            won&#039;t need very often.  They are explained in detail
            in the docstrings of the slot definitions.
          </p>
          <p>
            Unless you are in a Lisp without MP capabilities, you can
            have several active instances of
            <clix:ref>ACCEPTOR</clix:ref> (listening on different
            ports) at the same time.
          </p>
        </clix:description>
      </clix:class>

      <clix:class name='ssl-acceptor'>
        <clix:description>Create and <clix:ref>START</clix:ref> an instance of this class
        (instead of <clix:ref>ACCEPTOR</clix:ref>) if you want an https server.  There are two
        required initargs, <code>:SSL-CERTIFICATE-FILE</code> and <code>:SSL-PRIVATEKEY-FILE</code>, for
        pathname designators denoting the certificate file and the key file in
        PEM format.  On LispWorks, you can have both in one file in which case
        the second initarg is optional.  You can also use the
        <code>:SSL-PRIVATEKEY-PASSWORD</code> initarg to provide a password
        (as a string) for the key file (or <code>NIL</code>, the default, for
        no password).
        <p>
          The default port for <clix:ref>SSL-ACCEPTOR</clix:ref> instances is 443 instead of 80
        </p>
        </clix:description>
      </clix:class>

      <clix:function generic='true' name='start'>
        <clix:lambda-list>acceptor
        </clix:lambda-list>
        <clix:returns>acceptor
        </clix:returns>
        <clix:description>Starts <clix:arg>acceptor</clix:arg> so that it begins accepting
        connections.  Returns the acceptor.
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='stop'>
        <clix:lambda-list>acceptor &amp;key soft</clix:lambda-list>
        <clix:returns>acceptor
        </clix:returns>
        <clix:description>Stops the <clix:arg>acceptor</clix:arg> so
        that it no longer accepts requests.  If
        <clix:arg>soft</clix:arg> is true, and there are any requests
        in progress, wait until all requests are fully processed, but
        meanwhile do not accept new requests.  Note that
        <clix:arg>soft</clix:arg> must not be set when calling
        <clix:ref>stop</clix:ref> from within a request handler, as
        that will deadlock.
        </clix:description>
      </clix:function>

      <clix:special-variable name='*acceptor*'>
        <clix:description>The current ACCEPTOR object in the context of a request.
        </clix:description>
      </clix:special-variable>

      <clix:function generic='true' name='acceptor-listen-backlog'>
        <clix:lambda-list>listen-backlog
        </clix:lambda-list>
        <clix:returns>number-of-pending-connections
        </clix:returns>
        <clix:description>
          Number of pending connections allowed in the listen socket
          before the kernel rejects further incoming connections.
          Non-LispWorks only.
        </clix:description>
      </clix:function>

      <clix:readers generic='true'>
        <clix:listed-reader generic='true' name='acceptor-address'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>address
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='acceptor-port'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>port
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='acceptor-read-timeout'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>read-timeout
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='acceptor-ssl-certificate-file'>
          <clix:lambda-list>ssl-acceptor
          </clix:lambda-list>
          <clix:returns>ssl-certificate-file
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='acceptor-ssl-privatekey-file'>
          <clix:lambda-list>ssl-acceptor
          </clix:lambda-list>
          <clix:returns>ssl-privatekey-file
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='acceptor-ssl-privatekey-password'>
          <clix:lambda-list>ssl-acceptor
          </clix:lambda-list>
          <clix:returns>ssl-privatekey-password
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='acceptor-write-timeout'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>write-timeout
          </clix:returns>
        </clix:listed-reader>

        <clix:description>
          These are readers for various slots of <clix:ref>ACCEPTOR</clix:ref>
          objects (and some of them obviously only make sense
          for <clix:ref>SSL-ACCEPTOR</clix:ref> objects).  See the docstrings of
          these slots for more information and note that there are corresponding
          initargs for all of them.
        </clix:description>
      </clix:readers>

      <clix:accessors generic='true'>
        <clix:listed-accessor generic='true' name='acceptor-access-log-destination'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>(or pathname null)
          </clix:returns>
        </clix:listed-accessor>

        <clix:listed-accessor generic='true' name='acceptor-document-root'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>(or pathname null)
          </clix:returns>
        </clix:listed-accessor>

        <clix:listed-accessor generic='true' name='acceptor-error-template-directory'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>(or pathname null)
          </clix:returns>
        </clix:listed-accessor>

        <clix:listed-accessor generic='true' name='acceptor-input-chunking-p'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>input-chunking-p
          </clix:returns>
        </clix:listed-accessor>

        <clix:listed-accessor generic='true' name='acceptor-message-log-destination'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>(or pathname null)
          </clix:returns>
        </clix:listed-accessor>

        <clix:listed-accessor generic='true' name='acceptor-name'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>name
          </clix:returns>
        </clix:listed-accessor>

        <clix:listed-accessor generic='true' name='acceptor-output-chunking-p'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>output-chunking-p
          </clix:returns>
        </clix:listed-accessor>

        <clix:listed-accessor generic='true' name='acceptor-persistent-connections-p'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>persistent-connections-p
          </clix:returns>
        </clix:listed-accessor>

        <clix:listed-accessor generic='true' name='acceptor-reply-class'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>reply-class
          </clix:returns>
        </clix:listed-accessor>

        <clix:listed-accessor generic='true' name='acceptor-request-class'>
          <clix:lambda-list>acceptor
          </clix:lambda-list>
          <clix:returns>request-class
          </clix:returns>
        </clix:listed-accessor>

        <clix:description>
          These are accessors for various slots of <clix:ref>ACCEPTOR</clix:ref>
          objects.  See the docstrings of these slots for more information and
          note that there are corresponding initargs for all of them.
        </clix:description>
      </clix:accessors>

      <clix:function generic='true' name='acceptor-ssl-p'>
        <clix:lambda-list>acceptor
        </clix:lambda-list>
        <clix:returns>generalized-boolean
        </clix:returns>
        <clix:description>Returns a true value if <clix:arg>acceptor</clix:arg> uses SSL
        connections.  The default is to unconditionally return <code>NIL</code> and
        subclasses of <clix:ref>ACCEPTOR</clix:ref> must specialize this method to signal that
        they&#039;re using secure connections - see the <clix:ref>SSL-ACCEPTOR</clix:ref> class.
        </clix:description>
      </clix:function>

      <clix:special-variable name='*default-connection-timeout*'>
        <clix:description>The default connection timeout used when an
        acceptor is reading from and writing to a socket stream.  Note that
        some Lisps allow you to set different timeouts for reading and writing
        and you can specify both values via initargs when you create
        an <a href="#acceptors">acceptor</a>.
        </clix:description>
      </clix:special-variable>

      <clix:function generic='true' name='acceptor-remove-session'>
        <clix:lambda-list>acceptor session
        </clix:lambda-list>
        <clix:returns>|
        </clix:returns>
        <clix:description>
          This function is called whenever a session in
          <clix:ref>ACCEPTOR</clix:ref> is being destroyed because of
          a session timout or an explicit
          <clix:ref>REMOVE-SESSION</clix:ref> call.
        </clix:description>
      </clix:function>

    </clix:subchapter>

    <clix:subchapter name="acceptor-behaviour" title="Customizing acceptor behaviour">

      If you want to modify what acceptors do, you should subclass
      <clix:ref>ACCEPTOR</clix:ref> (or <clix:ref>SSL-ACCEPTOR</clix:ref>) and
      specialize the generic functions that constitute their behaviour (see
      example below).  The life of an acceptor looks like this: It is started
      with the function <clix:ref>START</clix:ref> which immediately calls
      <clix:ref>START-LISTENING</clix:ref> and then applies the function
      <clix:ref>EXECUTE-ACCEPTOR</clix:ref> to its <a
      href="#taskmasters">taskmaster</a>.  This function will eventually call
      <clix:ref>ACCEPT-CONNECTIONS</clix:ref> which is responsible for setting
      things up to wait for clients to connect.  For each incoming connection
      which comes in, <clix:ref>HANDLE-INCOMING-CONNECTION</clix:ref> is applied
      to the taskmaster which will either call
      <clix:ref>PROCESS-CONNECTION</clix:ref> directly, or will create a thread
      to call it.  <clix:ref>PROCESS-CONNECTION</clix:ref> calls
      <clix:ref>INITIALIZE-CONNECTION-STREAM</clix:ref> before it does anything
      else, then it selects and calls a function which handles the <a
      href="#requests">request</a>, and finally it sends the <a
      href="#replies">reply</a> to the client before it calls
      <clix:ref>RESET-CONNECTION-STREAM</clix:ref>.  If the connection is
      persistent, this procedure is repeated (except for the intialization step)
      in a loop until the connection is closed.  The acceptor is stopped with
      <clix:ref>STOP</clix:ref>.

      <p>
        If you just want to use the standard acceptors that come with
        Hunchentoot, you don't need to know anything about the functions
        listed in this section.
      </p>

      <clix:function generic='true' name='start-listening'>
        <clix:lambda-list>acceptor
        </clix:lambda-list>
        <clix:returns>|
        </clix:returns>
        <clix:description>Sets up a listen socket for the given acceptor and
        enables it to listen to incoming connections.  This function is called
        from the thread that starts the acceptor initially and may return
        errors resulting from the listening operation (like &#039;address in use&#039;
        or similar).
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='accept-connections'>
        <clix:lambda-list>acceptor
        </clix:lambda-list>
        <clix:returns>nil
        </clix:returns>
        <clix:description>In a loop, accepts a connection and hands it over
        to the acceptor's taskmaster for processing using
        <clix:ref>HANDLE-INCOMING-CONNECTION</clix:ref>. On LispWorks, this
        function returns immediately, on other Lisps it returns only once the
        acceptor has been stopped.
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='process-connection'>
        <clix:lambda-list>acceptor socket
        </clix:lambda-list>
        <clix:returns>nil
        </clix:returns>
        <clix:description>
          This function is called by the taskmaster when a new client
          connection has been established.  Its arguments are the
          <clix:ref>ACCEPTOR</clix:ref> object and a LispWorks socket
          handle or a usocket socket stream object in
          <clix:arg>socket</clix:arg>.  It reads the request headers,
          sets up the <a href="#requests">request</a> and <a
          href="#replies">reply</a> objects, and hands over to
          <clix:ref>PROCESS-REQUEST</clix:ref> which calls
          <clix:ref>HANDLE-REQUEST</clix:ref> to select and call a
          handler for the request and sends its reply to the client.
          This is done in a loop until the stream has to be closed or
          until a connection timeout occurs.  It is probably not a
          good idea to re-implement this method until you really,
          really know what you're doing.
          <p>
            Handlers may call to the
            <clix:ref>DETACH-SOCKET</clix:ref> generic function to
            indicate that no further requests should be handled on
            the connection by Hunchentoot, and that responsibility for
            the socket is assumed by third-party software.  This can
            be used by specialized handlers that wish to hand over
            connection polling or processing to functions outside of
            Hunchentoot, i.e. for connection multiplexing or
            implementing specialized client protocols.  Hunchentoot
            will finish processing the request and the
            <clix:ref>PROCESS-CONNECTION</clix:ref> function will
            return without closing the connection.  At that point,
            the acceptor may interact with the socket in whatever
            fashion required.
          </p>
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='detach-socket'>
        <clix:lambda-list>acceptor
        </clix:lambda-list>
        <clix:returns>stream
        </clix:returns>
        <clix:description>
          Indicate to Hunchentoot that it should stop serving requests
          on the current request's socket.  Hunchentoot will finish
          processing the current request and then return from
          <clix:ref>PROCESS-CONNECTION</clix:ref> without closing the
          connection to the client.
          <clix:ref>DETACH-SOCKET</clix:ref> can only be called from
          within a request handler function.
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='initialize-connection-stream'>
        <clix:lambda-list>acceptor stream
        </clix:lambda-list>
        <clix:returns>stream
        </clix:returns>
        <clix:description>
          Can be used to modify the stream which is used to
          communicate between client and server before the request is
          read.  The default method of <clix:ref>ACCEPTOR</clix:ref>
          does nothing, but see for example the method defined for
          <clix:ref>SSL-ACCEPTOR</clix:ref>.  All methods of this
          generic function <em>must</em> return the stream to use.
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='reset-connection-stream'>
        <clix:lambda-list>acceptor stream
        </clix:lambda-list>
        <clix:returns>stream
        </clix:returns>
        <clix:description>
          Resets the stream which is used to communicate
          between client and server after one request has been served so that it
          can be used to process the next request.  This generic function is
          called after a request has been processed and <em>must</em> return the
          stream.
        </clix:description>
      </clix:function>

      <clix:function name="acceptor-log-access" generic="true">
        <clix:lambda-list>acceptor &amp;key return-code</clix:lambda-list>
        <clix:description>
          Function to call to log access to the acceptor.  The
          <clix:arg>return-code</clix:arg> keyword argument contains additional
          information about the request to log.  In addition, it can use the
          standard request and reply accessor functions that are available to
          handler functions to find out more information about the request.
        </clix:description>
      </clix:function>

      <clix:function name="acceptor-log-message" generic="true">
        <clix:lambda-list>acceptor log-level format-string &amp;rest format-arguments</clix:lambda-list>
        <clix:description>
          Function to call to log messages by the <clix:arg>acceptor</clix:arg>.  It must accept
          a severity level for the message, which will be one of :ERROR, :INFO,
          or :WARNING, a format string and an arbitary number of formatting
          arguments.
        </clix:description>
      </clix:function>

      <clix:function name="acceptor-status-message" generic="true">
        <clix:lambda-list>acceptor http-return-code &amp;key &amp;allow-other-keys</clix:lambda-list>
        <clix:description>
          This function is called when a request's handler has been
          called but failed to provide content to send back to the
          client.  It converts the
          <clix:arg>HTTP-STATUS-CODE</clix:arg> to some request
          contents, typically a human readable description of the
          status code to be displayed to the user.

          If an ERROR-TEMPLATE-DIRECTORY is set in the current
          acceptor and the directory contains a file corresponding to
          HTTP-STATUS-CODE named &lt;code&gt;.html, that file is sent
          to the client after variable substitution.  Variables are
          referenced by ${&lt;variable-name&gt;}.

          Additional keyword arguments may be provided which are made
          available to the templating logic as substitution variables.
          These variables can be interpolated into error message
          templates in, which contains the current URL relative to the
          server and without GET parameters.

          In addition to the variables corresponding to keyword
          arguments, the script-name, lisp-implementation-type,
          lisp-implementation-version and hunchentoot-version
          variables are available.
        </clix:description>
      </clix:function>
    </clix:subchapter>

  <clix:subchapter name="subclassing-acceptors"
		   title="An example of how to subclass ACCEPTOR">

    This example shows how to subclass <clix:ref>ACCEPTOR</clix:ref> in order to
    provide Hunchentoot with basic virtual host support.&#x00A0; It assumes
    Hunchentoot is sitting behind an Internet-facing reverse-proxy web server
    that maps the host (or domain) part of incoming HTTP requests to unique
    localhost ports.

