Mercurial > hgbook
view en/ch11-mq.xml @ 775:29f0f79cf614
Update paragraph IDs
author | Bryan O'Sullivan <bos@serpentine.com> |
---|---|
date | Thu, 16 Apr 2009 23:46:45 -0700 |
parents | 3b640272a966 |
children | ef53d025f410 |
line wrap: on
line source
<!-- vim: set filetype=docbkxml shiftwidth=2 autoindent expandtab tw=77 : --> <chapter id="chap:mq"> <?dbhtml filename="managing-change-with-mercurial-queues.html"?> <title>Managing change with Mercurial Queues</title> <sect1 id="sec:mq:patch-mgmt"> <title>The patch management problem</title> <para id="x_3ac">Here is a common scenario: you need to install a software package from source, but you find a bug that you must fix in the source before you can start using the package. You make your changes, forget about the package for a while, and a few months later you need to upgrade to a newer version of the package. If the newer version of the package still has the bug, you must extract your fix from the older source tree and apply it against the newer version. This is a tedious task, and it's easy to make mistakes.</para> <para id="x_3ad">This is a simple case of the <quote>patch management</quote> problem. You have an <quote>upstream</quote> source tree that you can't change; you need to make some local changes on top of the upstream tree; and you'd like to be able to keep those changes separate, so that you can apply them to newer versions of the upstream source.</para> <para id="x_3ae">The patch management problem arises in many situations. Probably the most visible is that a user of an open source software project will contribute a bug fix or new feature to the project's maintainers in the form of a patch.</para> <para id="x_3af">Distributors of operating systems that include open source software often need to make changes to the packages they distribute so that they will build properly in their environments.</para> <para id="x_3b0">When you have few changes to maintain, it is easy to manage a single patch using the standard <command>diff</command> and <command>patch</command> programs (see <xref linkend="sec:mq:patch"/> for a discussion of these tools). Once the number of changes grows, it starts to make sense to maintain patches as discrete <quote>chunks of work,</quote> so that for example a single patch will contain only one bug fix (the patch might modify several files, but it's doing <quote>only one thing</quote>), and you may have a number of such patches for different bugs you need fixed and local changes you require. In this situation, if you submit a bug fix patch to the upstream maintainers of a package and they include your fix in a subsequent release, you can simply drop that single patch when you're updating to the newer release.</para> <para id="x_3b1">Maintaining a single patch against an upstream tree is a little tedious and error-prone, but not difficult. However, the complexity of the problem grows rapidly as the number of patches you have to maintain increases. With more than a tiny number of patches in hand, understanding which ones you have applied and maintaining them moves from messy to overwhelming.</para> <para id="x_3b2">Fortunately, Mercurial includes a powerful extension, Mercurial Queues (or simply <quote>MQ</quote>), that massively simplifies the patch management problem.</para> </sect1> <sect1 id="sec:mq:history"> <title>The prehistory of Mercurial Queues</title> <para id="x_3b3">During the late 1990s, several Linux kernel developers started to maintain <quote>patch series</quote> that modified the behavior of the Linux kernel. Some of these series were focused on stability, some on feature coverage, and others were more speculative.</para> <para id="x_3b4">The sizes of these patch series grew rapidly. In 2002, Andrew Morton published some shell scripts he had been using to automate the task of managing his patch queues. Andrew was successfully using these scripts to manage hundreds (sometimes thousands) of patches on top of the Linux kernel.</para> <sect2 id="sec:mq:quilt"> <title>A patchwork quilt</title> <para id="x_3b5">In early 2003, Andreas Gruenbacher and Martin Quinson borrowed the approach of Andrew's scripts and published a tool called <quote>patchwork quilt</quote> <citation>web:quilt</citation>, or simply <quote>quilt</quote> (see <citation>gruenbacher:2005</citation> for a paper describing it). Because quilt substantially automated patch management, it rapidly gained a large following among open source software developers.