view en/ch13-hgext.xml @ 749:7e7c47481e4f

Oops, this is the real merge for my hg's oddity
author Dongsheng Song <dongsheng.song@gmail.com>
date Fri, 20 Mar 2009 16:43:35 +0800
parents en/ch14-hgext.xml@d0160b0b1a9e
children 1c13ed2130a7
line wrap: on
line source

<!-- vim: set filetype=docbkxml shiftwidth=2 autoindent expandtab tw=77 : -->

<chapter id="chap.hgext">
  <?dbhtml filename="adding-functionality-with-extensions.html"?>
  <title>Adding functionality with extensions</title>

  <para>While the core of Mercurial is quite complete from a
    functionality standpoint, it's deliberately shorn of fancy
    features.  This approach of preserving simplicity keeps the
    software easy to deal with for both maintainers and users.</para>

  <para>However, Mercurial doesn't box you in with an inflexible
    command set: you can add features to it as
    <emphasis>extensions</emphasis> (sometimes known as
    <emphasis>plugins</emphasis>).  We've already discussed a few of
    these extensions in earlier chapters.</para>
  <itemizedlist>
    <listitem><para>Section <xref linkend="sec.tour-merge.fetch"/>
	covers the <literal role="hg-ext">fetch</literal> extension;
	this combines pulling new changes and merging them with local
	changes into a single command, <command
	  role="hg-ext-fetch">fetch</command>.</para>
    </listitem>
    <listitem><para>In chapter <xref linkend="chap.hook"/>, we covered
	several extensions that are useful for hook-related
	functionality: <literal role="hg-ext">acl</literal> adds
	access control lists; <literal
	  role="hg-ext">bugzilla</literal> adds integration with the
	Bugzilla bug tracking system; and <literal
	  role="hg-ext">notify</literal> sends notification emails on
	new changes.</para>
    </listitem>
    <listitem><para>The Mercurial Queues patch management extension is
	so invaluable that it merits two chapters and an appendix all
	to itself. Chapter <xref linkend="chap.mq"/> covers the
	basics; chapter <xref
	  linkend="chap.mq-collab"/> discusses advanced topics;
	and appendix <xref linkend="chap.mqref"/> goes into detail on
	each
	command.</para>
    </listitem></itemizedlist>

  <para>In this chapter, we'll cover some of the other extensions that
    are available for Mercurial, and briefly touch on some of the
    machinery you'll need to know about if you want to write an
    extension of your own.</para>
  <itemizedlist>
    <listitem><para>In section <xref linkend="sec.hgext.inotify"/>,
	we'll discuss the possibility of <emphasis>huge</emphasis>
	performance improvements using the <literal
	  role="hg-ext">inotify</literal> extension.</para>
    </listitem></itemizedlist>

  <sect1 id="sec.hgext.inotify">
    <title>Improve performance with the <literal
	role="hg-ext">inotify</literal> extension</title>

    <para>Are you interested in having some of the most common
      Mercurial operations run as much as a hundred times faster?
      Read on!</para>

    <para>Mercurial has great performance under normal circumstances.
      For example, when you run the <command role="hg-cmd">hg
	status</command> command, Mercurial has to scan almost every
      directory and file in your repository so that it can display
      file status.  Many other Mercurial commands need to do the same
      work behind the scenes; for example, the <command
	role="hg-cmd">hg diff</command> command uses the status
      machinery to avoid doing an expensive comparison operation on
      files that obviously haven't changed.</para>

    <para>Because obtaining file status is crucial to good
      performance, the authors of Mercurial have optimised this code
      to within an inch of its life.  However, there's no avoiding the
      fact that when you run <command role="hg-cmd">hg
	status</command>, Mercurial is going to have to perform at
      least one expensive system call for each managed file to
      determine whether it's changed since the last time Mercurial
      checked.  For a sufficiently large repository, this can take a
      long time.</para>

    <para>To put a number on the magnitude of this effect, I created a
      repository containing 150,000 managed files.  I timed <command
	role="hg-cmd">hg status</command> as taking ten seconds to
      run, even when <emphasis>none</emphasis> of those files had been
      modified.</para>