    <pre>(asdf:load-system "hunchentoot")
(asdf:load-system "drakma")

;;; Subclass ACCEPTOR
(defclass vhost (tbnl:acceptor)
  ;; slots
  ((dispatch-table
    :initform '()
    :accessor dispatch-table
    :documentation "List of dispatch functions"))
  ;; options
  (:default-initargs                    ; default-initargs must be used
   :address "127.0.0.1"))               ; because ACCEPTOR uses it

;;; Specialise ACCEPTOR-DISPATCH-REQUEST for VHOSTs
(defmethod tbnl:acceptor-dispatch-request ((vhost vhost) request)
  ;; try REQUEST on each dispatcher in turn
  (mapc (lambda (dispatcher)
	  (let ((handler (funcall dispatcher request)))
	    (when handler               ; Handler found. FUNCALL it and return result
	      (return-from tbnl:acceptor-dispatch-request (funcall handler)))))
	(dispatch-table vhost))
  (call-next-method))

;;; ======================================================================
;;; Now all we need to do is test it

;;; Instantiate VHOSTs
(defvar vhost1 (make-instance 'vhost :port 50001))
(defvar vhost2 (make-instance 'vhost :port 50002))

;;; Populate each dispatch table
(push
 (tbnl:create-prefix-dispatcher "/foo" 'foo1)
 (dispatch-table vhost1))
(push
 (tbnl:create-prefix-dispatcher "/foo" 'foo2)
 (dispatch-table vhost2))

;;; Define handlers
(defun foo1 () "Hello")
(defun foo2 () "Goodbye")

;;; Start VHOSTs
(tbnl:start vhost1)
(tbnl:start vhost2)

;;; Make some requests
(drakma:http-request "http://127.0.0.1:50001/foo")
;;; =|
;;; 127.0.0.1 - [2012-06-08 14:30:39] "GET /foo HTTP/1.1" 200 5 "-" "Drakma/1.2.6 (SBCL 1.0.56; Linux; 2.6.32-5-686; http://weitz.de/drakma/)"
;;; =>
;;; "Hello"
;;; 200
;;; ((:CONTENT-LENGTH . "5") (:DATE . "Fri, 08 Jun 2012 14:30:39 GMT")
;;;  (:SERVER . "Hunchentoot 1.2.3") (:CONNECTION . "Close")
;;;  (:CONTENT-TYPE . "text/html; charset=utf-8"))
;;; #&lt;PURI:URI http://127.0.0.1:50001/foo&gt;
;;; #&lt;FLEXI-STREAMS:FLEXI-IO-STREAM {CA90059}&gt;
;;; T
;;; "OK"
(drakma:http-request "http://127.0.0.1:50002/foo")
;;; =|
;;; 127.0.0.1 - [2012-06-08 14:30:47] "GET /foo HTTP/1.1" 200 7 "-" "Drakma/1.2.6 (SBCL 1.0.56; Linux; 2.6.32-5-686; http://weitz.de/drakma/)"
;;; =&gt;
;;; "Goodbye"
;;; 200
;;; ((:CONTENT-LENGTH . "7") (:DATE . "Fri, 08 Jun 2012 14:30:47 GMT")
;;;  (:SERVER . "Hunchentoot 1.2.3") (:CONNECTION . "Close")
;;;  (:CONTENT-TYPE . "text/html; charset=utf-8"))
;;; #&lt;PURI:URI http://127.0.0.1:50002/foo&gt;
;;; #&lt;FLEXI-STREAMS:FLEXI-IO-STREAM {CAE8059}&gt;
;;; T
;;; "OK"</pre>

    How to make each VHOST write to separate access log streams (or files) is
    left as an exercise to the reader.

  </clix:subchapter>

    <clix:subchapter name="taskmasters" title="Taskmasters">
      As a "normal" Hunchentoot user, you can completely ignore
      taskmasters and skip this section.  But if you're still reading,
      here are the dirty details: Each <a
      href="#acceptors">acceptor</a> has a taskmaster associated with
      it at creation time.  It is the taskmaster's job to distribute
      the work of accepting and handling incoming connections.  The
      acceptor calls the taskmaster if appropriate and the taskmaster
      calls back into the acceptor.  This is done using the generic
      functions described in this and the <a
      href="#acceptor-behaviour">previous</a> section.  Hunchentoot
      comes with two standard taskmaster implementations - one (which
      is the default used on multi-threaded Lisps) which starts a new
      thread for each incoming connection and one which handles all
      requests sequentially.  It should for example be relatively
      straightforward to create a taskmaster which allocates threads
      from a fixed pool instead of creating a new one for each
      connection.

      <p>
        You can control the resources consumed by a threaded taskmaster via
        two initargs. <code>:max-thread-count</code> lets you set the maximum
        number of request threads that can be processes simultaneously.  If
        this is <code>nil</code>, the is no thread limit imposed.

        <code>:max-accept-count</code> lets you set the maximum number of requests
        that can be outstanding (i.e. being processed or queued for processing).

        If <code>:max-thread-count</code> is supplied and <code>:max-accept-count</code>
        is <code>NIL</code>, then a <clix:ref>+HTTP-SERVICE-UNAVAILABLE+</clix:ref>
        error will be generated if there are more than the max-thread-count
        threads processing requests.  If both <code>:max-thread-count</code>
        and <code>:max-accept-count</code> are supplied, then max-thread-count
        must be less than max-accept-count; if more than max-thread-count
        requests are being processed, then requests up to max-accept-count
        will be queued until a thread becomes available.  If more than
        max-accept-count requests are outstanding, then a <clix:ref>+HTTP-SERVICE-UNAVAILABLE+</clix:ref>
        error will be generated.

        In a load-balanced environment with multiple Hunchentoot servers, it's
        reasonable to provide <code>:max-thread-count</code> but leave
        <code>:max-accept-count</code> null.   This will immediately result
        in <clix:ref>+HTTP-SERVICE-UNAVAILABLE+</clix:ref> when one server is
        out of resources, so the load balancer can try to find another server.

        In an environment with a single Hunchentoot server, it's reasonable
        to provide both <code>:max-thread-count</code> and a somewhat larger value
        for <code>:max-accept-count</code>.  This will cause a server that's almost
        out of resources to wait a bit; if the server is completely out of resources,
        then the reply will be <clix:ref>+HTTP-SERVICE-UNAVAILABLE+</clix:ref>.
        The default for these values is 100 and 120, respectively.
      </p>

      <p>
        If you want to implement your own taskmasters, you should subclass
        <clix:ref>TASKMASTER</clix:ref> or one of its subclasses,
        <clix:ref>SINGLE-THREADED-TASKMASTER</clix:ref> or
        <clix:ref>ONE-THREAD-PER-CONNECTION-TASKMASTER</clix:ref>, and
        specialize the generic functions in this section.
      </p>

      <clix:class name='taskmaster'>
        <clix:description>
          An instance of this class is responsible for distributing
          the work of handling requests for its acceptor.  This is an
          "abstract" class in the sense that usually only instances of
          subclasses of <clix:ref>TASKMASTER</clix:ref> will be used.
        </clix:description>
      </clix:class>

      <clix:class name='one-thread-per-connection-taskmaster'>
        <clix:description>
          A taskmaster that starts one thread for listening to
          incoming requests and one thread for each incoming
          connection.
          <p>
            This is the default taskmaster implementation for multi-threaded Lisp
            implementations.
          </p>
        </clix:description>
      </clix:class>

      <clix:class name='single-threaded-taskmaster'>
        <clix:description>
          A taskmaster that runs synchronously in the
          thread where the <clix:ref>START</clix:ref> function was invoked (or
          in the case of LispWorks in the thread started
          by <a href="http://www.lispworks.com/documentation/lw51/LWRM/html/lwref-61.htm#marker-910861"><code>COMM:START-UP-SERVER</code></a>).
          This is the simplest possible taskmaster implementation in that its
          methods do nothing but calling their acceptor &quot;sister&quot;
          methods - <clix:ref>EXECUTE-ACCEPTOR</clix:ref> calls <clix:ref>ACCEPT-CONNECTIONS</clix:ref>,
          <clix:ref>HANDLE-INCOMING-CONNECTION</clix:ref> calls <clix:ref>PROCESS-CONNECTION</clix:ref>.
        </clix:description>
      </clix:class>

      <clix:class name='multi-threaded-taskmaster'>
        <clix:description>
          This is an abstract class for taskmasters that use multiple threads;
          it is not a concrete class and you should not instantiate it with
          <code>MAKE-INSTANCE</code>.
          Instead, you should instantiate its subclass
          <clix:ref>ONE-THREAD-PER-CONNECTION-TASKMASTER</clix:ref> described above.
          <clix:ref>MULTI-THREADED-TASKMASTER</clix:ref>
          is intended to be inherited from by extensions to Hunchentoot,
          such as <a href="http://common-lisp.net/project/qitab/">quux-hunchentoot</a>'s
          <code>THREAD-POOLING-TASKMASTER</code>,
          though at the moment, doing so only inherits one slot and one method,
          on <clix:ref>EXECUTE-ACCEPTOR</clix:ref>,
          to have it start a new thread for the acceptor,
          then saved in said slot.
        </clix:description>
      </clix:class>

      <clix:function generic='true' name='execute-acceptor'>
        <clix:lambda-list>taskmaster
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>This is a callback called by the acceptor once it
        has performed all initial processing to start listening for incoming
        connections (see <clix:ref>START-LISTENING</clix:ref>).  It usually calls the
        <clix:ref>ACCEPT-CONNECTIONS</clix:ref> method of the acceptor, but depending on the
        taskmaster instance the method might be called from a new thread.
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='handle-incoming-connection'>
        <clix:lambda-list>taskmaster socket
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>
          This function is called by the acceptor to start
          processing of requests on a new incoming connection.  <clix:arg>socket</clix:arg> is the
          usocket instance that represents the new connection (or a socket
          handle on LispWorks).  The taskmaster starts processing requests on
          the incoming connection by calling the <clix:ref>PROCESS-CONNECTION</clix:ref>
          method of the acceptor instance.  The <clix:arg>socket</clix:arg> argument is passed to
          <clix:ref>PROCESS-CONNECTION</clix:ref> as an argument.

          If the taskmaster is a multi-threaded taskmaster, <clix:ref>HANDLE-INCOMING-THREAD</clix:ref>
          will call <clix:ref>CREATE-REQUEST-HANDLER-THREAD</clix:ref>, which will call
          <clix:ref>PROCESS-CONNECTION</clix:ref> in a new thread.
          <clix:ref>HANDLE-INCOMING-THREAD</clix:ref> might issue a
          <clix:ref>+HTTP-SERVICE-UNAVAILABLE+</clix:ref> error
          if there are too many request threads or it might block waiting for a
          request thread to finish.
              </clix:description>
            </clix:function>

            <clix:function generic='true' name='start-thread'>
            <clix:lambda-list>taskmaster thunk &amp;key
            </clix:lambda-list>
            <clix:returns>thread
            </clix:returns>
              <clix:description>This function is a callback that
                starts a new thread that will call the given <clix:arg>thunk</clix:arg>
                in the context of the proper <clix:arg>taskmaster</clix:arg>,
                with appropriate context-dependent keyword arguments.
                <clix:ref>ONE-THREAD-PER-CONNECTION-TASKMASTER</clix:ref> uses it in
                <clix:ref>EXECUTE-ACCEPTOR</clix:ref>
                and <clix:ref>CREATE-REQUEST-HANDLER-THREAD</clix:ref>,
                but specialized taskmasters may define more functions that use it.
                By default, it just creates a thread calling the thunk
                with a specified <clix:arg>name</clix:arg> keyword argument.
                Specialized taskmasters may wrap special bindings and condition handlers
                around the thunk call, register the thread in a management table, etc.
              </clix:description>
            </clix:function>

            <clix:function generic='true' name='create-request-handler-thread'>
            <clix:lambda-list>taskmaster socket
            </clix:lambda-list>
            <clix:returns>thread
            </clix:returns>
              <clix:description>This function is called by <clix:ref>HANDLE-INCOMING-THREAD</clix:ref>
          to create a new thread which calls <clix:ref>PROCESS-CONNECTION</clix:ref>.
          If you specialize this function, you must be careful to have the thread
          call <clix:ref>DECREMENT-TASKMASTER-REQUEST-COUNT</clix:ref> before
          it exits.  A typical method will look like this:

                <pre>(defmethod create-request-handler-thread ((taskmaster monitor-taskmaster) socket)
  (bt:make-thread
   (lambda ()
     (with-monitor-error-handlers
         (unwind-protect
              (with-monitor-variable-bindings
                  (process-connection (taskmaster-acceptor taskmaster) socket))
           (decrement-taskmaster-request-count taskmaster))))))</pre>







        </clix:description>
      </clix:function>

      <clix:function generic='true' name='shutdown'>
        <clix:lambda-list>taskmaster
        </clix:lambda-list>
        <clix:returns>taskmaster
        </clix:returns>
        <clix:description>Shuts down the taskmaster, i.e. frees all resources
        that were set up by it.  For example, a multi-threaded taskmaster
        might terminate all threads that are currently associated with it.
        This function is called by the acceptor's <clix:ref>STOP</clix:ref> method.
        </clix:description>
      </clix:function>

      <clix:accessor generic='true' name='taskmaster-acceptor'>
        <clix:lambda-list>taskmaster
        </clix:lambda-list>
        <clix:returns>acceptor
        </clix:returns>
        <clix:description>
          This is an accessor for the slot of a <clix:ref>TASKMASTER</clix:ref>
          object that links back to the <a href="#acceptors">acceptor</a> it is
          associated with.
        </clix:description>
      </clix:accessor>

    </clix:subchapter>

    <clix:subchapter name="request-dispatch" title="Request dispatch and handling">

      The main job of <clix:ref>HANDLE-REQUEST</clix:ref> is to select
      and call a function which handles the request, i.e. which looks
      at the data the client has sent and prepares an appropriate
      reply to send back.  This is by default implemented as follows:
      <p>
        The ACCEPTOR class defines a
        <clix:ref>ACCEPTOR-DISPATCH-REQUEST</clix:ref> generic
        function which is used to actually dispatch the request.  This
        function is called by the default method of
        <clix:ref>HANDLE-REQUEST</clix:ref>.  Each
        <clix:ref>ACCEPTOR-DISPATCH-REQUEST</clix:ref> method looks at
        the request object and depending on its contents decides to
        either handle the request or call the next method.
      </p>
      <p>
        In order to dispatch a request, Hunchentoot calls the
        <clix:ref>ACCEPTOR-DISPATCH-REQUEST</clix:ref> generic
        functions.  The method for <clix:ref>ACCEPTOR</clix:ref> tries
        to serve a static file relative to it's
        <clix:ref>ACCEPTOR-DOCUMENT-ROOT</clix:ref>.  Application
        specific acceptor subclasses will typically perform URL
        parsing and dispatching according to the policy that is
        required.
      </p>
      <p>
        The default method of <clix:ref>HANDLE-REQUEST</clix:ref> sets
        up <a href="#logging">standard logging and error handling</a>
        before it calls the acceptor's request dispatcher.
      </p>
      <p>
        Request handlers do their work by modifying
        the <a href="#replies">reply object</a> if necessary and by eventually
        returning the response body in the form of a string or a binary
        sequence.  As an alternative, they can also
        call <clix:ref>SEND-HEADERS</clix:ref> and write directly to a stream.
      </p>
    </clix:subchapter>