</para> <para id="x_3b6">Quilt manages a <emphasis>stack of patches</emphasis> on top of a directory tree. To begin, you tell quilt to manage a directory tree, and tell it which files you want to manage; it stores away the names and contents of those files. To fix a bug, you create a new patch (using a single command), edit the files you need to fix, then <quote>refresh</quote> the patch.</para> <para id="x_3b7">The refresh step causes quilt to scan the directory tree; it updates the patch with all of the changes you have made. You can create another patch on top of the first, which will track the changes required to modify the tree from <quote>tree with one patch applied</quote> to <quote>tree with two patches applied</quote>.</para> <para id="x_3b8">You can <emphasis>change</emphasis> which patches are applied to the tree. If you <quote>pop</quote> a patch, the changes made by that patch will vanish from the directory tree. Quilt remembers which patches you have popped, though, so you can <quote>push</quote> a popped patch again, and the directory tree will be restored to contain the modifications in the patch. Most importantly, you can run the <quote>refresh</quote> command at any time, and the topmost applied patch will be updated. This means that you can, at any time, change both which patches are applied and what modifications those patches make.</para> <para id="x_3b9">Quilt knows nothing about revision control tools, so it works equally well on top of an unpacked tarball or a Subversion working copy.</para> </sect2> <sect2 id="sec:mq:quilt-mq"> <title>From patchwork quilt to Mercurial Queues</title> <para id="x_3ba">In mid-2005, Chris Mason took the features of quilt and wrote an extension that he called Mercurial Queues, which added quilt-like behavior to Mercurial.</para> <para id="x_3bb">The key difference between quilt and MQ is that quilt knows nothing about revision control systems, while MQ is <emphasis>integrated</emphasis> into Mercurial. Each patch that you push is represented as a Mercurial changeset. Pop a patch, and the changeset goes away.</para> <para id="x_3bc">Because quilt does not care about revision control tools, it is still a tremendously useful piece of software to know about for situations where you cannot use Mercurial and MQ.</para> </sect2> </sect1> <sect1> <title>The huge advantage of MQ</title> <para id="x_3bd">I cannot overstate the value that MQ offers through the unification of patches and revision control.</para> <para id="x_3be">A major reason that patches have persisted in the free software and open source world&emdash;in spite of the availability of increasingly capable revision control tools over the years&emdash;is the <emphasis>agility</emphasis> they offer.</para> <para id="x_3bf">Traditional revision control tools make a permanent, irreversible record of everything that you do. While this has great value, it's also somewhat stifling. If you want to perform a wild-eyed experiment, you have to be careful in how you go about it, or you risk leaving unneeded&emdash;or worse, misleading or destabilising&emdash;traces of your missteps and errors in the permanent revision record.</para> <para id="x_3c0">By contrast, MQ's marriage of distributed revision control with patches makes it much easier to isolate your work. Your patches live on top of normal revision history, and you can make them disappear or reappear at will. If you don't like a patch, you can drop it. If a patch isn't quite as you want it to be, simply fix it&emdash;as many times as you need to, until you have refined it into the form you desire.</para> <para id="x_3c1">As an example, the integration of patches with revision control makes understanding patches and debugging their effects&emdash;and their interplay with the code they're based on&emdash;<emphasis>enormously</emphasis> easier. Since every applied patch has an associated changeset, you can give <command role="hg-cmd">hg log</command> a file name to see which changesets and patches affected the file. You can use the <command role="hg-cmd">hg bisect</command> command to binary-search through all changesets and applied patches to see where a bug got introduced or fixed. You can use the <command role="hg-cmd">hg annotate</command> command to see which changeset or patch modified a particular line of a source file. And so on.</para> </sect1> <sect1 id="sec:mq:patch"> <title>Understanding patches</title> <para id="x_3c2">Because MQ doesn't hide its patch-oriented nature, it is helpful to understand what patches are, and a little about the tools that work with them.