    <para>Many modern operating systems contain a file notification
      facility. If a program signs up to an appropriate service, the
      operating system will notify it every time a file of interest is
      created, modified, or deleted.  On Linux systems, the kernel
      component that does this is called
      <literal>inotify</literal>.</para>

    <para>Mercurial's <literal role="hg-ext">inotify</literal>
      extension talks to the kernel's <literal>inotify</literal>
      component to optimise <command role="hg-cmd">hg status</command>
      commands.  The extension has two components.  A daemon sits in
      the background and receives notifications from the
      <literal>inotify</literal> subsystem.  It also listens for
      connections from a regular Mercurial command.  The extension
      modifies Mercurial's behaviour so that instead of scanning the
      filesystem, it queries the daemon.  Since the daemon has perfect
      information about the state of the repository, it can respond
      with a result instantaneously, avoiding the need to scan every
      directory and file in the repository.</para>

    <para>Recall the ten seconds that I measured plain Mercurial as
      taking to run <command role="hg-cmd">hg status</command> on a
      150,000 file repository.  With the <literal
	role="hg-ext">inotify</literal> extension enabled, the time
      dropped to 0.1 seconds, a factor of <emphasis>one
	hundred</emphasis> faster.</para>

    <para>Before we continue, please pay attention to some
      caveats.</para>
    <itemizedlist>
      <listitem><para>The <literal role="hg-ext">inotify</literal>
	  extension is Linux-specific.  Because it interfaces directly
	  to the Linux kernel's <literal>inotify</literal> subsystem,
	  it does not work on other operating systems.</para>
      </listitem>
      <listitem><para>It should work on any Linux distribution that
	  was released after early 2005.  Older distributions are
	  likely to have a kernel that lacks
	  <literal>inotify</literal>, or a version of
	  <literal>glibc</literal> that does not have the necessary
	  interfacing support.</para>
      </listitem>
      <listitem><para>Not all filesystems are suitable for use with
	  the <literal role="hg-ext">inotify</literal> extension.
	  Network filesystems such as NFS are a non-starter, for
	  example, particularly if you're running Mercurial on several
	  systems, all mounting the same network filesystem.  The
	  kernel's <literal>inotify</literal> system has no way of
	  knowing about changes made on another system.  Most local
	  filesystems (e.g. ext3, XFS, ReiserFS) should work
	  fine.</para>
      </listitem></itemizedlist>

    <para>The <literal role="hg-ext">inotify</literal> extension is
      not yet shipped with Mercurial as of May 2007, so it's a little
      more involved to set up than other extensions.  But the
      performance improvement is worth it!</para>