    <clix:subchapter name="easy-handlers" title="Using the easy-handler framework">
      <p>
        The <clix:ref>EASY-ACCEPTOR</clix:ref> class defines a method
        for <clix:ref>ACCEPTOR-DISPATCH-REQUEST</clix:ref> that walks
        through the list <clix:ref>*DISPATCH-TABLE*</clix:ref> which
        consists of <em>dispatch functions</em>.  Each of these
        functions accepts the request object as its only argument and
        either returns a request handler to handle the request or
        <code>NIL</code> which means that the next dispatcher in the
        list will be tried. A <em>request handler</em> is a function
        of zero arguments which relies on the special variable
        <clix:ref>*REQUEST*</clix:ref> to access the request instance
        being serviced. If all dispatch functions return
        <code>NIL</code>, the next
        <clix:ref>ACCEPTOR-DISPATCH-REQUEST</clix:ref> will be called.
      </p>
      <p>
        <strong>N.B.</strong> All functions and variables in this
        section are related to the easy request dispatch mechanism and
        are meaningless if you're using your own request dispatcher.
      </p>

      <clix:class name='easy-acceptor'>
        <clix:description>
          This class defines no additional slots with respect to
          <clix:ref>ACCEPTOR</clix:ref>.  It only serves as an
          additional type for dispatching calls to
          <clix:ref>ACCEPTOR-DISPATCH-REQUEST</clix:ref>.  In order to
          use the easy handler framework, acceptors of this class or
          one of its subclasses must be used.
        </clix:description>
      </clix:class>

      <clix:class name='easy-ssl-acceptor'>
        <clix:description>
          This class mixes the <clix:ref>SSL-ACCEPTOR</clix:ref> and
          the <clix:ref>EASY-ACCEPTOR</clix:ref> classes.  It is used
          when both ssl and the easy handler framework are required.
        </clix:description>
      </clix:class>

      <clix:special-variable name='*dispatch-table*'>
        <clix:description>
          A global list of dispatch functions.  The initial value is a
          list consisting of the symbol
          <clix:ref>DISPATCH-EASY-HANDLERS</clix:ref>.
        </clix:description>
      </clix:special-variable>

      <clix:function name="create-prefix-dispatcher">
        <clix:lambda-list>prefix handler</clix:lambda-list>
        <clix:returns>dispatch-fn</clix:returns>
        <clix:description>
          A convenience function which will return a dispatcher that
          returns <clix:arg>handler</clix:arg> whenever the path part of
          the request URI starts with the
          string <clix:arg>prefix</clix:arg>.
        </clix:description>
      </clix:function>

      <clix:function name="create-regex-dispatcher">
        <clix:lambda-list>regex handler</clix:lambda-list>
        <clix:returns>dispatch-fn</clix:returns>
        <clix:description>
          A convenience function which will return a dispatcher that
          returns <clix:arg>handler</clix:arg> whenever the path part of
          the request URI matches
          the <a href="http://weitz.de/cl-ppcre/">CL-PPCRE</a> regular
          expression <clix:arg>regex</clix:arg> (which can be a string, an
          s-expression, or a scanner).
        </clix:description>
      </clix:function>

      <clix:function name="create-folder-dispatcher-and-handler">
        <clix:lambda-list>uri-prefix base-path <clix:lkw>optional</clix:lkw> content-type</clix:lambda-list>
        <clix:returns>dispatch-fn</clix:returns>
        <clix:description>
          Creates and returns a dispatch function which will dispatch to
          a handler function which emits the file relative
          to <clix:arg>base-path</clix:arg> that is denoted by the URI of
          the request relative
          to <clix:arg>uri-prefix</clix:arg>.  <clix:arg>uri-prefix</clix:arg>
          must be a string ending with a
          slash, <clix:arg>base-path</clix:arg> must be a pathname
          designator for an existing directory.
          Uses <clix:ref>HANDLE-STATIC-FILE</clix:ref> internally.
          <p>
            If <clix:arg>content-type</clix:arg> is <em>not</em>
            <code>NIL</code>, it will be used as a the content type for
            all files in the folder.  Otherwise (which is the default)
            the content type of each file will be
            determined <a href="#handle-static-file">as usual</a>.
          </p>
        </clix:description>
      </clix:function>

      <clix:function name='create-static-file-dispatcher-and-handler'>
        <clix:lambda-list>uri path
        <clix:lkw>optional
        </clix:lkw> content-type
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>
          Creates and returns a request dispatch function which will
          dispatch to a handler function which emits the file denoted
          by the pathname designator PATH with content type
          CONTENT-TYPE if the SCRIPT-NAME of the request matches the
          string URI.  If CONTENT-TYPE is NIL, tries to determine the
          content type via the file&#039;s suffix.
        </clix:description>
      </clix:function>

      <clix:function macro="true" name="define-easy-handler">
        <clix:lambda-list>description lambda-list [[declaration* | documentation]] form*</clix:lambda-list>
        <clix:description>
          Defines a handler as if
          by <a href="http://www.lispworks.com/documentation/HyperSpec/Body/m_defun.htm">
            <code>DEFUN</code></a> and optionally registers it with a
          URI so that it will be found
          by <clix:ref>DISPATCH-EASY-HANDLERS</clix:ref>.
          <p>
            <clix:arg>description</clix:arg> is either a
            symbol <clix:arg>name</clix:arg> or a list matching the
            <a href="http://www.lispworks.com/documentation/HyperSpec/Body/03_de.htm">destructuring
              lambda list</a>
          </p>
          <pre>(name &amp;key uri acceptor-names default-parameter-type default-request-type).</pre>
          <clix:arg>lambda-list</clix:arg> is a list the elements of which
          are either a symbol <clix:arg>var</clix:arg> or a list matching
          the destructuring lambda list
          <pre>(var &amp;key real-name parameter-type init-form request-type).</pre>
          The resulting handler will be a Lisp function with the
          name <clix:arg>name</clix:arg> and keyword parameters named by
          the <clix:arg>var</clix:arg> symbols.
          Each <clix:arg>var</clix:arg> will be bound to the value of the
          GET or POST parameter called <clix:arg>real-name</clix:arg> (a
          string) before the body of the function is executed.
          If <clix:arg>real-name</clix:arg> is not provided, it will be
          computed
          by <a href="http://www.lispworks.com/documentation/HyperSpec/Body/f_stg_up.htm#string-downcase">downcasing</a>
          the symbol name of <clix:arg>var</clix:arg>.
          <p>
            If <clix:arg>uri</clix:arg> (which is evaluated) is provided,
            then it must be a string or
            a <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_f.htm#function_designator">function
              designator</a> for a unary function.  In this case, the
            handler will be returned
            by <clix:ref>DISPATCH-EASY-HANDLERS</clix:ref>,
            if <clix:arg>uri</clix:arg> is a string and
            the <a href="#script-name">script name</a> of the current
            request is <clix:arg>uri</clix:arg>, or
            if <clix:arg>uri</clix:arg> designates a function and applying
            this function to
            the <a href="#*request*">current <code>REQUEST</code>
              object</a> returns a true value.
          </p>
          <p>
            <clix:arg>acceptor-names</clix:arg> (which is evaluated) can be a
            list of symbols which means that the handler will only be
            returned by <clix:ref>DISPATCH-EASY-HANDLERS</clix:ref> in
            acceptors which have one of these names
            (see <clix:ref>ACCEPTOR-NAME</clix:ref>).  <clix:arg>acceptor-names</clix:arg> can also be the
            symbol <code>T</code> which means that the handler will be
            returned by <clix:ref>DISPATCH-EASY-HANDLERS</clix:ref>
            in <em>every</em> acceptor.
          </p>
          <p>
            Whether the GET or POST parameter (or both) will be taken into
            consideration, depends on <clix:arg>request-type</clix:arg>
            which can
            be <code>:GET</code>, <code>:POST</code>, <code>:BOTH</code>,
            or <code>NIL</code>.  In the last case, the value of
            <clix:arg>default-request-type</clix:arg> (the default of which
            is <code>:BOTH</code>) will be used.
          </p>
          <p>
            The value of <clix:arg>var</clix:arg> will usually be a string
            (unless it resulted from a <a href="#upload">file upload</a>
            in which case it won't be converted at all), but
            if <clix:arg>parameter-type</clix:arg> (which is evaluated) is
            provided, the string will be converted to another Lisp type by
            the following rules:
          </p>
          <p>
            If the corresponding GET or POST parameter wasn't provided by
            the client, <clix:arg>var</clix:arg>'s value will
            be <code>NIL</code>.  If <clix:arg>parameter-type</clix:arg>
            is <code>'STRING</code>,
            <clix:arg>var</clix:arg>'s value remains as is.
            If <clix:arg>parameter-type</clix:arg> is <code>'INTEGER</code>
            and the parameter string consists solely of decimal
            digits, <clix:arg>var</clix:arg>'s value will be the
            corresponding integer, otherwise <code>NIL</code>.
            If <clix:arg>parameter-type</clix:arg> is
            <code>'KEYWORD</code>, <clix:arg>var</clix:arg>'s value will be
            the keyword obtained
            by <a href="http://www.lispworks.com/documentation/HyperSpec/Body/f_intern.htm">interning</a>
            the <a href="http://www.lispworks.com/documentation/HyperSpec/Body/f_stg_up.htm#string-upcase">upcased</a>
            parameter string into
            the <a href="http://www.lispworks.com/documentation/HyperSpec/Body/11_abc.htm">keyword
              package</a>.  If <clix:arg>parameter-type</clix:arg>
            is <code>'CHARACTER</code> and the parameter string is of
            length one, <clix:arg>var</clix:arg>'s value will be the single
            character of this string, otherwise <code>NIL</code>.
            If <clix:arg>parameter-type</clix:arg>
            is <code>'BOOLEAN</code>, <clix:arg>var</clix:arg>'s value will
            always be <code>T</code> (unless it is <code>NIL</code> by the
            first rule above, of course).
            If <clix:arg>parameter-type</clix:arg> is any other atom, it is
            supposed to be
            a <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_f.htm#function_designator">function
              designator</a> for a unary function which will be called to
            convert the string to something else.
          </p>
          <p>
            Those were the rules for <em>simple</em> parameter types, but
            <clix:arg>parameter-type</clix:arg> can also be a list starting
            with one of the symbols
            <code>LIST</code>, <code>ARRAY</code>,
            or <code>HASH-TABLE</code>.  The second value of the list must
            always be a simple parameter type as in the last paragraph -
            we'll call it the <em>inner type</em> below.
          </p>
          <p>
            In the case of <code>'LIST</code>, all GET/POST parameters
            called <clix:arg>real-name</clix:arg> will be collected,
            converted to the inner type as by the rules above, and
            assembled into a list which will be the value of
            <clix:arg>var</clix:arg>.
          </p>
          <p>
            In the case of <code>'ARRAY</code>, all GET/POST parameters
            which have a name like the result of
          </p>
          <pre>(format nil "~A[~A]" real-name n)</pre>
          where <clix:arg>n</clix:arg> is a non-negative integer, will be
          assembled into an array where the <clix:arg>n</clix:arg>th element
          will be set accordingly, after conversion to the inner type.
          The array, which will become the value
          of <clix:arg>var</clix:arg>, will be big enough to hold all
          matching parameters, but not bigger.  Array elements not set as
          described above will be <code>NIL</code>.  Note
          that <code>VAR</code> will always be bound to an array, which
          may be empty, so it will never be <code>NIL</code>, even if no
          appropriate GET/POST parameters are found.
          <p>
            The full form of a <code>'HASH-TABLE</code> parameter type is
          </p>
          <pre>(hash-table inner-type key-type test-function)</pre>
          but <clix:arg>key-type</clix:arg>
          and <clix:arg>test-function</clix:arg> can be left out in which
          case they default to <code>'STRING</code>
          and <code>'EQUAL</code>, respectively.  For this parameter type,
          all GET/POST parameters which have a name like the result of
          <pre>(format nil "~A{~A}" real-name key)</pre>
          (where <clix:arg>key</clix:arg> is a string that doesn't contain
          curly brackets) will become the values (after conversion
          to <clix:arg>inner-type</clix:arg>) of a hash table with test
          function <clix:arg>test-function</clix:arg>
          where <clix:arg>key</clix:arg> (after conversion
          to <clix:arg>key-type</clix:arg>) will be the corresponding key.
          Note that <clix:arg>var</clix:arg> will always be bound to a hash
          table, which may be empty, so it will never be <code>NIL</code>,
          even if no appropriate GET/POST parameters are found.
          <p>
            To make matters even more complicated, the three compound
            parameter types also have an abbreviated form - just one of
            the symbols <code>LIST</code>, <code>ARRAY</code>,
            or <code>HASH-TABLE</code>.  In this case, the inner type will
            default to <code>'STRING</code>.
          </p>
          <p>
            If <clix:arg>parameter-type</clix:arg> is not provided
            or <code>NIL</code>, <clix:arg>default-parameter-type</clix:arg>
            (the default of which is <code>'STRING</code>) will be used
            instead.
          </p>
          <p>
            If the result of the computations above would be
            that <clix:arg>var</clix:arg> would be bound
            to <code>NIL</code>, then <clix:arg>init-form</clix:arg> (if
            provided) will be evaluated instead,
            and <clix:arg>var</clix:arg> will be bound to the result of this
            evaluation.
          </p>
          <p>
            Handlers built with this macro are constructed in such a way
            that the resulting Lisp function is useful even outside of
            Hunchentoot.  Specifically, all the parameter computations
            above will only happen if <clix:ref>*REQUEST*</clix:ref> is
            bound, i.e. if we're within a Hunchentoot request.
            Otherwise, <clix:arg>var</clix:arg> will always be bound to the
            result of evaluating <clix:arg>init-form</clix:arg> unless a
            corresponding keyword argument is provided.
          </p>
          <p>
            The <a href="#example">example code</a> that comes with
            Hunchentoot contains an example which demonstrates some of the
            features of <clix:ref>DEFINE-EASY-HANDLER</clix:ref>.
          </p>
        </clix:description>
      </clix:function>

      <clix:function name='dispatch-easy-handlers'>
        <clix:lambda-list>request
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>This is a dispatcher which returns the appropriate handler
        defined with <clix:ref>DEFINE-EASY-HANDLER</clix:ref>, if there is one.
        </clix:description>
      </clix:function>

    </clix:subchapter>

    <clix:subchapter name="requests" title="Request objects">

      For each incoming request, the <a href="#acceptors">acceptor</a> (in
      <clix:ref>PROCESS-CONNECTION</clix:ref>) creates a
      <clix:ref>REQUEST</clix:ref> object and makes it available to <a
      href="#request-dispatch">handlers</a> via the special variable
      <clix:ref>*REQUEST*</clix:ref>.  This object contains all relevant
      information about the request and this section collects the functions
      which can be used to query such an object.  In all function where
      <clix:arg>request</clix:arg> is an optional or keyword parameter, the
      default is <clix:ref>*REQUEST*</clix:ref>.

      <p>
        If you need more fine-grained control over the behaviour of request
        objects, you can subclass <clix:ref>REQUEST</clix:ref> and initialize
        the <a href="#acceptor-request-class"><code>REQUEST-CLASS</code></a>
        slot of the <clix:ref>ACCEPTOR</clix:ref> class accordingly.  The
        acceptor will generate request objects of the class named by this
        slot.
      </p>

      <clix:class name='request'>
        <clix:description>
          Objects of this class hold all the information
          about an incoming request.  They are created automatically by
          acceptors and can be accessed by the
          corresponding <a href="#request-dispatch">handler</a>.