</para> <para id="x_3c3">The traditional Unix <command>diff</command> command compares two files, and prints a list of differences between them. The <command>patch</command> command understands these differences as <emphasis>modifications</emphasis> to make to a file. Take a look below for a simple example of these commands in action.</para> &interaction.mq.dodiff.diff; <para id="x_3c4">The type of file that <command>diff</command> generates (and <command>patch</command> takes as input) is called a <quote>patch</quote> or a <quote>diff</quote>; there is no difference between a patch and a diff. (We'll use the term <quote>patch</quote>, since it's more commonly used.)</para> <para id="x_3c5">A patch file can start with arbitrary text; the <command>patch</command> command ignores this text, but MQ uses it as the commit message when creating changesets. To find the beginning of the patch content, <command>patch</command> searches for the first line that starts with the string <quote><literal>diff -</literal></quote>.</para> <para id="x_3c6">MQ works with <emphasis>unified</emphasis> diffs (<command>patch</command> can accept several other diff formats, but MQ doesn't). A unified diff contains two kinds of header. The <emphasis>file header</emphasis> describes the file being modified; it contains the name of the file to modify. When <command>patch</command> sees a new file header, it looks for a file with that name to start modifying.</para> <para id="x_3c7">After the file header comes a series of <emphasis>hunks</emphasis>. Each hunk starts with a header; this identifies the range of line numbers within the file that the hunk should modify. Following the header, a hunk starts and ends with a few (usually three) lines of text from the unmodified file; these are called the <emphasis>context</emphasis> for the hunk. If there's only a small amount of context between successive hunks, <command>diff</command> doesn't print a new hunk header; it just runs the hunks together, with a few lines of context between modifications.</para> <para id="x_3c8">Each line of context begins with a space character. Within the hunk, a line that begins with <quote><literal>-</literal></quote> means <quote>remove this line,</quote> while a line that begins with <quote><literal>+</literal></quote> means <quote>insert this line.</quote> For example, a line that is modified is represented by one deletion and one insertion.</para> <para id="x_3c9">We will return to some of the more subtle aspects of patches later (in <xref linkend="sec:mq:adv-patch"/>), but you should have enough information now to use MQ.</para> </sect1> <sect1 id="sec:mq:start"> <title>Getting started with Mercurial Queues</title> <para id="x_3ca">Because MQ is implemented as an extension, you must explicitly enable before you can use it. (You don't need to download anything; MQ ships with the standard Mercurial distribution.) To enable MQ, edit your <filename role="home">~/.hgrc</filename> file, and add the lines below.</para> <programlisting>[extensions] hgext.mq =</programlisting> <para id="x_3cb">Once the extension is enabled, it will make a number of new commands available. To verify that the extension is working, you can use <command role="hg-cmd">hg help</command> to see if the <command role="hg-ext-mq">qinit</command> command is now available.</para> &interaction.mq.qinit-help.help; <para id="x_3cc">You can use MQ with <emphasis>any</emphasis> Mercurial repository, and its commands only operate within that repository. To get started, simply prepare the repository using the <command role="hg-ext-mq">qinit</command> command.</para> &interaction.mq.tutorial.qinit; <para id="x_3cd">This command creates an empty directory called <filename role="special" class="directory">.hg/patches</filename>, where MQ will keep its metadata. As with many Mercurial commands, the <command role="hg-ext-mq">qinit</command> command prints nothing if it succeeds.</para> <sect2> <title>Creating a new patch</title> <para id="x_3ce">To begin work on a new patch, use the <command role="hg-ext-mq">qnew</command> command. This command takes one argument, the name of the patch to create.</para> <para id="x_3cf">MQ will use this as the name of an actual file in the <filename role="special" class="directory">.hg/patches</filename> directory, as you can see below.