    <para>The extension currently comes in two parts: a set of patches
      to the Mercurial source code, and a library of Python bindings
      to the <literal>inotify</literal> subsystem.</para>
    <note>
      <para>  There are <emphasis>two</emphasis> Python
	<literal>inotify</literal> binding libraries.  One of them is
	called <literal>pyinotify</literal>, and is packaged by some
	Linux distributions as <literal>python-inotify</literal>.
	This is <emphasis>not</emphasis> the one you'll need, as it is
	too buggy and inefficient to be practical.</para>
    </note>
    <para>To get going, it's best to already have a functioning copy
      of Mercurial installed.</para>
    <note>
      <para>  If you follow the instructions below, you'll be
	<emphasis>replacing</emphasis> and overwriting any existing
	installation of Mercurial that you might already have, using
	the latest <quote>bleeding edge</quote> Mercurial code. Don't
	say you weren't warned!</para>
    </note>
    <orderedlist>
      <listitem><para>Clone the Python <literal>inotify</literal>
	  binding repository.  Build and install it.</para>
	<programlisting>hg clone http://hg.kublai.com/python/inotify
cd inotify
python setup.py build --force
sudo python setup.py install --skip-build</programlisting>
      </listitem>
      <listitem><para>Clone the <filename
	    class="directory">crew</filename> Mercurial repository.
	  Clone the <literal role="hg-ext">inotify</literal> patch
	  repository so that Mercurial Queues will be able to apply
	  patches to your cope of the <filename
	    class="directory">crew</filename> repository.</para>
	<programlisting>hg clone http://hg.intevation.org/mercurial/crew
hg clone crew inotify
hg clone http://hg.kublai.com/mercurial/patches/inotify inotify/.hg/patches</programlisting>
      </listitem>
      <listitem><para>Make sure that you have the Mercurial Queues
	  extension, <literal role="hg-ext">mq</literal>, enabled.  If
	  you've never used MQ, read section <xref
	    linkend="sec.mq.start"/> to get started
	  quickly.</para>
      </listitem>
      <listitem><para>Go into the <filename
	    class="directory">inotify</filename> repo, and apply all
	  of the <literal role="hg-ext">inotify</literal> patches
	  using the <option role="hg-ext-mq-cmd-qpush-opt">hg
	    -a</option> option to the <command
	    role="hg-ext-mq">qpush</command> command.</para>
	<programlisting>cd inotify
hg qpush -a</programlisting>
      </listitem>
      <listitem><para>  If you get an error message from <command
	    role="hg-ext-mq">qpush</command>, you should not continue.
	  Instead, ask for help.</para>
      </listitem>
      <listitem><para>Build and install the patched version of
	  Mercurial.</para>
	<programlisting>python setup.py build --force
sudo python setup.py install --skip-build</programlisting>
      </listitem>
    </orderedlist>
    <para>Once you've build a suitably patched version of Mercurial,
      all you need to do to enable the <literal
	role="hg-ext">inotify</literal> extension is add an entry to
      your <filename role="special">~/.hgrc</filename>.</para>
    <programlisting>[extensions] inotify =</programlisting>
    <para>When the <literal role="hg-ext">inotify</literal> extension
      is enabled, Mercurial will automatically and transparently start
      the status daemon the first time you run a command that needs
      status in a repository.  It runs one status daemon per
      repository.</para>

    <para>The status daemon is started silently, and runs in the
      background.  If you look at a list of running processes after
      you've enabled the <literal role="hg-ext">inotify</literal>
      extension and run a few commands in different repositories,
      you'll thus see a few <literal>hg</literal> processes sitting
      around, waiting for updates from the kernel and queries from
      Mercurial.</para>

    <para>The first time you run a Mercurial command in a repository
      when you have the <literal role="hg-ext">inotify</literal>
      extension enabled, it will run with about the same performance
      as a normal Mercurial command.  This is because the status
      daemon needs to perform a normal status scan so that it has a
      baseline against which to apply later updates from the kernel.
      However, <emphasis>every</emphasis> subsequent command that does
      any kind of status check should be noticeably faster on
      repositories of even fairly modest size.  Better yet, the bigger
      your repository is, the greater a performance advantage you'll
      see.  The <literal role="hg-ext">inotify</literal> daemon makes
      status operations almost instantaneous on repositories of all
      sizes!</para>

    <para>If you like, you can manually start a status daemon using
      the <command role="hg-ext-inotify">inserve</command> command.
      This gives you slightly finer control over how the daemon ought
      to run.  This command will of course only be available when the
      <literal role="hg-ext">inotify</literal> extension is
      enabled.</para>

    <para>When you're using the <literal
	role="hg-ext">inotify</literal> extension, you should notice
      <emphasis>no difference at all</emphasis> in Mercurial's
      behaviour, with the sole exception of status-related commands
      running a whole lot faster than they used to.  You should
      specifically expect that commands will not print different
      output; neither should they give different results. If either of
      these situations occurs, please report a bug.</para>

  </sect1>
  <sect1 id="sec.hgext.extdiff">
    <title>Flexible diff support with the <literal
	role="hg-ext">extdiff</literal> extension</title>

    <para>Mercurial's built-in <command role="hg-cmd">hg
	diff</command> command outputs plaintext unified diffs.</para>

    &interaction.extdiff.diff;

    <para>If you would like to use an external tool to display
      modifications, you'll want to use the <literal
	role="hg-ext">extdiff</literal> extension.  This will let you
      use, for example, a graphical diff tool.</para>