          You should not mess with the slots of these objects directly, but you
          can subclass <clix:ref>REQUEST</clix:ref> in order to implement your
          own behaviour.  See
          the <a href="#acceptor-request-class"><code>REQUEST-CLASS</code></a>
          slot of the <clix:ref>ACCEPTOR</clix:ref> class.
        </clix:description>
      </clix:class>

      <clix:special-variable name='*request*'>
        <clix:description>The current REQUEST object while in the context of a request.
        </clix:description>
      </clix:special-variable>

      <clix:function name='real-remote-addr'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>string{, list}
        </clix:returns>
        <clix:description>
          Returns the &#039;<code>X-Forwarded-For</code>&#039; incoming http header as the
          second value in the form of a list of IP addresses and the first
          element of this list as the first value if this header exists.
          Otherwise returns the value of <clix:ref>REMOTE-ADDR</clix:ref> as the only value.
        </clix:description>
      </clix:function>

      <clix:function name='parameter'>
        <clix:lambda-list>name
        <clix:lkw>optional
        </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>string
        </clix:returns>
        <clix:description>
          Returns the GET or the POST parameter with name
          <clix:arg>name</clix:arg> (a string) - or <code>NIL</code>
          if there is none.  If both a GET and a POST parameter with
          the same name exist the GET parameter is returned.  Search
          is case-sensitive.  See also
          <clix:ref>GET-PARAMETER</clix:ref> and
          <clix:ref>POST-PARAMETER</clix:ref>.
        </clix:description>
      </clix:function>

      <clix:function name="get-parameter">
        <clix:lambda-list>name <clix:lkw>optional</clix:lkw> request</clix:lambda-list>
        <clix:returns>string</clix:returns>
        <clix:description>
          Returns the value of the GET parameter (as provided via the
          request URI) named by the string <clix:arg>name</clix:arg> as a
          string (or <code>NIL</code> if there ain't no GET parameter
          with this name). Note that only the first value will be
          returned if the client provided more than one GET parameter
          with the name <clix:arg>name</clix:arg>. See
          also <clix:ref>GET-PARAMETERS*</clix:ref>.
        </clix:description>
      </clix:function>

      <clix:function name="post-parameter">
        <clix:lambda-list>name <clix:lkw>optional</clix:lkw> request</clix:lambda-list>
        <clix:returns>string</clix:returns>
        <clix:description>
          Returns the value of the POST parameter (as provided in the
          request's body) named by the
          string <clix:arg>name</clix:arg>. Note that only the first value
          will be returned if the client provided more than one POST
          parameter with the name <clix:arg>name</clix:arg>.  This value
          will usually be a string (or <code>NIL</code> if there ain't
          no POST parameter with this name). If, however, the browser
          sent a <a class="none" name="upload">file</a> through
          a <a href="http://www.faqs.org/rfcs/rfc2388.html">
            <code>multipart/form-data</code>
          </a> form, the value of this function is a three-element list
          <pre>(path file-name content-type)</pre>
          where <clix:arg>path</clix:arg> is a pathname denoting the place
          were the uploaded file was
          stored, <clix:arg>file-name</clix:arg> (a string) is the file
          name sent by the browser, and <clix:arg>content-type</clix:arg>
          (also a string) is the content type sent by the browser. The
          file denoted by <clix:arg>path</clix:arg> will be deleted after
          the request has been handled - you have to move or copy it
          somewhere else if you want to keep it.
          <p>
            POST parameters will only be computed if the content type of
            the request body was <code>multipart/form-data</code>
            or <code>application/x-www-form-urlencoded</code>.  Although
            this function is called <code>POST-PARAMETER</code>, you can
            instruct Hunchentoot to compute these parameters for other
            request methods by
            setting <clix:ref>*METHODS-FOR-POST-PARAMETERS*</clix:ref>.
          </p>
          <p>
            See also <clix:ref>POST-PARAMETERS</clix:ref>
            and <clix:ref>*TMP-DIRECTORY*</clix:ref>.
          </p>
        </clix:description>
      </clix:function>

      <clix:function name="get-parameters*">
        <clix:lambda-list><clix:lkw>optional</clix:lkw> request</clix:lambda-list>
        <clix:returns>alist</clix:returns>
        <clix:description>
          Returns
          an <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_a.htm#alist">alist</a>
          of all GET parameters (as provided via the request
          URI). The <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_c.htm#car">car</a>
          of each element of this list is the parameter's name while
          the <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_c.htm#cdr">cdr</a>
          is its value (as a string). The elements of this list are in
          the same order as they were within the request URI. See
          also <clix:ref>GET-PARAMETER</clix:ref>.
        </clix:description>
      </clix:function>

      <clix:function name="post-parameters*">
        <clix:lambda-list><clix:lkw>optional</clix:lkw> request</clix:lambda-list>
        <clix:returns>alist</clix:returns>
        <clix:description>
          Returns
          an <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_a.htm#alist">alist</a>
          of all POST parameters (as provided via the request's
          body). The <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_c.htm#car">car</a>
          of each element of this list is the parameter's name while
          the <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_c.htm#cdr">cdr</a>
          is its value. The elements of this list are in the same order
          as they were within the request's body.
          <p>
            See also <clix:ref>POST-PARAMETER</clix:ref>.
          </p>
        </clix:description>
      </clix:function>

      <clix:special-variable name='*methods-for-post-parameters*'>
        <clix:description>A list of the request method types (as keywords) for which
        Hunchentoot will try to compute <clix:arg>post-parameters</clix:arg>.
        </clix:description>
      </clix:special-variable>

      <clix:function name='cookie-in'>
        <clix:lambda-list>name
        <clix:lkw>optional
        </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>string
        </clix:returns>
        <clix:description>
          Returns the cookie with the name <clix:arg>name</clix:arg> (a string) as sent by the
          browser - or <code>NIL</code> if there is none.
        </clix:description>
      </clix:function>

      <clix:function name='cookies-in*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>alist
        </clix:returns>
        <clix:description>Returns an alist of all cookies associated with the <clix:ref>REQUEST</clix:ref> object
        <clix:arg>request</clix:arg>.
        </clix:description>
      </clix:function>

      <clix:function name='host'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>host
        </clix:returns>
        <clix:description>Returns the &#039;Host&#039; incoming http header value.
        </clix:description>
      </clix:function>

      <clix:function name='query-string*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>string
        </clix:returns>
        <clix:description>
          Returns the query string of the <clix:ref>REQUEST</clix:ref> object <clix:arg>request</clix:arg>. That&#039;s
          the part behind the question mark (i.e. the GET parameters).
        </clix:description>
      </clix:function>

      <clix:function name='referer'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>
          Returns the &#039;Referer&#039; (sic!) http header.
        </clix:description>
      </clix:function>

      <clix:function name='request-method*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>keyword
        </clix:returns>
        <clix:description>
          Returns the request method as a Lisp keyword.
        </clix:description>
      </clix:function>

      <clix:function name='request-uri*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>uri
        </clix:returns>
        <clix:description>
          Returns the request URI.
        </clix:description>
      </clix:function>

      <clix:function name='server-protocol*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>keyword
        </clix:returns>
        <clix:description>
          Returns the request protocol as a Lisp keyword.
        </clix:description>
      </clix:function>

      <clix:function name='user-agent'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>
          Returns the &#039;User-Agent&#039; http header.
        </clix:description>
      </clix:function>

      <clix:function name='header-in*'>
        <clix:lambda-list>name
        <clix:lkw>optional
        </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>header
        </clix:returns>
        <clix:description>
          Returns the incoming header with name
          <clix:arg>name</clix:arg>.  <clix:arg>name</clix:arg> can be
          a keyword (recommended) or a string.
        </clix:description>
      </clix:function>

      <clix:function name='headers-in*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>alist
        </clix:returns>
        <clix:description>
          Returns an alist of the incoming headers associated with the
          <clix:ref>REQUEST</clix:ref> object
          <clix:arg>request</clix:arg>.
        </clix:description>
      </clix:function>

      <clix:function name='remote-addr*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>address
        </clix:returns>
        <clix:description>
          Returns the address the current request originated from.
        </clix:description>
      </clix:function>

      <clix:function name='remote-port*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>port
        </clix:returns>
        <clix:description>
          Returns the port the current request originated from.
        </clix:description>
      </clix:function>

      <clix:function name='local-addr*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>address
        </clix:returns>
        <clix:description>
          The IP address of the local system that the client connected to.
        </clix:description>
      </clix:function>

      <clix:function name='local-port*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>port
        </clix:returns>
        <clix:description>
          The TCP port number of the local system that the client connected to.
        </clix:description>
      </clix:function>

      <clix:function name='script-name*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>script-name
        </clix:returns>
        <clix:description>
          Returns the file name of the <clix:ref>REQUEST</clix:ref>
          object <clix:arg>request</clix:arg>. That&#039;s the
          requested URI without the query string (i.e the GET
          parameters).
        </clix:description>
      </clix:function>

      <clix:accessor name='aux-request-value'>
        <clix:lambda-list>symbol
        <clix:lkw>optional
        </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>value, present-p
        </clix:returns>
        <clix:description>
          This accessor can be used to associate arbitrary
          data with the the symbol <clix:arg>symbol</clix:arg> in the <clix:ref>REQUEST</clix:ref> object
          <clix:arg>request</clix:arg>. <clix:arg>present-p</clix:arg> is true if such data was found, otherwise <code>NIL</code>.
        </clix:description>
      </clix:accessor>

      <clix:function name='delete-aux-request-value'>
        <clix:lambda-list>symbol
        <clix:lkw>optional
        </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>|
        </clix:returns>
        <clix:description>
          Removes the value associated with <clix:arg>symbol</clix:arg> from the <clix:ref>REQUEST</clix:ref> object
          <clix:arg>request</clix:arg>.
        </clix:description>
      </clix:function>

      <clix:function name='authorization'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> request
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>
          Returns as two values the user and password (if any) as
          encoded in the &#039;AUTHORIZATION&#039; header.  Returns
          <code>NIL</code> if there is no such header.
        </clix:description>
      </clix:function>

      <clix:special-variable name='*hunchentoot-default-external-format*'>
        <clix:description>
          The external format used to compute the <clix:ref>REQUEST</clix:ref> object.
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*file-upload-hook*'>
        <clix:description>
          If this is not <code>NIL</code>, it should be a unary
          function which will be called with a pathname for each file
          which is <a href="#upload">uploaded</a> to Hunchentoot.  The
          pathname denotes the temporary file to which the uploaded
          file is written.  The hook is called directly before the
          file is created. At this point,
          <clix:ref>*REQUEST*</clix:ref> is already bound to the
          current <clix:ref>REQUEST</clix:ref> object, but obviously
          you can't access the post parameters yet.
        </clix:description>
      </clix:special-variable>

      <clix:function name="raw-post-data">
        <clix:lambda-list>
          <clix:lkw>key</clix:lkw>
          request external-format force-text force-binary want-stream
        </clix:lambda-list>
        <clix:returns>raw-body-or-stream</clix:returns>
        <clix:description>
          Returns the content sent by the client in the request body if
          there was any (unless the content type
          was <code>multipart/form-data</code> in which
          case <code>NIL</code> is returned).  By default, the result is
          a string if the type of the <code>Content-Type</code>
          <a href="http://www.faqs.org/rfcs/rfc1590.html">media type</a>
          is <code>"text"</code>, and a vector of octets otherwise.  In
          the case of a string, the external format to be used to decode
          the content will be determined from the <code>charset</code>
          parameter sent by the client (or
          otherwise <clix:ref>*HUNCHENTOOT-DEFAULT-EXTERNAL-FORMAT*</clix:ref>
          will be used).
          <p>
            You can also provide an external format explicitly (through
            <clix:arg>external-format</clix:arg>) in which case the result
            will unconditionally be a string.  Likewise, you can provide
            a true value for <clix:arg>force-text</clix:arg> which will
            force Hunchentoot to act as if the type of the media type
            had been <code>"text"</code>
            (with <clix:arg>external-format</clix:arg> taking precedence
            if provided).  Or you can provide a true value
            for <clix:arg>force-binary</clix:arg> which means that you
            want a vector of octets at any rate.  (If both
            <clix:arg>force-text</clix:arg>
            and <clix:arg>force-binary</clix:arg> are true, an error will
            be signaled.)
          </p>
          <p>
            If, however, you provide a true value
            for <clix:arg>want-stream</clix:arg>, the other parameters are
            ignored and you'll get the content (flexi) stream to read
            from it yourself.  It is then your responsibility to read
            the correct amount of data, because otherwise you won't be
            able to return a response to the client.  The stream will
            have
            its <a href="http://weitz.de/flexi-streams/#flexi-streams">octet
              position</a> set to <code>0</code>.  If the client provided
            a <code>Content-Length</code> header, the stream will also
            have a
            corresponding <a href="http://weitz.de/flexi-streams/#flexi-streams">bound</a>,
            so no matter whether the client used chunked encoding or
            not, you can always read until EOF.
          </p>
          <p>
            If the content type of the request
            was <code>multipart/form-data</code>
            or <code>application/x-www-form-urlencoded</code>, the
            content has been read by Hunchentoot already and you can't
            read from the stream anymore.
          </p>
          <p>
            You can call <clix:ref>RAW-POST-DATA</clix:ref> more than once
            per request, but you can't mix calls which have different
            values for <clix:arg>want-stream</clix:arg>.
          </p>
          <p>
            Note that this function is slightly misnamed because a
            client can send content even if the request method is not
            POST.
          </p>
        </clix:description>
      </clix:function>


      <clix:function name='recompute-request-parameters'>
        <clix:lambda-list>
          <clix:lkw>key
          </clix:lkw> request external-format
        </clix:lambda-list>
        <clix:returns>|
        </clix:returns>
        <clix:description>
          Recomputes the GET and POST parameters for the <clix:ref>REQUEST</clix:ref> object
          <clix:arg>request</clix:arg>.  This only makes sense if you&#039;re switching external formats
          during the request.
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='process-request'>
        <clix:lambda-list>request
        </clix:lambda-list>
        <clix:returns>nil
        </clix:returns>
        <clix:description>
          This function is called by <clix:ref>PROCESS-CONNECTION</clix:ref>
          after the incoming headers have been read.  It
          calls <clix:ref>HANDLE-REQUEST</clix:ref> (and is more or less just a
          thin wrapper around it) to select and call a
          <a href="#request-dispatch">handler</a> and send the output of this handler to
          the client.  Note that <clix:ref>PROCESS-CONNECTION</clix:ref> is
          called once per connection and loops in case of a persistent
          connection while <clix:ref>PROCESS-REQUEST</clix:ref> is called anew
          for each request.
          <p>
            The return value of this function is ignored.
          </p>
          <p>
            Like <clix:ref>PROCESS-CONNECTION</clix:ref>, this is another function
            the behaviour of which you should only modify if you really, really
            know what you're doing.
          </p>
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='handle-request'>
        <clix:lambda-list>acceptor request
        </clix:lambda-list>
        <clix:returns>content
        </clix:returns>
        <clix:description>
          This function is called by <clix:ref>PROCESS-REQUEST</clix:ref> once
          the request has been read and a <clix:ref>REQUEST</clix:ref> object
          has been created.  Its job is to actually handle the request, i.e. to
          return something to the client.
          <p>
            The default method calls the
            acceptor's <a href="#request-dispatch">request dispatcher</a>, but you
            can of course implement a different behaviour.  The default method
            also sets up <a href="#logging">standard error handling</a> for
            the <a href="#request-dispatch">handler</a>.
          </p>
          <p>
            Might be a good place to bind or rebind special variables which can
            then be accessed by your <a href="#request-dispatch">handlers</a>.
          </p>
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='acceptor-dispatch-request'>
        <clix:lambda-list>acceptor request
        </clix:lambda-list>
        <clix:returns>content
        </clix:returns>
        <clix:description>
          This function is called to actually dispatch the request
          once the standard logging and error handling has been set
          up.  <clix:ref>ACCEPTOR</clix:ref> subclasses implement
          methods for this function in order to perform their own
          request routing.  If a method does not want to handle the
          request, it is supposed to invoke <a
          href="http://www.lispworks.com/documentation/HyperSpec/Body/f_call_n.htm">CALL-NEXT-METHOD</a>
          so that the next <clix:ref>ACCEPTOR</clix:ref> in the
          inheritance chain gets a chance to handle the request.
        </clix:description>
      </clix:function>