</para> &interaction.mq.tutorial.qnew; <para id="x_3d0">Also newly present in the <filename role="special" class="directory">.hg/patches</filename> directory are two other files, <filename role="special">series</filename> and <filename role="special">status</filename>. The <filename role="special">series</filename> file lists all of the patches that MQ knows about for this repository, with one patch per line. Mercurial uses the <filename role="special">status</filename> file for internal book-keeping; it tracks all of the patches that MQ has <emphasis>applied</emphasis> in this repository.</para> <note> <para id="x_3d1"> You may sometimes want to edit the <filename role="special">series</filename> file by hand; for example, to change the sequence in which some patches are applied. However, manually editing the <filename role="special">status</filename> file is almost always a bad idea, as it's easy to corrupt MQ's idea of what is happening.</para> </note> <para id="x_3d2">Once you have created your new patch, you can edit files in the working directory as you usually would. All of the normal Mercurial commands, such as <command role="hg-cmd">hg diff</command> and <command role="hg-cmd">hg annotate</command>, work exactly as they did before.</para> </sect2> <sect2> <title>Refreshing a patch</title> <para id="x_3d3">When you reach a point where you want to save your work, use the <command role="hg-ext-mq">qrefresh</command> command to update the patch you are working on.</para> &interaction.mq.tutorial.qrefresh; <para id="x_3d4">This command folds the changes you have made in the working directory into your patch, and updates its corresponding changeset to contain those changes.</para> <para id="x_3d5">You can run <command role="hg-ext-mq">qrefresh</command> as often as you like, so it's a good way to <quote>checkpoint</quote> your work. Refresh your patch at an opportune time; try an experiment; and if the experiment doesn't work out, <command role="hg-cmd">hg revert</command> your modifications back to the last time you refreshed.</para> &interaction.mq.tutorial.qrefresh2; </sect2> <sect2> <title>Stacking and tracking patches</title> <para id="x_3d6">Once you have finished working on a patch, or need to work on another, you can use the <command role="hg-ext-mq">qnew</command> command again to create a new patch. Mercurial will apply this patch on top of your existing patch.</para> &interaction.mq.tutorial.qnew2; <para id="x_3d7">Notice that the patch contains the changes in our prior patch as part of its context (you can see this more clearly in the output of <command role="hg-cmd">hg annotate</command>).</para> <para id="x_3d8">So far, with the exception of <command role="hg-ext-mq">qnew</command> and <command role="hg-ext-mq">qrefresh</command>, we've been careful to only use regular Mercurial commands. However, MQ provides many commands that are easier to use when you are thinking about patches, as illustrated below.</para> &interaction.mq.tutorial.qseries; <itemizedlist> <listitem><para id="x_3d9">The <command role="hg-ext-mq">qseries</command> command lists every patch that MQ knows about in this repository, from oldest to newest (most recently <emphasis>created</emphasis>).</para> </listitem> <listitem><para id="x_3da">The <command role="hg-ext-mq">qapplied</command> command lists every patch that MQ has <emphasis>applied</emphasis> in this repository, again from oldest to newest (most recently applied).</para> </listitem></itemizedlist> </sect2> <sect2> <title>Manipulating the patch stack</title> <para id="x_3db">The previous discussion implied that there must be a difference between <quote>known</quote> and <quote>applied</quote> patches, and there is. MQ can manage a patch without it being applied in the repository.</para> <para id="x_3dc">An <emphasis>applied</emphasis> patch has a corresponding changeset in the repository, and the effects of the patch and changeset are visible in the working directory. You can undo the application of a patch using the <command role="hg-ext-mq">qpop</command> command. MQ still <emphasis>knows about</emphasis>, or manages, a popped patch, but the patch no longer has a corresponding changeset in the repository, and the working directory does not contain the changes made by the patch. <xref linkend="fig:mq:stack"/> illustrates the difference between applied and tracked patches.</para> <figure id="fig:mq:stack"> <title>Applied and unapplied patches in the MQ patch stack</title> <mediaobject> <imageobject><imagedata fileref="figs/mq-stack.png"/></imageobject> <textobject><phrase>XXX add text</phrase></textobject> </mediaobject> </figure> <para id="x_3de">You can reapply an unapplied, or popped, patch using the <command role="hg-ext-mq">qpush</command> command. This creates a new changeset to correspond to the patch, and the patch's changes once again become present in the working directory. See below for examples of <command role="hg-ext-mq">qpop</command> and <command role="hg-ext-mq">qpush</command> in action.