    <para>The <literal role="hg-ext">extdiff</literal> extension is
      bundled with Mercurial, so it's easy to set up.  In the <literal
	role="rc-extensions">extensions</literal> section of your
      <filename role="special">~/.hgrc</filename>, simply add a
      one-line entry to enable the extension.</para>
    <programlisting>[extensions]
extdiff =</programlisting>
    <para>This introduces a command named <command
	role="hg-ext-extdiff">extdiff</command>, which by default uses
      your system's <command>diff</command> command to generate a
      unified diff in the same form as the built-in <command
	role="hg-cmd">hg diff</command> command.</para>
    
    &interaction.extdiff.extdiff;

    <para>The result won't be exactly the same as with the built-in
      <command role="hg-cmd">hg diff</command> variations, because the
      output of <command>diff</command> varies from one system to
      another, even when passed the same options.</para>

    <para>As the <quote><literal>making snapshot</literal></quote>
      lines of output above imply, the <command
	role="hg-ext-extdiff">extdiff</command> command works by
      creating two snapshots of your source tree.  The first snapshot
      is of the source revision; the second, of the target revision or
      working directory.  The <command
	role="hg-ext-extdiff">extdiff</command> command generates
      these snapshots in a temporary directory, passes the name of
      each directory to an external diff viewer, then deletes the
      temporary directory.  For efficiency, it only snapshots the
      directories and files that have changed between the two
      revisions.</para>

    <para>Snapshot directory names have the same base name as your
      repository. If your repository path is <filename
	class="directory">/quux/bar/foo</filename>, then <filename
	class="directory">foo</filename> will be the name of each
      snapshot directory.  Each snapshot directory name has its
      changeset ID appended, if appropriate.  If a snapshot is of
      revision <literal>a631aca1083f</literal>, the directory will be
      named <filename class="directory">foo.a631aca1083f</filename>.
      A snapshot of the working directory won't have a changeset ID
      appended, so it would just be <filename
	class="directory">foo</filename> in this example.  To see what
      this looks like in practice, look again at the <command
	role="hg-ext-extdiff">extdiff</command> example above.  Notice
      that the diff has the snapshot directory names embedded in its
      header.</para>

    <para>The <command role="hg-ext-extdiff">extdiff</command> command
      accepts two important options. The <option
	role="hg-ext-extdiff-cmd-extdiff-opt">hg -p</option> option
      lets you choose a program to view differences with, instead of
      <command>diff</command>.  With the <option
	role="hg-ext-extdiff-cmd-extdiff-opt">hg -o</option> option,
      you can change the options that <command
	role="hg-ext-extdiff">extdiff</command> passes to the program
      (by default, these options are
      <quote><literal>-Npru</literal></quote>, which only make sense
      if you're running <command>diff</command>).  In other respects,
      the <command role="hg-ext-extdiff">extdiff</command> command
      acts similarly to the built-in <command role="hg-cmd">hg
	diff</command> command: you use the same option names, syntax,
      and arguments to specify the revisions you want, the files you
      want, and so on.</para>

    <para>As an example, here's how to run the normal system
      <command>diff</command> command, getting it to generate context
      diffs (using the <option role="cmd-opt-diff">-c</option> option)
      instead of unified diffs, and five lines of context instead of
      the default three (passing <literal>5</literal> as the argument
      to the <option role="cmd-opt-diff">-C</option> option).</para>

      &interaction.extdiff.extdiff-ctx;

    <para>Launching a visual diff tool is just as easy.  Here's how to
      launch the <command>kdiff3</command> viewer.</para>
    <programlisting>hg extdiff -p kdiff3 -o</programlisting>

    <para>If your diff viewing command can't deal with directories,
      you can easily work around this with a little scripting.  For an
      example of such scripting in action with the <literal
	role="hg-ext">mq</literal> extension and the
      <command>interdiff</command> command, see section <xref
	linkend="mq-collab.tips.interdiff"/>.</para>