      <clix:readers generic='true'>
        <clix:listed-reader generic='true' name='cookies-in'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>cookies
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='get-parameters'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>get-parameters
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='header-in'>
          <clix:lambda-list>name request
          </clix:lambda-list>
          <clix:returns>result
          </clix:returns>
          <clix:description>
          </clix:description>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='headers-in'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>headers
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='post-parameters'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>post-parameters
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='query-string'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>query-string
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='remote-addr'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>address
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='remote-port'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>port
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='local-addr'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>address
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='local-port'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>port
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='request-acceptor'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>acceptor
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='request-method'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>method
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='request-uri'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>uri
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='server-protocol'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>protocol
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='script-name'>
          <clix:lambda-list>request
          </clix:lambda-list>
          <clix:returns>result
          </clix:returns>
        </clix:listed-reader>

        <clix:description>
          These are various generic readers which are used
          to read information about a <clix:ref>REQUEST</clix:ref> object.  If you are writing a
          <a href="#request-dispatch">handler</a>, you should <em>not</em> use these readers but instead utilize the
          corresponding functions with an asterisk at the end of their name,
          also listed in this section.  These generic readers are only
          exported for users who want to create their own subclasses of
          <clix:ref>REQUEST</clix:ref>.
        </clix:description>

      </clix:readers>

    </clix:subchapter>

    <clix:subchapter name="replies" title="Reply objects">

      For each incoming request, the <a href="#acceptors">acceptor</a>
      (in <clix:ref>PROCESS-CONNECTION</clix:ref>) creates
      a <clix:ref>REPLY</clix:ref> object and makes it available
      to <a href="#request-dispatch">handlers</a> via the special variable
      <clix:ref>*REPLY*</clix:ref>.  This object contains all relevant
      information (except for the content body) about the reply that will be
      sent to the client and this section collects the functions which can
      be used to query and modify such an object.  In all function
      where <clix:arg>reply</clix:arg> is an optional or keyword parameter,
      the default is <clix:ref>*REPLY*</clix:ref>.

      <p>
        If you need more fine-grained control over the behaviour of reply
        objects, you can subclass <clix:ref>REPLY</clix:ref> and initialize
        the <a href="#acceptor-reply-class"><code>REPLY-CLASS</code></a>
        slot of the <clix:ref>ACCEPTOR</clix:ref> class accordingly.  The
        acceptor will generate reply objects of the class named by this
        slot.
      </p>

      <clix:class name='reply'>
        <clix:description>
          Objects of this class hold all the information about an
          outgoing reply.  They are created automatically by
          Hunchentoot and can be accessed and modified by the
          corresponding <a href="#request-dispatch">handler</a>.
        <p>
          You should not mess with the slots of these objects directly, but you
          can subclass <clix:ref>REPLY</clix:ref> in order to implement your own behaviour.  See the
          <a href="#acceptor-reply-class"><code>:reply-class</code></a> initarg
          of the <clix:ref>ACCEPTOR</clix:ref> class.
        </p>
        </clix:description>
      </clix:class>

      <clix:special-variable name='*reply*'>
        <clix:description>
          The current <clix:ref>REPLY</clix:ref> object in the context of a request.
        </clix:description>
      </clix:special-variable>

      <clix:accessor name='header-out'>
        <clix:lambda-list>name
        <clix:lkw>optional
        </clix:lkw> reply
        </clix:lambda-list>
        <clix:returns>string
        </clix:returns>
        <clix:description>
          <clix:ref>HEADER-OUT</clix:ref> returns the outgoing http
          header named by the keyword <clix:arg>name</clix:arg> if
          there is one, otherwise <code>NIL</code>. <code>SETF</code>
          of <clix:ref>HEADER-OUT</clix:ref> changes the current value
          of the header named <clix:arg>name</clix:arg>. If no header
          named <clix:arg>name</clix:arg> exists, it is created. For
          backwards compatibility, <clix:arg>name</clix:arg> can also
          be a string in which case the association between a header
          and its name is case-insensitive.
          <p>
            Note that the header 'Set-Cookie' cannot be queried by
            <clix:ref>HEADER-OUT</clix:ref> and must not be set by
            <code>SETF</code> of <clix:ref>HEADER-OUT</clix:ref>.  See
            also <clix:ref>HEADERS-OUT*</clix:ref>,
            <clix:ref>CONTENT-TYPE*</clix:ref>,
            <clix:ref>CONTENT-LENGTH*</clix:ref>,
            <clix:ref>COOKIES-OUT*</clix:ref>, and
            <clix:ref>COOKIE-OUT</clix:ref>.
          </p>
        </clix:description>
      </clix:accessor>

      <clix:function name='headers-out*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> reply
        </clix:lambda-list>
        <clix:returns>alist
        </clix:returns>
        <clix:description>Returns an alist of the outgoing headers associated with the
        <clix:ref>REPLY</clix:ref> object <clix:arg>reply</clix:arg>.  See also <clix:ref>HEADER-OUT</clix:ref>.
        </clix:description>
      </clix:function>

      <clix:accessor name='content-length*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> reply
        </clix:lambda-list>
        <clix:returns>content-length
        </clix:returns>
        <clix:description>
          The outgoing &#039;Content-Length&#039; http header of <clix:arg>reply</clix:arg>.
        </clix:description>
      </clix:accessor>

      <clix:accessor name='content-type*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> reply
        </clix:lambda-list>
        <clix:returns>content-type
        </clix:returns>
        <clix:description>
          The outgoing &#039;Content-Type&#039; http header of <clix:arg>reply</clix:arg>.
        </clix:description>
      </clix:accessor>

      <clix:function name='cookie-out'>
        <clix:lambda-list>name
        <clix:lkw>optional
        </clix:lkw> reply
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>
          Returns the current value of the outgoing <a
          href="#cookies">cookie</a> named
          <clix:arg>name</clix:arg>. Search is case-sensitive.
        </clix:description>
      </clix:function>

      <clix:accessor name='cookies-out*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> reply
        </clix:lambda-list>
        <clix:returns>alist
        </clix:returns>
        <clix:description>
          Returns or sets an alist of the outgoing <a
          href="#cookies">cookies</a> associated with the
          <clix:ref>REPLY</clix:ref> object
          <clix:arg>reply</clix:arg>.
        </clix:description>
      </clix:accessor>

      <clix:accessor name='return-code*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> reply
        </clix:lambda-list>
        <clix:returns>return-code
        </clix:returns>
        <clix:description>
          Gets or sets the http return code of
          <clix:arg>reply</clix:arg>.  The return code of each
          <clix:ref>REPLY</clix:ref> object is initially set to
          <clix:ref>+HTTP-OK+</clix:ref>.
        </clix:description>
      </clix:accessor>

      <clix:function name="send-headers">
        <clix:returns>stream</clix:returns>
        <clix:description>
          Sends the initial status line and all headers as determined
          by the <clix:ref>REPLY</clix:ref>
          object <clix:ref>*REPLY*</clix:ref>.  Returns
          a <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_b.htm#binary">binary</a>
          stream to which the body of the reply can be written.  Once
          this function has been called, further changes
          to <clix:ref>*REPLY*</clix:ref> don't have any effect.
          Also, automatic handling of errors (i.e. sending the
          corresponding status code to the browser, etc.) is turned
          off for this request and functions
          like <clix:ref>REDIRECT</clix:ref> or
          to <clix:ref>ABORT-REQUEST-HANDLER</clix:ref> won't have the
          desired effect once the headers are sent.
          <p>
            If your handlers return the full body as a string or as an
            array of octets, you should <em>not</em> call this function.
            If a handler calls <clix:ref>SEND-HEADERS</clix:ref> , its
            return value is ignored.
          </p>
        </clix:description>
      </clix:function>

      <clix:accessor name='reply-external-format*'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> reply
        </clix:lambda-list>
        <clix:returns>external-format
        </clix:returns>
        <clix:description>
          Gets or sets the external format of <clix:arg>reply</clix:arg> which is used for character output.
        </clix:description>
      </clix:accessor>

      <clix:special-variable name='*default-content-type*'>
        <clix:description>
          The default content-type header which is returned to the client.
        </clix:description>
      </clix:special-variable>

      <clix:constants>
        <clix:listed-constant name="+http-continue+"/>
        <clix:listed-constant name="+http-switching-protocols+"/>
        <clix:listed-constant name="+http-ok+"/>
        <clix:listed-constant name="+http-created+"/>
        <clix:listed-constant name="+http-accepted+"/>
        <clix:listed-constant name="+http-non-authoritative-information+"/>
        <clix:listed-constant name="+http-no-content+"/>
        <clix:listed-constant name="+http-reset-content+"/>
        <clix:listed-constant name="+http-partial-content+"/>
        <clix:listed-constant name="+http-multi-status+"/>
        <clix:listed-constant name="+http-multiple-choices+"/>
        <clix:listed-constant name="+http-moved-permanently+"/>
        <clix:listed-constant name="+http-moved-temporarily+"/>
        <clix:listed-constant name="+http-see-other+"/>
        <clix:listed-constant name="+http-not-modified+"/>
        <clix:listed-constant name="+http-use-proxy+"/>
        <clix:listed-constant name="+http-temporary-redirect+"/>
        <clix:listed-constant name="+http-bad-request+"/>
        <clix:listed-constant name="+http-authorization-required+"/>
        <clix:listed-constant name="+http-payment-required+"/>
        <clix:listed-constant name="+http-forbidden+"/>
        <clix:listed-constant name="+http-not-found+"/>
        <clix:listed-constant name="+http-method-not-allowed+"/>
        <clix:listed-constant name="+http-not-acceptable+"/>
        <clix:listed-constant name="+http-proxy-authentication-required+"/>
        <clix:listed-constant name="+http-request-time-out+"/>
        <clix:listed-constant name="+http-conflict+"/>
        <clix:listed-constant name="+http-gone+"/>
        <clix:listed-constant name="+http-length-required+"/>
        <clix:listed-constant name="+http-precondition-failed+"/>
        <clix:listed-constant name="+http-request-entity-too-large+"/>
        <clix:listed-constant name="+http-request-uri-too-large+"/>
        <clix:listed-constant name="+http-unsupported-media-type+"/>
        <clix:listed-constant name="+http-requested-range-not-satisfiable+"/>
        <clix:listed-constant name="+http-expectation-failed+"/>
        <clix:listed-constant name="+http-failed-dependency+"/>
        <clix:listed-constant name="+http-internal-server-error+"/>
        <clix:listed-constant name="+http-not-implemented+"/>
        <clix:listed-constant name="+http-bad-gateway+"/>
        <clix:listed-constant name="+http-service-unavailable+"/>
        <clix:listed-constant name="+http-gateway-time-out+"/>
        <clix:listed-constant name="+http-version-not-supported+"/>
        <clix:description>
          The values of these constants are 100, 101, 200, 201, 202,
          203, 204, 205, 206, 207, 300, 301, 302, 303, 304, 305, 307,
          400, 401, 402, 403, 404, 405, 406, 407, 408, 409, 410, 411,
          412, 413, 414, 415, 416, 417, 424, 500, 501, 502, 503, 504,
          and 505. See <clix:ref>RETURN-CODE</clix:ref>.
        </clix:description>
      </clix:constants>

      <clix:readers generic='true'>
        <clix:listed-reader generic='true' name='content-length'>
          <clix:lambda-list>reply
          </clix:lambda-list>
          <clix:returns>content-length
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='content-type'>
          <clix:lambda-list>reply
          </clix:lambda-list>
          <clix:returns>content-type
          </clix:returns>
        </clix:listed-reader>

        <clix:listed-reader generic='true' name='headers-out'>
          <clix:lambda-list>reply
          </clix:lambda-list>
          <clix:returns>headers-out
          </clix:returns>
        </clix:listed-reader>

        <clix:description>
          These are various generic readers which are used
          to read information about a <clix:ref>REPLY</clix:ref> object.  If you are writing a
          <a href="#request-dispatch">handler</a>, you should <em>not</em> use these readers but instead utilize the
          corresponding functions with an asterisk at the end of their name,
          also listed in this section.  These generic readers are only
          exported for users who want to create their own subclasses of
          <clix:ref>REPLY</clix:ref>.
        </clix:description>
      </clix:readers>

      <clix:accessors generic='true'>
        <clix:listed-accessor generic='true' name='cookies-out'>
          <clix:lambda-list>reply
          </clix:lambda-list>
          <clix:returns>result
          </clix:returns>
        </clix:listed-accessor>

        <clix:listed-accessor generic='true' name='return-code'>
          <clix:lambda-list>reply
          </clix:lambda-list>
          <clix:returns>result
          </clix:returns>
        </clix:listed-accessor>

        <clix:listed-accessor generic='true' name='reply-external-format'>
          <clix:lambda-list>reply
          </clix:lambda-list>
          <clix:returns>result
          </clix:returns>
        </clix:listed-accessor>

        <clix:description>
          These are various generic accessors which are
          used to query and modify a <clix:ref>REPLY</clix:ref> objects.  If
          you are writing a
          <a href="#request-dispatch">handler</a>, you should <em>not</em> use these
          accessors but instead utilize the corresponding functions with an
          asterisk at the end of their name, also listed in this section.
          These generic accessors are only exported for users who want to
          create their own subclasses of
          <clix:ref>REPLY</clix:ref>.
        </clix:description>
      </clix:accessors>