</para> &interaction.mq.tutorial.qpop; <para id="x_3df">Notice that once we have popped a patch or two patches, the output of <command role="hg-ext-mq">qseries</command> remains the same, while that of <command role="hg-ext-mq">qapplied</command> has changed.</para> </sect2> <sect2> <title>Pushing and popping many patches</title> <para id="x_3e0">While <command role="hg-ext-mq">qpush</command> and <command role="hg-ext-mq">qpop</command> each operate on a single patch at a time by default, you can push and pop many patches in one go. The <option role="hg-ext-mq-cmd-qpush-opt">hg -a</option> option to <command role="hg-ext-mq">qpush</command> causes it to push all unapplied patches, while the <option role="hg-ext-mq-cmd-qpop-opt">-a</option> option to <command role="hg-ext-mq">qpop</command> causes it to pop all applied patches. (For some more ways to push and pop many patches, see <xref linkend="sec:mq:perf"/> below.)</para> &interaction.mq.tutorial.qpush-a; </sect2> <sect2> <title>Safety checks, and overriding them</title> <para id="x_3e1">Several MQ commands check the working directory before they do anything, and fail if they find any modifications. They do this to ensure that you won't lose any changes that you have made, but not yet incorporated into a patch. The example below illustrates this; the <command role="hg-ext-mq">qnew</command> command will not create a new patch if there are outstanding changes, caused in this case by the <command role="hg-cmd">hg add</command> of <filename>file3</filename>.</para> &interaction.mq.tutorial.add; <para id="x_3e2">Commands that check the working directory all take an <quote>I know what I'm doing</quote> option, which is always named <option>-f</option>. The exact meaning of <option>-f</option> depends on the command. For example, <command role="hg-cmd">hg qnew <option role="hg-ext-mq-cmd-qnew-opt">hg -f</option></command> will incorporate any outstanding changes into the new patch it creates, but <command role="hg-cmd">hg qpop <option role="hg-ext-mq-cmd-qpop-opt">hg -f</option></command> will revert modifications to any files affected by the patch that it is popping. Be sure to read the documentation for a command's <option>-f</option> option before you use it!</para> </sect2> <sect2> <title>Working on several patches at once</title> <para id="x_3e3">The <command role="hg-ext-mq">qrefresh</command> command always refreshes the <emphasis>topmost</emphasis> applied patch. This means that you can suspend work on one patch (by refreshing it), pop or push to make a different patch the top, and work on <emphasis>that</emphasis> patch for a while.</para> <para id="x_3e4">Here's an example that illustrates how you can use this ability. Let's say you're developing a new feature as two patches. The first is a change to the core of your software, and the second&emdash;layered on top of the first&emdash;changes the user interface to use the code you just added to the core. If you notice a bug in the core while you're working on the UI patch, it's easy to fix the core. Simply <command role="hg-ext-mq">qrefresh</command> the UI patch to save your in-progress changes, and <command role="hg-ext-mq">qpop</command> down to the core patch. Fix the core bug, <command role="hg-ext-mq">qrefresh</command> the core patch, and <command role="hg-ext-mq">qpush</command> back to the UI patch to continue where you left off.</para> </sect2> </sect1> <sect1 id="sec:mq:adv-patch"> <title>More about patches</title> <para id="x_3e5">MQ uses the GNU <command>patch</command> command to apply patches, so it's helpful to know a few more detailed aspects of how <command>patch</command> works, and about patches themselves.</para> <sect2> <title>The strip count</title> <para id="x_3e6">If you look at the file headers in a patch, you will notice that the pathnames usually have an extra component on the front that isn't present in the actual path name. This is a holdover from the way that people used to generate patches (people still do this, but it's somewhat rare with modern revision control tools).