    <sect2>
      <title>Defining command aliases</title>

      <para>It can be cumbersome to remember the options to both the
	<command role="hg-ext-extdiff">extdiff</command> command and
	the diff viewer you want to use, so the <literal
	  role="hg-ext">extdiff</literal> extension lets you define
	<emphasis>new</emphasis> commands that will invoke your diff
	viewer with exactly the right options.</para>

      <para>All you need to do is edit your <filename
	  role="special">~/.hgrc</filename>, and add a section named
	<literal role="rc-extdiff">extdiff</literal>.  Inside this
	section, you can define multiple commands.  Here's how to add
	a <literal>kdiff3</literal> command.  Once you've defined
	this, you can type <quote><literal>hg kdiff3</literal></quote>
	and the <literal role="hg-ext">extdiff</literal> extension
	will run <command>kdiff3</command> for you.</para>
      <programlisting>[extdiff]
cmd.kdiff3 =</programlisting>
      <para>If you leave the right hand side of the definition empty,
	as above, the <literal role="hg-ext">extdiff</literal>
	extension uses the name of the command you defined as the name
	of the external program to run.  But these names don't have to
	be the same.  Here, we define a command named
	<quote><literal>hg wibble</literal></quote>, which runs
	<command>kdiff3</command>.</para>
      <programlisting>[extdiff]
 cmd.wibble = kdiff3</programlisting>

      <para>You can also specify the default options that you want to
	invoke your diff viewing program with.  The prefix to use is
	<quote><literal>opts.</literal></quote>, followed by the name
	of the command to which the options apply.  This example
	defines a <quote><literal>hg vimdiff</literal></quote> command
	that runs the <command>vim</command> editor's
	<literal>DirDiff</literal> extension.</para>
      <programlisting>[extdiff]
 cmd.vimdiff = vim
opts.vimdiff = -f '+next' '+execute "DirDiff" argv(0) argv(1)'</programlisting>

    </sect2>
  </sect1>
  <sect1 id="sec.hgext.transplant">
    <title>Cherrypicking changes with the <literal
	role="hg-ext">transplant</literal> extension</title>

    <para>Need to have a long chat with Brendan about this.</para>

  </sect1>
  <sect1 id="sec.hgext.patchbomb">
    <title>Send changes via email with the <literal
	role="hg-ext">patchbomb</literal> extension</title>

    <para>Many projects have a culture of <quote>change
	review</quote>, in which people send their modifications to a
      mailing list for others to read and comment on before they
      commit the final version to a shared repository.  Some projects
      have people who act as gatekeepers; they apply changes from
      other people to a repository to which those others don't have
      access.</para>

    <para>Mercurial makes it easy to send changes over email for
      review or application, via its <literal
	role="hg-ext">patchbomb</literal> extension.  The extension is
      so named because changes are formatted as patches, and it's usual
      to send one changeset per email message.  Sending a long series
      of changes by email is thus much like <quote>bombing</quote> the
      recipient's inbox, hence <quote>patchbomb</quote>.</para>

    <para>As usual, the basic configuration of the <literal
	role="hg-ext">patchbomb</literal> extension takes just one or
      two lines in your <filename role="special">
	/.hgrc</filename>.</para>
    <programlisting>[extensions]
patchbomb =</programlisting>
    <para>Once you've enabled the extension, you will have a new
      command available, named <command
	role="hg-ext-patchbomb">email</command>.</para>

    <para>The safest and best way to invoke the <command
	role="hg-ext-patchbomb">email</command> command is to
      <emphasis>always</emphasis> run it first with the <option
	role="hg-ext-patchbomb-cmd-email-opt">hg -n</option> option.
      This will show you what the command <emphasis>would</emphasis>
      send, without actually sending anything.  Once you've had a
      quick glance over the changes and verified that you are sending
      the right ones, you can rerun the same command, with the <option
	role="hg-ext-patchbomb-cmd-email-opt">hg -n</option> option
      removed.</para>