    </clix:subchapter>

    <clix:subchapter name="sessions" title="Sessions">
      Hunchentoot supports <em>sessions</em>: Once a <a href="#request-dispatch">request
      handler</a> has called <clix:ref>START-SESSION</clix:ref>, Hunchentoot
      uses either cookies or (if the client doesn't send the cookies
      back) <a href="#*rewrite-for-session-urls*">rewrites URLs</a> to keep
      track of this client, i.e. to provide a kind of 'state' for the
      stateless http protocol. The session associated with the client is a
      <a href="#session">CLOS object</a> which can be used
      to <a href="#session-value">store arbitrary data</a> between requests.
      <p>
        Hunchentoot makes some reasonable effort to prevent eavesdroppers from
        hijacking sessions (see below), but this should not be considered
        really secure. Don't store sensitive data in sessions and rely solely
        on the session mechanism as a safeguard against malicious users who
        want to get at this data!
      </p>
      <p>
        For each request there's one <clix:ref>SESSION</clix:ref> object which is accessible to the
        <a href="#handler">handler</a> via the special
        variable <clix:ref>*SESSION*</clix:ref>. This object holds all the
        information available about the session and can be accessed with the
        functions described in this chapter. Note that the internal structure
        of <clix:ref>SESSION</clix:ref> objects should be considered opaque
        and may change in future releases of Hunchentoot.
      </p>
      <p>
        Sessions are automatically <a href="#session-verify">verified</a> for
        validity and age when the <clix:ref>REQUEST</clix:ref> object is
        instantiated, i.e. if <clix:ref>*SESSION*</clix:ref> is not NIL then
        this session is valid (as far as Hunchentoot is concerned) and
        not <a href="#session-too-old-p">too old</a>.  Old sessions
        are <a href="#session-gc">automatically removed</a>.
      </p>
      <p>
        Hunchentoot also provides a <clix:ref>SESSION-REGENERATE-COOKIE-VALUE</clix:ref>
        function that creates a new cookie value. This helps to prevent against
        <a href="https://www.owasp.org/index.php/Session_fixation">session fixation
        attacks</a>, and should be used when a user logs in according to the application.
      </p>

      <clix:class name='session'>
        <clix:description>
          <clix:ref>SESSION</clix:ref> objects are
          automatically maintained by Hunchentoot.  They should not be created
          explicitly with <code>MAKE-INSTANCE</code> but implicitly
          with <clix:ref>START-SESSION</clix:ref> and they should be treated as
          opaque objects.
          <p>
            You can ignore Hunchentoot's <clix:ref>SESSION</clix:ref> objects and
            <a href="#session-behaviour">implement your own sessions</a> if you provide corresponding methods for
            <clix:ref>SESSION-COOKIE-VALUE</clix:ref>
            and <clix:ref>SESSION-VERIFY</clix:ref>.
          </p>
        </clix:description>
      </clix:class>

      <clix:function name='start-session'>
        <clix:lambda-list>
        </clix:lambda-list>
        <clix:returns>session
        </clix:returns>
        <clix:description>
          Returns the current <clix:ref>SESSION</clix:ref>
          object. If there is no current session, creates one and updates the
          corresponding data structures. In this case the function will also
          send a session cookie to the browser.
        </clix:description>
      </clix:function>

      <clix:accessor name='session-value'>
        <clix:lambda-list>symbol
        <clix:lkw>optional
        </clix:lkw> session
        </clix:lambda-list>
        <clix:returns>value, present-p
        </clix:returns>
        <clix:description>
          This accessor can be used to associate arbitrary data with the the
          symbol <clix:arg>symbol</clix:arg> in the <clix:ref>SESSION</clix:ref>
          object <clix:arg>session</clix:arg>. <clix:arg>present-p</clix:arg> is
          true if such data was found, otherwise <code>NIL</code>. The default
          value for <clix:arg>session</clix:arg> is
          <clix:ref>*SESSION*</clix:ref>.
          <p>
            If <code>SETF</code> of <clix:ref>SESSION-VALUE</clix:ref> is called
            with <clix:arg>session</clix:arg> being <code>NIL</code> then a
            session is automatically instantiated
            with <clix:ref>START-SESSION</clix:ref>.
          </p>
        </clix:description>
      </clix:accessor>

      <clix:function name='delete-session-value'>
        <clix:lambda-list>symbol
        <clix:lkw>optional
        </clix:lkw> session
        </clix:lambda-list>
        <clix:returns>|
        </clix:returns>
        <clix:description>
          Removes the value associated with
          <clix:arg>symbol</clix:arg> from
          <clix:arg>session</clix:arg> if there is one.
        </clix:description>
      </clix:function>

      <clix:special-variable name='*session*'>
        <clix:description>
          The current session while in the context of a request, or
          <code>NIL</code>.
        </clix:description>
      </clix:special-variable>

      <clix:function name='remove-session'>
        <clix:lambda-list>session
        </clix:lambda-list>
        <clix:returns>|
        </clix:returns>
        <clix:description>
          Completely removes the <clix:ref>SESSION</clix:ref> object
          <clix:arg>session</clix:arg> from Hunchentoot&#039;s
          internal <a href="#session-db">session database</a>.
        </clix:description>
      </clix:function>

      <clix:function name='reset-sessions'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> acceptor
        </clix:lambda-list>
        <clix:returns>|
        </clix:returns>
        <clix:description>
          Removes <em>all</em> stored sessions of
          <clix:arg>acceptor</clix:arg>.  The default for
          <clix:arg>acceptor</clix:arg> is
          <clix:ref>*ACCEPTOR*</clix:ref>.
        </clix:description>
      </clix:function>

      <clix:function name='regenerate-session-cookie-value'>
        <clix:lambda-list>session
        </clix:lambda-list>
        <clix:returns>cookie
        </clix:returns>
        <clix:description>
          Regenerates the session cookie value. This should be used
          when a user logs in according to the application to prevent
          against session fixation attacks. The cookie value being
          dependent on ID, USER-AGENT, REMOTE-ADDR, START, and
          *SESSION-SECRET*, the only value we can change is START to
          regenerate a new value. Since we're generating a new cookie,
          it makes sense to have the session being restarted, in
          time. That said, because of this fact, calling this function
          twice in the same second will regenerate twice the same
          value.
        </clix:description>
      </clix:function>

      <clix:special-variable name='*rewrite-for-session-urls*'>
        <clix:description>
          Whether HTML pages should possibly be rewritten for cookie-less
          session-management.
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*content-types-for-url-rewrite*'>
        <clix:description>
          The content types for which url-rewriting is OK. See
          <clix:ref>*REWRITE-FOR-SESSION-URLS*</clix:ref>.
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*use-remote-addr-for-sessions*'>
        <clix:description>
          Whether the client&#039;s remote IP (as returned by <clix:ref>REAL-REMOTE-ADDR</clix:ref>)
          should be encoded into the session string.  If this value is true, a
          session will cease to be accessible if the client&#039;s remote IP changes.
          <p>
            This might for example be an issue if the client uses a proxy server
            which doesn&#039;t send correct &#039;X-Forwarded-For&#039; headers.
          </p>
        </clix:description>
      </clix:special-variable>

      <clix:function generic='true' name='session-remote-addr'>
        <clix:lambda-list>session
        </clix:lambda-list>
        <clix:returns>remote-addr
        </clix:returns>
        <clix:description>
          The remote IP address of the client when this session was started (as
          returned by <clix:ref>REAL-REMOTE-ADDR</clix:ref>).
        </clix:description>
      </clix:function>

      <clix:special-variable name='*use-user-agent-for-sessions*'>
        <clix:description>Whether the &#039;User-Agent&#039; header should
        be encoded into the session string.  If this value is true, a session
        will cease to be accessible if the client sends a different
        &#039;User-Agent&#039; header.
        </clix:description>
      </clix:special-variable>

      <clix:function generic='true' name='session-user-agent'>
        <clix:lambda-list>session
        </clix:lambda-list>
        <clix:returns>user-agent
        </clix:returns>
        <clix:description>
          The incoming &#039;User-Agent&#039; header that
          was sent when this session was created.
        </clix:description>
      </clix:function>

      <clix:accessor generic='true' name='session-max-time'>
        <clix:lambda-list>session
        </clix:lambda-list>
        <clix:returns>max-time
        </clix:returns>
        <clix:description>
          Gets or sets the time (in seconds) after
          which <clix:arg>session</clix:arg> expires if it's not used.
        </clix:description>
      </clix:accessor>


      <clix:special-variable name='*session-max-time*'>
        <clix:description>
          The default time (in seconds) after which a session times out.
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*session-gc-frequency*'>
        <clix:description>
          A session GC (see function <clix:ref>SESSION-GC</clix:ref>) will happen every
          <clix:ref>*SESSION-GC-FREQUENCY*</clix:ref> requests (counting only
          requests which create a new session) if this variable is
          not <code>NIL</code>.  See <clix:ref>SESSION-CREATED</clix:ref>.
        </clix:description>
      </clix:special-variable>

      <clix:function name='session-gc'>
        <clix:lambda-list>
        </clix:lambda-list>
        <clix:returns>|
        </clix:returns>
        <clix:description>
          Removes sessions from the current session database which are
          too old - see <clix:ref>SESSION-TOO-OLD-P</clix:ref>.
        </clix:description>
      </clix:function>

      <clix:function name='session-too-old-p'>
        <clix:lambda-list>session
        </clix:lambda-list>
        <clix:returns>generalized-boolean
        </clix:returns>
        <clix:description>
          Returns true if the <clix:ref>SESSION</clix:ref> object <clix:arg>session</clix:arg> has not been active in
          the last <code>(session-max-timesession)</code> seconds.
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='session-id'>
        <clix:lambda-list>session
        </clix:lambda-list>
        <clix:returns>session-id
        </clix:returns>
        <clix:description>
          The unique ID (an INTEGER) of the session.
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='session-start'>
        <clix:lambda-list>session
        </clix:lambda-list>
        <clix:returns>universal-time
        </clix:returns>
        <clix:description>
          The time this session was started.
        </clix:description>
      </clix:function>

    </clix:subchapter>

    <clix:subchapter name="session-behaviour" title="Customizing session behaviour">

      For everyday session usage, you will probably just
      use <clix:ref>START-SESSION</clix:ref>,
      <clix:ref>SESSION-VALUE</clix:ref>,
      and maybe <clix:ref>DELETE-SESSION-VALUE</clix:ref>
      and <clix:ref>*SESSION*</clix:ref>.  However, there are two ways to
      customize the way Hunchentoot maintains sessions.
      <p>
        One way is to mostly leave the session mechanism intact but to tweak
        it a bit:
        <ul>
          <li>The publicly visible part of a session is encoded using a
          <a href="#*session-secret*">secret</a> which you can set yourself.</li>
          <li>And it is stored using a cookie (or GET
          parameter) <a href="#session-cookie-name">name</a> that you can
          override.</li>
          <li>Each session receives a <a href="#next-session-id">new ID</a> when
          it is created and you can implement a more robust way to do that.</li>
          <li>You can arrange to be called whenever a session
          is <a href="#session-created">created</a> to trigger some action.  You
          might also do this to invent your own
          session <a href="#session-gc">garbage collection</a>.</li>
          <li>By default, all sessions are stored in a global alist in memory.
          You can't change the alist part, but you can distribute your sessions
          over different <a href="#session-db">"databases"</a>.</li>
          <li>By default, every operation which modifies sessions or one of the
          session databases is guarded by a global lock, but you can arrange to
          <a href="#session-db-lock">provide</a> different locks for this.</li>
        </ul>
      </p>
      <p>
        The other way to customize Hunchentoot's sessions is to completely
        replace them.  This is actually pretty easy: Create your own class to
        store state (which doesn't have to and probably shouldn't inherit
        from <clix:ref>SESSION</clix:ref>) and implement methods for
        <clix:ref>SESSION-VERIFY</clix:ref>
        and <clix:ref>SESSION-COOKIE-VALUE</clix:ref> - that's it.
        Hunchentoot will continue to use cookies and/or to rewrite URLs to
        keep track of session state and it will store "the current session"
        (whatever that is in your implementation)
        in <clix:ref>*SESSION*</clix:ref>.  Everything else (like persisting
        sessions, GC, getting and setting values) you'll have to take care of
        yourself and the other session functions
        (like <clix:ref>START-SESSION</clix:ref> or
        <clix:ref>SESSION-VALUE</clix:ref>) won't work anymore.  (Almost)
        total freedom, but a lot of responsibility as well... :)
      </p>

      <clix:special-variable name='*session-secret*'>
        <clix:description>
          A random ASCII string that&#039;s used to encode the public
          session data.  This variable is initially unbound and will
          be set (using <clix:ref>RESET-SESSION-SECRET</clix:ref>) the
          first time a session is created, if necessary.  You can
          prevent this from happening if you set the value yourself
          before starting <a href="#acceptors">acceptors</a>.
        </clix:description>
      </clix:special-variable>

      <clix:function name='reset-session-secret'>
        <clix:lambda-list>
        </clix:lambda-list>
        <clix:returns>secret
        </clix:returns>
        <clix:description>
          Sets <clix:ref>*SESSION-SECRET*</clix:ref> to a
          new random value. All old sessions will cease to be valid.
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='session-cookie-name'>
        <clix:lambda-list>acceptor
        </clix:lambda-list>
        <clix:returns>name
        </clix:returns>
        <clix:description>
          Returns the name (a string) of the cookie (or
          the GET parameter) which is used to store a session on the client
          side.  The default is to use the
          string <code>&quot;hunchentoot-session&quot;</code>, but you can
          specialize this function if you want another name.
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='session-created'>
        <clix:lambda-list>acceptor new-session
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>
          This function is called whenever a new session
          has been created.  There&#039;s a default method which might trigger
          a <a href="#session-gc">session GC</a> based on the value of
          <clix:ref>*SESSION-GC-FREQUENCY*</clix:ref>.
          <p>
            The return value is ignored.
          </p>
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='next-session-id'>
        <clix:lambda-list>acceptor
        </clix:lambda-list>
        <clix:returns>id
        </clix:returns>
        <clix:description>
          Returns the next sequential session ID, an
          integer, which should be unique per session.  The default method uses
          a simple global counter and isn&#039;t guarded by a lock.  For a
          high-performance production environment you might consider using a
          more robust implementation.
        </clix:description>
      </clix:function>

      <clix:accessor generic='true' name='session-db'>
        <clix:lambda-list>acceptor
        </clix:lambda-list>
        <clix:returns>database
        </clix:returns>
        <clix:description>
          Returns the current session database which is an
          alist where each car is a session&#039;s ID and the cdr is the
          corresponding <clix:ref>SESSION</clix:ref> object itself.  The default
          is to use a global list for all acceptors.
        </clix:description>
      </clix:accessor>

      <clix:function generic='true' name='session-db-lock'>
        <clix:lambda-list>acceptor
        <clix:lkw>key
        </clix:lkw> whole-db-p
        </clix:lambda-list>
        <clix:returns>lock
        </clix:returns>
        <clix:description>
          A function which returns a lock that will be
          used to prevent concurrent access to sessions.  The first argument
          will be the <a href="#acceptors">acceptor</a> that handles the
          current <a href="#requests">request</a>, the second argument is true
          if the whole (current) session database is modified.  If it
          is <code>NIL</code>, only one existing session in the database is
          modified.
          <p>
            This function can return <code>NIL</code> which means that sessions or
            session databases will be modified without a lock held (for example
            for single-threaded environments).  The default is to always return a
            global lock (ignoring the <clix:arg>acceptor</clix:arg> argument) for
            Lisps that support threads and <code>NIL</code> otherwise.
          </p>
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='session-verify'>
        <clix:lambda-list>request
        </clix:lambda-list>
        <clix:returns>session-or-nil
        </clix:returns>
        <clix:description>
          Tries to get a session identifier from the cookies
          (or alternatively from the GET parameters) sent by the client (see
          <clix:ref>SESSION-COOKIE-NAME</clix:ref>
          and <clix:ref>SESSION-COOKIE-VALUE</clix:ref>).  This identifier is
          then checked for validity against the <clix:ref>REQUEST</clix:ref>
          object
          <clix:arg>request</clix:arg>.  On success the corresponding session object (if not too
          old) is returned (and updated).  Otherwise <code>NIL</code> is returned.
          <p>
            A default method is provided and you only need to write your own one
            if you want to maintain your own sessions.
          </p>
        </clix:description>
      </clix:function>