</para> <para id="x_3e7">Alice would unpack a tarball, edit her files, then decide that she wanted to create a patch. So she'd rename her working directory, unpack the tarball again (hence the need for the rename), and use the <option role="cmd-opt-diff">-r</option> and <option role="cmd-opt-diff">-N</option> options to <command>diff</command> to recursively generate a patch between the unmodified directory and the modified one. The result would be that the name of the unmodified directory would be at the front of the left-hand path in every file header, and the name of the modified directory would be at the front of the right-hand path.</para> <para id="x_3e8">Since someone receiving a patch from the Alices of the net would be unlikely to have unmodified and modified directories with exactly the same names, the <command>patch</command> command has a <option role="cmd-opt-patch">-p</option> option that indicates the number of leading path name components to strip when trying to apply a patch. This number is called the <emphasis>strip count</emphasis>.</para> <para id="x_3e9">An option of <quote><literal>-p1</literal></quote> means <quote>use a strip count of one</quote>. If <command>patch</command> sees a file name <filename>foo/bar/baz</filename> in a file header, it will strip <filename>foo</filename> and try to patch a file named <filename>bar/baz</filename>. (Strictly speaking, the strip count refers to the number of <emphasis>path separators</emphasis> (and the components that go with them ) to strip. A strip count of one will turn <filename>foo/bar</filename> into <filename>bar</filename>, but <filename>/foo/bar</filename> (notice the extra leading slash) into <filename>foo/bar</filename>.)</para> <para id="x_3ea">The <quote>standard</quote> strip count for patches is one; almost all patches contain one leading path name component that needs to be stripped. Mercurial's <command role="hg-cmd">hg diff</command> command generates path names in this form, and the <command role="hg-cmd">hg import</command> command and MQ expect patches to have a strip count of one.</para> <para id="x_3eb">If you receive a patch from someone that you want to add to your patch queue, and the patch needs a strip count other than one, you cannot just <command role="hg-ext-mq">qimport</command> the patch, because <command role="hg-ext-mq">qimport</command> does not yet have a <literal>-p</literal> option (see <ulink role="hg-bug" url="http://www.selenic.com/mercurial/bts/issue311">issue 311</ulink>). Your best bet is to <command role="hg-ext-mq">qnew</command> a patch of your own, then use <command>patch -pN</command> to apply their patch, followed by <command role="hg-cmd">hg addremove</command> to pick up any files added or removed by the patch, followed by <command role="hg-ext-mq">hg qrefresh</command>. This complexity may become unnecessary; see <ulink role="hg-bug" url="http://www.selenic.com/mercurial/bts/issue311">issue 311</ulink> for details. </para> </sect2> <sect2> <title>Strategies for applying a patch</title> <para id="x_3ec">When <command>patch</command> applies a hunk, it tries a handful of successively less accurate strategies to try to make the hunk apply. This falling-back technique often makes it possible to take a patch that was generated against an old version of a file, and apply it against a newer version of that file.</para> <para id="x_3ed">First, <command>patch</command> tries an exact match, where the line numbers, the context, and the text to be modified must apply exactly. If it cannot make an exact match, it tries to find an exact match for the context, without honouring the line numbering information. If this succeeds, it prints a line of output saying that the hunk was applied, but at some <emphasis>offset</emphasis> from the original line number.</para> <para id="x_3ee">If a context-only match fails, <command>patch</command> removes the first and last lines of the context, and tries a <emphasis>reduced</emphasis> context-only match. If the hunk with reduced context succeeds, it prints a message saying that it applied the hunk with a <emphasis>fuzz factor</emphasis> (the number after the fuzz factor indicates how many lines of context <command>patch</command> had to trim before the patch applied).</para> <para id="x_3ef">When neither of these techniques works, <command>patch</command> prints a message saying that the hunk in question was rejected. It saves rejected hunks (also simply called <quote>rejects</quote>) to a file with the same name, and an added <filename role="special">.rej</filename> extension. It also saves an unmodified copy of the file with a <filename role="special">.orig</filename> extension; the copy of the file without any extensions will contain any changes made by hunks that <emphasis>did</emphasis> apply cleanly. If you have a patch that modifies <filename>foo</filename> with six hunks, and one of them fails to apply, you will have: an unmodified <filename>foo.orig</filename>, a <filename>foo.rej</filename> containing one hunk, and <filename>foo</filename>, containing the changes made by the five successful hunks.