    <para>The <command role="hg-ext-patchbomb">email</command> command
      accepts the same kind of revision syntax as every other
      Mercurial command.  For example, this command will send every
      revision between 7 and <literal>tip</literal>, inclusive.</para>
    <programlisting>hg email -n 7:tip</programlisting>
    <para>You can also specify a <emphasis>repository</emphasis> to
      compare with.  If you provide a repository but no revisions, the
      <command role="hg-ext-patchbomb">email</command> command will
      send all revisions in the local repository that are not present
      in the remote repository.  If you additionally specify revisions
      or a branch name (the latter using the <option
	role="hg-ext-patchbomb-cmd-email-opt">hg -b</option> option),
      this will constrain the revisions sent.</para>

    <para>It's perfectly safe to run the <command
	role="hg-ext-patchbomb">email</command> command without the
      names of the people you want to send to: if you do this, it will
      just prompt you for those values interactively.  (If you're
      using a Linux or Unix-like system, you should have enhanced
      <literal>readline</literal>-style editing capabilities when
      entering those headers, too, which is useful.)</para>

    <para>When you are sending just one revision, the <command
	role="hg-ext-patchbomb">email</command> command will by
      default use the first line of the changeset description as the
      subject of the single email message it sends.</para>

    <para>If you send multiple revisions, the <command
	role="hg-ext-patchbomb">email</command> command will usually
      send one message per changeset.  It will preface the series with
      an introductory message, in which you should describe the
      purpose of the series of changes you're sending.</para>

    <sect2>
      <title>Changing the behaviour of patchbombs</title>

      <para>Not every project has exactly the same conventions for
	sending changes in email; the <literal
	  role="hg-ext">patchbomb</literal> extension tries to
	accommodate a number of variations through command line
	options.</para>
      <itemizedlist>
	<listitem><para>You can write a subject for the introductory
	    message on the command line using the <option
	      role="hg-ext-patchbomb-cmd-email-opt">hg -s</option>
	    option.  This takes one argument, the text of the subject
	    to use.</para>
	</listitem>
	<listitem><para>To change the email address from which the
	    messages originate, use the <option
	      role="hg-ext-patchbomb-cmd-email-opt">hg -f</option>
	    option.  This takes one argument, the email address to
	    use.</para>
	</listitem>
	<listitem><para>The default behaviour is to send unified diffs
	    (see section <xref linkend="sec.mq.patch"/> for a
	    description of the
	    format), one per message.  You can send a binary bundle
	    instead with the <option
	      role="hg-ext-patchbomb-cmd-email-opt">hg -b</option>
	    option.</para>
	</listitem>
	<listitem><para>Unified diffs are normally prefaced with a
	    metadata header.  You can omit this, and send unadorned
	    diffs, with the <option
	      role="hg-ext-patchbomb-cmd-email-opt">hg
	      --plain</option> option.</para>
	</listitem>
	<listitem><para>Diffs are normally sent <quote>inline</quote>,
	    in the same body part as the description of a patch.  This
	    makes it easiest for the largest number of readers to
	    quote and respond to parts of a diff, as some mail clients
	    will only quote the first MIME body part in a message. If
	    you'd prefer to send the description and the diff in
	    separate body parts, use the <option
	      role="hg-ext-patchbomb-cmd-email-opt">hg -a</option>
	    option.</para>
	</listitem>
	<listitem><para>Instead of sending mail messages, you can
	    write them to an <literal>mbox</literal>-format mail
	    folder using the <option
	      role="hg-ext-patchbomb-cmd-email-opt">hg -m</option>
	    option.  That option takes one argument, the name of the
	    file to write to.</para>
	</listitem>
	<listitem><para>If you would like to add a
	    <command>diffstat</command>-format summary to each patch,
	    and one to the introductory message, use the <option
	      role="hg-ext-patchbomb-cmd-email-opt">hg -d</option>
	    option.  The <command>diffstat</command> command displays
	    a table containing the name of each file patched, the
	    number of lines affected, and a histogram showing how much
	    each file is modified.  This gives readers a qualitative
	    glance at how complex a patch is.</para>
	</listitem></itemizedlist>

    </sect2>
  </sect1>
</chapter>

<!--
local variables: 
sgml-parent-document: ("00book.xml" "book" "chapter")
end:
-->