      <clix:function generic='true' name='session-cookie-value'>
        <clix:lambda-list>session
        </clix:lambda-list>
        <clix:returns>string
        </clix:returns>
        <clix:description>
          Returns a string which can be used to safely
          restore the session <clix:arg>session</clix:arg> if as session has
          already been established.  This is used as the value stored in the
          session cookie or in the corresponding GET parameter and verified
          by <clix:ref>SESSION-VERIFY</clix:ref>.
          <p>
            A default
            method is provided and there&#039;s no reason to change it unless you
            want to use your own session objects.
          </p>
        </clix:description>
      </clix:function>

    </clix:subchapter>

    <clix:subchapter name="cookies" title="Cookies">

      Outgoing cookies are stored in the request's <clix:ref>REPLY</clix:ref>
      object (see <clix:ref>COOKIE-OUT</clix:ref>
      and <clix:ref>COOKIES-OUT*</clix:ref>). They are CLOS objects
      defined like this:

      <pre>(defclass cookie ()
  ((name :initarg :name
         :reader <a class="noborder" name="cookie-name">cookie-name</a>
         :type string
         :documentation "The name of the cookie - a string.")
   (value :initarg :value
          :accessor <a class="noborder" name="cookie-value">cookie-value</a>
          :initform ""
          :documentation "The value of the cookie. Will be URL-encoded when sent to the browser.")
   (expires :initarg :expires
            :initform nil
            :accessor <a class="noborder" name="cookie-expires">cookie-expires</a>
            :documentation "The time (a universal time) when the cookie expires (or NIL).")
   (max-age :initarg :max-age
            :initform nil
            :accessor <a class="noborder" name="cookie-max-age">cookie-max-age</a>
            :documentation "The time delta (in seconds) after which the cookie expires (or NIL).")
   (path :initarg :path
         :initform nil
         :accessor <a class="noborder" name="cookie-path">cookie-path</a>
         :documentation "The path this cookie is valid for (or NIL).")
   (domain :initarg :domain
           :initform nil
           :accessor <a class="noborder" name="cookie-domain">cookie-domain</a>
           :documentation "The domain this cookie is valid for (or NIL).")
   (secure :initarg :secure
           :initform nil
           :accessor <a class="noborder" name="cookie-secure">cookie-secure</a>
           :documentation "A generalized boolean denoting whether this is a secure cookie.")
   (http-only :initarg :http-only
              :initform nil
              :accessor <a class="noborder" name="cookie-http-only">cookie-http-only</a>
              :documentation "A generalized boolean denoting whether this is a <a href="http://msdn2.microsoft.com/en-us/library/ms533046.aspx">HttpOnly</a> cookie.")))
      </pre>

      The <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_r.htm#reader">reader</a>
      <clix:ref>COOKIE-NAME</clix:ref> and
      the <a href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_a.htm#accessor">accessors</a>
      <clix:ref>COOKIE-VALUE</clix:ref>, <clix:ref>COOKIE-EXPIRES</clix:ref>, <clix:ref>COOKIE-MAX-AGE</clix:ref>,
      <clix:ref>COOKIE-PATH</clix:ref>, <clix:ref>COOKIE-DOMAIN</clix:ref>, <clix:ref>COOKIE-SECURE</clix:ref>,
      and <clix:ref>COOKIE-HTTP-ONLY</clix:ref> are all exported from
      the <code>HUNCHENTOOT</code> package.  For now, the class name itself is <em>not</em> exported.

      <clix:function name="set-cookie">
        <clix:lambda-list>
          name <clix:lkw>key</clix:lkw> value expires path
          domain secure http-only reply
        </clix:lambda-list>
        <clix:returns>cookie</clix:returns>
        <clix:description>
          Creates a <code>COOKIE</code> object from the parameters
          provided to this function and adds it to the outgoing cookies
          of the <a href="#replies"><code>REPLY</code> object</a>
          <clix:arg>reply</clix:arg>. If a cookie with the same name
          (case-sensitive) already exists, it is replaced. The default
          for <clix:arg>reply</clix:arg>
          is <clix:ref>*REPLY*</clix:ref>. The default
          for <clix:arg>value</clix:arg> is the empty string.
        </clix:description>
      </clix:function>

      <clix:function name="set-cookie*">
        <clix:lambda-list>cookie <clix:lkw>optional</clix:lkw> reply</clix:lambda-list>
        <clix:returns>cookie</clix:returns>
        <clix:description>
          Adds the <code>COOKIE</code> object <clix:arg>cookie</clix:arg>
          to the outgoing cookies of
          the <a href="#replies"><code>REPLY</code> object</a>
          <clix:arg>reply</clix:arg>. If a cookie with the same name
          (case-sensitive) already exists, it is replaced. The default
          for <clix:arg>reply</clix:arg> is <clix:ref>*REPLY*</clix:ref>.
        </clix:description>
      </clix:function>
    </clix:subchapter>

    <clix:subchapter name="logging" title="Logging">
      Hunchentoot can log accesses and diagnostic messages to two
      separate destinations, which can be either files in the file
      system or streams.  Logging can also be disabled by setting the
      <clix:code>ACCESS-LOG-DESTINATION</clix:code> and
      <clix:code>MESSAGE-LOG-DESTINATION</clix:code> slots in the
      <clix:ref>ACCEPTOR</clix:ref> to <code>NIL</code>.  The two
      slots can be initialized by providing the
      :ACCESS-LOG-DESTINATION and :MESSAGE-LOG-DESTINATION
      initialization arguments when creating the acceptor or set by
      setting the slots through its
      <clix:ref>ACCEPTOR-MESSAGE-LOG-DESTINATION</clix:ref> and
      <clix:ref>ACCEPTOR-ACCESS-LOG-DESTINATION</clix:ref> accessors.
      <p>
        When the path for the message or accept log is set to a
        variable holding an output stream, hunchentoots writes
        corresponding log entries to that stream.  By default,
        Hunchentoot logs to *STANDARD-ERROR*.
      </p>
      <p>
        Access logging is done in a format similar to what
        the Apache web server can write so that logfile analysis using
        standard tools is possible.  Errors during request processing are
        logged to a separate file.
      </p>
      <p>
        The standard logging mechanism is deliberately simple and slow.  The
        log files are opened for each log entry and closed again after
        writing, and access to them is protected by a global lock.  Derived
        acceptor classes can implement methods for the
        <clix:ref>ACCEPTOR-LOG-MESSAGE</clix:ref> and
        <clix:ref>ACCEPTOR-LOG-ACCESS</clix:ref> generic functions in order to
        log differently (e.g. to a central logging server or in a different
        file format.
      </p>
      <p>
        Errors happening within a <a href="#request-dispatch">handler</a>
        which are not caught by the handler itself are handled by
        Hunchentoot by logging them to the established
        <clix:ref>ACCEPTOR-MESSAGE-LOG-DESTINATION</clix:ref>.
      </p>

      <clix:function name='log-message*'>
        <clix:lambda-list>log-level format-string
        <clix:lkw>rest
        </clix:lkw> format-arguments
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>
          Convenience function which calls the message
          logger of the current acceptor (if there is one) with the same
          arguments it accepts.  Returns <code>NIL</code> if there is no message
          logger or whatever the message logger returns.
          <p>
            This is the function which Hunchentoot itself uses to log errors it
            catches during request processing.
          </p>
        </clix:description>
      </clix:function>

      <clix:special-variable name='*log-lisp-errors-p*'>
        <clix:description>
          Whether Lisp errors in request handlers should be logged.
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*log-lisp-backtraces-p*'>
        <clix:description>
          Whether Lisp backtraces should be logged.  Only
          has an effect if <clix:ref>*LOG-LISP-ERRORS-P*</clix:ref> is true
          as well.
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*log-lisp-warnings-p*'>
        <clix:description>
          Whether Lisp warnings in request handlers should be logged.
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*lisp-errors-log-level*'>
        <clix:description>
          Log level for Lisp errors.  Should be one
          of <code>:ERROR</code> (the default), <code>:WARNING</code>,
          or <code>:INFO</code>.
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*lisp-warnings-log-level*'>
        <clix:description>
          Log level for Lisp warnings.
          Should be one of <code>:ERROR</code>, <code>:WARNING</code>
          (the default), or <code>:INFO</code>.
        </clix:description>
      </clix:special-variable>
    </clix:subchapter>

    <clix:subchapter name="conditions" title="Conditions and error handling">
      <p>
        This section describes how Hunchentoot deals with exceptional
        situations.  See also the secion about <a href="#logging">logging</a>.
      </p>
      <p>
        When an error occurs while processing a request, Hunchentoot's
        default behavior is to catch the error, log it and
        optionally display it to the client in the HTML response.
        This behavior can be customized through the values of a number
        of special variables, which are documented below.
      </p>

      <clix:special-variable name='*catch-errors-p*'>
        <clix:description>
          If the value of this variable is <code>NIL</code>
          (the default is <code>T</code>), then errors which happen while a
          request is handled aren't <a href="#logging">caught as usual</a>, but
          instead your
          Lisp's <a
          href="http://www.lispworks.com/documentation/HyperSpec/Body/26_glo_d.htm#debugger">debugger</a>
          is <a
          href="http://www.lispworks.com/documentation/HyperSpec/Body/f_invoke.htm">invoked</a>.
          This variable should obviously always be set to a <em>true</em> value
          in a production environment.
          See <clix:ref>MAYBE-INVOKE-DEBUGGER</clix:ref>
          if you want to fine-tune this behaviour.
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*show-lisp-errors-p*'>
        <clix:description>
          Whether Lisp errors should be shown in HTML output.  Note
          that this only affects canned responses generated by Lisp.
          If an error template is present for the "internal server
          error" status code, this special variable is not used (see
          <clix:ref>acceptor-status-message</clix:ref>).
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*show-lisp-backtraces-p*'>
        <clix:description>
          Whether Lisp backtraces should be shown in HTML output if
          <clix:ref>*SHOW-LISP-ERRORS-P*</clix:ref> is true and an error occurs.
        </clix:description>
      </clix:special-variable>

      <clix:function generic='true' name='maybe-invoke-debugger'>
        <clix:lambda-list>condition
        </clix:lambda-list>
        <clix:returns>|
        </clix:returns>
        <clix:description>
          This generic function is called whenever a
          <a
              href="http://www.lispworks.com/documentation/HyperSpec/Body/09_.htm">condition</a> <code><i>condition</i></code>
          is signaled in Hunchentoot.  You might want to specialize it on
          specific condition classes for debugging purposes.  The default
          method <a
          href="http://www.lispworks.com/documentation/HyperSpec/Body/f_invoke.htm">invokes
          the debugger</a> with <clix:arg>condition</clix:arg> if
          <clix:ref>*CATCH-ERRORS-P*</clix:ref> is <code>NIL</code>.
        </clix:description>
      </clix:function>

      <clix:condition name='hunchentoot-condition'>
        <clix:description>
          Superclass for all conditions related to Hunchentoot.
        </clix:description>
      </clix:condition>

      <clix:condition name='hunchentoot-error'>
        <clix:description>
          Superclass for all errors related to Hunchentoot and a subclass of <clix:ref>HUNCHENTOOT-CONDITION</clix:ref>.
        </clix:description>
      </clix:condition>

      <clix:condition name='parameter-error'>
        <clix:description>
          Signalled if a function was called with incosistent or illegal parameters.  A subclass of <clix:ref>HUNCHENTOOT-ERROR</clix:ref>.
        </clix:description>
      </clix:condition>

      <clix:condition name='hunchentoot-warning'>
        <clix:description>
          Superclass for all warnings related to Hunchentoot and a subclass of <clix:ref>HUNCHENTOOT-CONDITION</clix:ref>.
        </clix:description>
      </clix:condition>

    </clix:subchapter>

    <clix:subchapter name="misc" title="Miscellaneous">

      Various functions and variables which didn't fit into one of the
      other categories.

      <clix:function name='abort-request-handler'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> result
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>
          This function can be called by a request handler
          at any time to immediately abort handling the request.  This works as
          if the handler had returned <clix:arg>result</clix:arg>.  See the
          source code of <clix:ref>REDIRECT</clix:ref> for an example.
        </clix:description>
      </clix:function>

      <clix:function name="handle-if-modified-since">
        <clix:lambda-list>time <clix:lkw>optional</clix:lkw> request</clix:lambda-list>
        <clix:returns>|</clix:returns>
        <clix:description>
          This function is designed to be used inside
          a <a href="#request-dispatch">handler</a>. If the client has sent an
          'If-Modified-Since' header
          (see <a href="http://www.faqs.org/rfcs/rfc2616.html">RFC2616</a>,
          section 14.25) and the time specified matches the universal
          time
          <clix:arg>time</clix:arg> then the
          header <clix:ref>+HTTP-NOT-MODIFIED+</clix:ref> with no content
          is immediately returned to the client.
          <p>
            Note that for this function to be useful you should usually
            send 'Last-Modified' headers back to the client. See the
            code
            of <clix:ref>CREATE-STATIC-FILE-DISPATCHER-AND-HANDLER</clix:ref>
            for an example.
          </p>
        </clix:description>
      </clix:function>

      <clix:function name="handle-static-file">
        <clix:lambda-list>path <clix:lkw>optional</clix:lkw> content-type</clix:lambda-list>
        <clix:returns>nil</clix:returns>
        <clix:description>
          Sends the file denoted by the pathname designator
          <clix:arg>path</clix:arg> with content type
          <clix:arg>content-type</clix:arg> to the client.  Sets the
          necessary handlers.  In particular the function employs
          <clix:ref>HANDLE-IF-MODIFIED-SINCE</clix:ref>.
          <p>
            If <clix:arg>content-type</clix:arg> is <code>NIL</code> the
            function tries to determine the correct content type from
            the file's suffix or falls back
            to <code>"application/octet-stream"</code> as a last resort.
          </p>
          <p>
            Note that this function
            calls <clix:ref>SEND-HEADERS</clix:ref> internally, so after
            you've called it, the headers are sent and the return value
            of your handler is ignored.
          </p>
        </clix:description>
      </clix:function>

      <clix:function name="redirect">
        <clix:lambda-list>target <clix:lkw>key</clix:lkw> host port protocol add-session-id code</clix:lambda-list>
        <clix:returns>|</clix:returns>
        <clix:description>
          Sends back appropriate headers to redirect the client
          to <clix:arg>target</clix:arg> (a string).
          <p>
            If <clix:arg>target</clix:arg> is a full URL starting with a
            scheme, <clix:arg>host</clix:arg>, <clix:arg>port</clix:arg>,
            and <clix:arg>protocol</clix:arg> are ignored.
            Otherwise, <clix:arg>target</clix:arg> should denote the path
            part of a URL, <clix:arg>protocol</clix:arg> must be one of
            the keywords <code>:HTTP</code> or <code>:HTTPS</code>, and
            the URL to redirect to will be constructed
            from <clix:arg>host</clix:arg>, <clix:arg>port</clix:arg>, <clix:arg>protocol</clix:arg>,
            and <clix:arg>target</clix:arg>.
          </p>
          <p>
            <clix:arg>code</clix:arg> must be a 3xx HTTP redirection
            status code to send to the client.  It defaults to 302
            ("Found").  If <clix:arg>host</clix:arg> is not provided,
            the current host (see <clix:ref>HOST</clix:ref>) will be
            used. If <clix:arg>protocol</clix:arg> is the keyword
            <code>:HTTPS</code>, the client will be redirected to a
            https URL, if it's <code>:HTTP</code> it'll be sent to a
            http URL.  If both <clix:arg>host</clix:arg> and
            <clix:arg>protocol</clix:arg> aren't provided, then the
            value of <clix:arg>protocol</clix:arg> will match the
            current request.
          </p>
        </clix:description>
      </clix:function>