</para> </sect2> <sect2> <title>Some quirks of patch representation</title> <para id="x_3f0">There are a few useful things to know about how <command>patch</command> works with files.</para> <itemizedlist> <listitem><para id="x_3f1">This should already be obvious, but <command>patch</command> cannot handle binary files.</para> </listitem> <listitem><para id="x_3f2">Neither does it care about the executable bit; it creates new files as readable, but not executable.</para> </listitem> <listitem><para id="x_3f3"><command>patch</command> treats the removal of a file as a diff between the file to be removed and the empty file. So your idea of <quote>I deleted this file</quote> looks like <quote>every line of this file was deleted</quote> in a patch.</para> </listitem> <listitem><para id="x_3f4">It treats the addition of a file as a diff between the empty file and the file to be added. So in a patch, your idea of <quote>I added this file</quote> looks like <quote>every line of this file was added</quote>.</para> </listitem> <listitem><para id="x_3f5">It treats a renamed file as the removal of the old name, and the addition of the new name. This means that renamed files have a big footprint in patches. (Note also that Mercurial does not currently try to infer when files have been renamed or copied in a patch.)</para> </listitem> <listitem><para id="x_3f6"><command>patch</command> cannot represent empty files, so you cannot use a patch to represent the notion <quote>I added this empty file to the tree</quote>.</para> </listitem></itemizedlist> </sect2> <sect2> <title>Beware the fuzz</title> <para id="x_3f7">While applying a hunk at an offset, or with a fuzz factor, will often be completely successful, these inexact techniques naturally leave open the possibility of corrupting the patched file. The most common cases typically involve applying a patch twice, or at an incorrect location in the file. If <command>patch</command> or <command role="hg-ext-mq">qpush</command> ever mentions an offset or fuzz factor, you should make sure that the modified files are correct afterwards.</para> <para id="x_3f8">It's often a good idea to refresh a patch that has applied with an offset or fuzz factor; refreshing the patch generates new context information that will make it apply cleanly. I say <quote>often,</quote> not <quote>always,</quote> because sometimes refreshing a patch will make it fail to apply against a different revision of the underlying files. In some cases, such as when you're maintaining a patch that must sit on top of multiple versions of a source tree, it's acceptable to have a patch apply with some fuzz, provided you've verified the results of the patching process in such cases.</para> </sect2> <sect2> <title>Handling rejection</title> <para id="x_3f9">If <command role="hg-ext-mq">qpush</command> fails to apply a patch, it will print an error message and exit. If it has left <filename role="special">.rej</filename> files behind, it is usually best to fix up the rejected hunks before you push more patches or do any further work.</para> <para id="x_3fa">If your patch <emphasis>used to</emphasis> apply cleanly, and no longer does because you've changed the underlying code that your patches are based on, Mercurial Queues can help; see <xref linkend="sec:mq:merge"/> for details.</para> <para id="x_3fb">Unfortunately, there aren't any great techniques for dealing with rejected hunks. Most often, you'll need to view the <filename role="special">.rej</filename> file and edit the target file, applying the rejected hunks by hand.</para> <para id="x_3fc">If you're feeling adventurous, Neil Brown, a Linux kernel hacker, wrote a tool called <command>wiggle</command> <citation>web:wiggle</citation>, which is more vigorous than <command>patch</command> in its attempts to make a patch apply.</para> <para id="x_3fd">Another Linux kernel hacker, Chris Mason (the author of Mercurial Queues), wrote a similar tool called <command>mpatch</command> <citation>web:mpatch</citation>, which takes a simple approach to automating the application of hunks rejected by <command>patch</command>. The <command>mpatch</command> command can help with four common reasons that a hunk may be rejected:</para> <itemizedlist> <listitem><para id="x_3fe">The context in the middle of a hunk has changed.