      <clix:function name="require-authorization">
        <clix:lambda-list><clix:lkw>optional</clix:lkw> realm</clix:lambda-list>
        <clix:returns>|</clix:returns>
        <clix:description>
          Sends back appropriate headers to require basic HTTP
          authentication
          (see <a href="http://www.faqs.org/rfcs/rfc2617.html">RFC2617</a>)
          for the realm <clix:arg>realm</clix:arg>. The default value
          for <clix:arg>realm</clix:arg> is <code>"Hunchentoot"</code>.
        </clix:description>
      </clix:function>

      <clix:function name='no-cache'>
        <clix:lambda-list>
        </clix:lambda-list>
        <clix:returns>|
        </clix:returns>
        <clix:description>
          Adds appropriate headers to completely prevent caching on most browsers.
        </clix:description>
      </clix:function>

      <clix:function name='ssl-p'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> acceptor
        </clix:lambda-list>
        <clix:returns>generalized-boolean
        </clix:returns>
        <clix:description>
          Whether the current connection to the client is secure.  See <clix:ref>ACCEPTOR-SSL-P</clix:ref>.
        </clix:description>
      </clix:function>

      <clix:function name='reason-phrase'>
        <clix:lambda-list>return-code
        </clix:lambda-list>
        <clix:returns>string
        </clix:returns>
        <clix:description>
          Returns a reason phrase for the HTTP return code <clix:arg>return-code</clix:arg>
          (which should be an integer) or <code>NIL</code> for return codes Hunchentoot
          doesn&#039;t know.
        </clix:description>
      </clix:function>

      <clix:function name='rfc-1123-date'>
        <clix:lambda-list>
          <clix:lkw>optional
          </clix:lkw> time
        </clix:lambda-list>
        <clix:returns>string
        </clix:returns>
        <clix:description>
          Generates a time string according to <a href="http://www.faqs.org/rfcs/rfc1123.html">RFC 1123</a>.  Default is current time.
          This can be used to send a 'Last-Modified' header - see <clix:ref>HANDLE-IF-MODIFIED-SINCE</clix:ref>.
        </clix:description>
      </clix:function>

      <clix:function name='url-encode'>
        <clix:lambda-list>string
        <clix:lkw>optional
        </clix:lkw> external-format
        </clix:lambda-list>
        <clix:returns>string
        </clix:returns>
        <clix:description>
          URL-encodes a string using the external format <clix:arg>external-format</clix:arg>.  The default for <clix:arg>external-format</clix:arg> is the value of <clix:ref>*HUNCHENTOOT-DEFAULT-EXTERNAL-FORMAT*</clix:ref>.
        </clix:description>
      </clix:function>

      <clix:function name='url-decode'>
        <clix:lambda-list>string
        <clix:lkw>optional
        </clix:lkw> external-format
        </clix:lambda-list>
        <clix:returns>string
        </clix:returns>
        <clix:description>
          Decodes a URL-encoded string which is assumed to
          be encoded using the external
          format <clix:arg>external-format</clix:arg>, i.e. this is the inverse
          of <clix:ref>URL-ENCODE</clix:ref>. It is assumed that you'll rarely
          need this function, if ever. But just in case - here it is.  The
          default for <clix:arg>external-format</clix:arg> is the value
          of <clix:ref>*HUNCHENTOOT-DEFAULT-EXTERNAL-FORMAT*</clix:ref>.
        </clix:description>
      </clix:function>

      <clix:function name='escape-for-html'>
        <clix:lambda-list>string
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>
          Escapes the characters #\&lt;, #\&gt;, #\&#039;, #\&quot;, and #\&amp; for HTML output.
        </clix:description>
      </clix:function>

      <clix:function name="http-token-p">
        <clix:lambda-list>object</clix:lambda-list>
        <clix:returns>generalized-boolean</clix:returns>
        <clix:description>
          This function tests whether <clix:arg>object</clix:arg> is a
          non-empty string which is a <em>token</em> according
          to <a href="http://www.faqs.org/rfcs/rfc2068.html">RFC
          2068</a> (i.e. whether it may be used for, say, cookie names).
        </clix:description>
      </clix:function>

      <clix:function name='mime-type'>
        <clix:lambda-list>pathspec
        </clix:lambda-list>
        <clix:returns>result
        </clix:returns>
        <clix:description>
          Given a pathname designator <clix:arg>pathspec</clix:arg> returns the <a href="http://en.wikipedia.org/wiki/Internet_media_type">MIME type</a>
          (as a string) corresponding to the suffix of the file denoted by
          <clix:arg>pathspec</clix:arg> (or <code>NIL</code>).
        </clix:description>
      </clix:function>

      <clix:function name='within-request-p'>
        <clix:lambda-list>
        </clix:lambda-list>
        <clix:returns>generalized-boolean
        </clix:returns>
        <clix:description>
          Returns true if in the context of a request. Otherwise, <code>NIL</code>.
        </clix:description>
      </clix:function>

      <clix:special-variable name="*tmp-directory*">
        <clix:description>
          This should be a pathname denoting a directory where temporary
          files can be stored. It is used for <a href="#upload">file
          uploads</a>.
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*header-stream*'>
        <clix:description>
          If this variable is not <code>NIL</code>, it should be bound to a stream to
          which incoming and outgoing headers will be written for debugging
          purposes.
        </clix:description>
      </clix:special-variable>


      <clix:special-variable name='*cleanup-function*'>
        <clix:description>
          A designator for a function without arguments which is called on a
          regular basis if <clix:ref>*CLEANUP-INTERVAL*</clix:ref> is not <code>NIL</code>.  The initial value is
          the name of a function which invokes a garbage collection on 32-bit
          versions of LispWorks.
          <p>
            This variable is only available on LispWorks.
          </p>
        </clix:description>
      </clix:special-variable>

      <clix:special-variable name='*cleanup-interval*'>
        <clix:description>
          Should be <code>NIL</code> or a positive integer.  The system calls
          <clix:ref>*CLEANUP-FUNCTION*</clix:ref>
          whenever <clix:ref>*CLEANUP-INTERVAL*</clix:ref> new worker threads
          (counted globally across all acceptors) have been created unless the
          value is <code>NIL</code>.  The initial value is 100.
          <p>
            This variable is only available on LispWorks.
          </p>
        </clix:description>
      </clix:special-variable>
    </clix:subchapter>
  </clix:chapter>

  <clix:chapter name="testing" title="Testing">
    Hunchentoot comes with a test script which verifies that the
    example web server responds as expected.  This test script uses the
    <a href="http://weitz.de/drakma/">Drakma</a> HTTP client library
    and thus shares a significant amount of its base code with
    Hunchentoot itself.  Still, running the test script is a useful
    confidence test, and it is also possible to run the script across
    machines in order to verify a new Hunchentoot (or, for that matter
    Drakma) port.
    <p>
      To run the confidence test, <a href="#teen-age">start
      the example web server</a>.  Then, in your Lisp
      listener, type
<pre>(<a class="noborder" href="hunchentoot-test:test-hunchentoot">hunchentoot-test:test-hunchentoot</a> "http://localhost:4242")</pre>
You will see some diagnostic output and a summary line that
reports whether any tests have failed.  (You can also use the
example certificate and key files in the test directory and
start and test an https server instead.)
    </p>

    <clix:function name="hunchentoot-test:test-hunchentoot">
      <clix:lambda-list>base-url <clix:lkw>key</clix:lkw></clix:lambda-list>
      <clix:returns>|</clix:returns>
      <clix:description>
        Runs the built-in confidence
        test.  <clix:arg>base-url</clix:arg> is the base URL to use
        for testing, it should not have a trailing slash.  The keyword
        arguments accepted are for future extension and should not
        currently be used.
        <p>
          The script expects the Hunchentoot example test server to be
          running at the given <clix:arg>base-url</clix:arg> and
          retrieves various pages from that server, expecting certain
          responses.
        </p>
      </clix:description>
    </clix:function>
  </clix:chapter>

  <clix:chapter name="debugging" title="Debugging">
    By default, Hunchentoot intercepts all errors that occur while
    executing request handlers, logs them to the log file and displays
    a static error page to the user.  While developing applications,
    you may want to change that behavior so that the debugger is
    invoked when an error occurs.  You can set
    the <clix:ref>*CATCH-ERRORS-P*</clix:ref> to <code>NIL</code> to
    make that happen.  Alternatively, you may want to have Hunchentoot
    display detailed error information in the error response page.
    You can set the <clix:ref>*SHOW-LISP-ERRORS-P*</clix:ref> to a
    true value to make that happen.  If you don't want to see Lisp
    backtraces in these error pages, you can
    set <clix:ref>*SHOW-LISP-BACKTRACES-P*</clix:ref>
    to <code>NIL</code>.
  </clix:chapter>

  <clix:chapter name="history" title="History">

    Hunchentoot's predecessor <a href="http://weitz.de/tbnl/">TBNL</a>
    (which is short for "To Be Named Later") grew over the years as a
    toolkit that I used for various commercial and private
    projects. In August2003, Daniel Barlow started
    a <a href="http://article.gmane.org/gmane.lisp.web/148">review of
      web APIs</a> on
    the <a href="http://www.red-bean.com/lispweb/">lispweb</a> mailing
    list and
    I <a href="http://article.gmane.org/gmane.lisp.web/153">described</a>
    the API of my hitherto-unreleased bunch of code (and christened it
    "TBNL").
    <p>
      It turned out that
      <a href="http://www.jeffcaldwell.com/">Jeff Caldwell</a> had
      worked on something similar so he emailed me and proposed to
      join our efforts. As I had no immediate plans to release my code
      (which was poorly organized, undocumented, and mostly
      CMUCL-specific), I gave it to Jeff and he worked towards a
      release. He added docstrings, refactored, added some stuff, and
      based it on KMRCL to make it portable across several Lisp
      implementations.
    </p>
    <p>
      Unfortunately, Jeff is at least as busy as I am so he didn't
      find the time to finish a full release.  But in spring2004 I
      needed a documented version of the code for a client of mine who
      thought it would be good if the toolkit were publicly available
      under an open source license. So I took Jeff's code, refactored
      again (to sync with the changes I had done in the meantime), and
      added documentation.  This resulted in TBNL0.1.0 (which
      initially required mod_lisp as its front-end).
    </p>
    <p>
      In March2005, Bob Hutchinson sent patches which enabled TBNL to
      use other front-ends than mod_lisp.  This made me aware that
      TBNL was already <em>almost</em> a full web server, so
      eventually I wrote Hunchentoot which <em>was</em> a full web
      server, implemented as a wrapper around TBNL.  Hunchentoot0.1.0
      was released at the end of 2005 and was originally
      LispWorks-only.
    </p>
    <p>
      Hunchentoot0.4.0, released in October2006, was the first
      release which also worked with other Common Lisp
      implementations.  It is a major rewrite and also incorporates
      most of TBNL and replaces it completely.
    </p>
    <p>
      Hunchentoot1.0.0, released in February 2009, is again a major
      rewrite and should be considered work in progress.  It moved to
      using
      the <a href="http://common-lisp.net/project/usocket/">usocket</a>
      and <a href="http://common-lisp.net/project/bordeaux-threads/">Bordeaux
      Threads</a> libraries for non-LispWorks Lisps, thereby removing most of
      the platform dependent code.  Threading behaviour was made
      controllable through the introduction of
      taskmasters.  <a href="http://www.cliki.net/mod_lisp">mod_lisp</a>
      support and several other things were removed in this release to
      simplify the code base (and partly due to the lack of interest).
      Several architectural changes (lots of them not
      backwards-compatible) were made to ease customization of
      Hunchentoot's behaviour.  A significant part of the 1.0.0
      redesign was done
      by <a href="http://netzhansa.blogspot.com/">Hans Hbner</a>.
    </p>
  </clix:chapter>

  <clix:chapter name="index" title="Symbol index">

    Here are all exported symbols of the <code>HUNCHENTOOT</code>
    package in alphabetical order linked to their corresponding
    documentation entries:

    <clix:index/>

  </clix:chapter>

  <clix:chapter name="ack" title="Acknowledgements">

    Thanks to Jeff Caldwell - TBNL would not have been released
    without his efforts.  Thanks
    to <a href="http://www.cliki.net/Stefan%20Scholl">Stefan
    Scholl</a> and Travis Cross for various additions and fixes to
    TBNL, to <a href="http://www.foldr.org/~michaelw/">Michael
    Weber</a> for initial file upload code, and
    to <a href="http://www.ltn.lv/~jonis/">Janis Dzerins</a> for
    his <a href="http://common-lisp.net/project/rfc2388/">RFC2388
    code</a>. Thanks to Bob Hutchison for his code for multiple
    front-ends (which made me realize that TBNL was already pretty
    close to a "real" web server) and the initial UTF-8 example.
    Thanks to <a href="http://netzhansa.blogspot.com/">Hans Hbner</a>
    for a lot of architectural and implementation enhancements for the
    1.0.0 release and also for transferring the documentation to sane
    XHTML.  Thanks to John
    Foderaro's <a href="http://opensource.franz.com/aserve/index.html">AllegroServe</a>
    for inspiration.  Thanks to <a href="http://www.htg1.de/">Uwe von
    Loh</a> for
    the <a href="http://www.htg1.de/hunchentoot/hunchentoot.html">Hunchentoot
    logo</a>.

    <p>
      Hunchentoot originally used code
      from <a href="http://www.cliki.net/ACL-COMPAT">ACL-COMPAT</a>,
      specifically the chunking code from Jochen Schmidt.  (This has been
      replaced by <a href="http://weitz.de/chunga/">Chunga</a>.)  When I ported
      Hunchentoot to other Lisps than LispWorks, I stole code from
      ACL-COMPAT, <a href="http://www.cliki.net/kmrcl">KMRCL</a>,
      and <a href="http://www.cliki.net/trivial-sockets">trivial-sockets</a> for
      implementation-dependent stuff like sockets and MP.  (This has been replaced by
      <a href="http://common-lisp.net/project/bordeaux-threads/">Bordeaux
      Threads</a>
      and <a href="http://common-lisp.net/project/usocket/">usocket</a>.)
    </p>
    <p>
      Parts of this documentation were prepared
      with <a href="http://weitz.de/documentation-template/">DOCUMENTATION-TEMPLATE</a>,
      no animals were harmed.
    </p>
  </clix:chapter>
  <p>
    <a href='http://weitz.de/index.html'>BACK TO MY HOMEPAGE
    </a>
  </p>
</clix:documentation>