</para> </listitem> <listitem><para id="x_3ff">A hunk is missing some context at the beginning or end.</para> </listitem> <listitem><para id="x_400">A large hunk might apply better&emdash;either entirely or in part&emdash;if it was broken up into smaller hunks.</para> </listitem> <listitem><para id="x_401">A hunk removes lines with slightly different content than those currently present in the file.</para> </listitem></itemizedlist> <para id="x_402">If you use <command>wiggle</command> or <command>mpatch</command>, you should be doubly careful to check your results when you're done. In fact, <command>mpatch</command> enforces this method of double-checking the tool's output, by automatically dropping you into a merge program when it has done its job, so that you can verify its work and finish off any remaining merges.</para> </sect2> </sect1> <sect1 id="sec:mq:perf"> <title>Getting the best performance out of MQ</title> <para id="x_403">MQ is very efficient at handling a large number of patches. I ran some performance experiments in mid-2006 for a talk that I gave at the 2006 EuroPython conference <citation>web:europython</citation>. I used as my data set the Linux 2.6.17-mm1 patch series, which consists of 1,738 patches. I applied these on top of a Linux kernel repository containing all 27,472 revisions between Linux 2.6.12-rc2 and Linux 2.6.17.</para> <para id="x_404">On my old, slow laptop, I was able to <command role="hg-cmd">hg qpush <option role="hg-ext-mq-cmd-qpush-opt">hg -a</option></command> all 1,738 patches in 3.5 minutes, and <command role="hg-cmd">hg qpop <option role="hg-ext-mq-cmd-qpop-opt">hg -a</option></command> them all in 30 seconds. (On a newer laptop, the time to push all patches dropped to two minutes.) I could <command role="hg-ext-mq">qrefresh</command> one of the biggest patches (which made 22,779 lines of changes to 287 files) in 6.6 seconds.</para> <para id="x_405">Clearly, MQ is well suited to working in large trees, but there are a few tricks you can use to get the best performance of it.</para> <para id="x_406">First of all, try to <quote>batch</quote> operations together. Every time you run <command role="hg-ext-mq">qpush</command> or <command role="hg-ext-mq">qpop</command>, these commands scan the working directory once to make sure you haven't made some changes and then forgotten to run <command role="hg-ext-mq">qrefresh</command>. On a small tree, the time that this scan takes is unnoticeable. However, on a medium-sized tree (containing tens of thousands of files), it can take a second or more.</para> <para id="x_407">The <command role="hg-ext-mq">qpush</command> and <command role="hg-ext-mq">qpop</command> commands allow you to push and pop multiple patches at a time. You can identify the <quote>destination patch</quote> that you want to end up at. When you <command role="hg-ext-mq">qpush</command> with a destination specified, it will push patches until that patch is at the top of the applied stack. When you <command role="hg-ext-mq">qpop</command> to a destination, MQ will pop patches until the destination patch is at the top.</para> <para id="x_408">You can identify a destination patch using either the name of the patch, or by number. If you use numeric addressing, patches are counted from zero; this means that the first patch is zero, the second is one, and so on.</para> </sect1> <sect1 id="sec:mq:merge"> <title>Updating your patches when the underlying code changes</title> <para id="x_409">It's common to have a stack of patches on top of an underlying repository that you don't modify directly. If you're working on changes to third-party code, or on a feature that is taking longer to develop than the rate of change of the code beneath, you will often need to sync up with the underlying code, and fix up any hunks in your patches that no longer apply. This is called <emphasis>rebasing</emphasis> your patch series.</para> <para id="x_40a">The simplest way to do this is to <command role="hg-cmd">hg qpop <option role="hg-ext-mq-cmd-qpop-opt">hg -a</option></command> your patches, then <command role="hg-cmd">hg pull</command> changes into the underlying repository, and finally <command role="hg-cmd">hg qpush <option role="hg-ext-mq-cmd-qpop-opt">hg -a</option></command> your patches again. MQ will stop pushing any time it runs across a patch that fails to apply during conflicts, allowing you to fix your conflicts, <command role="hg-ext-mq">qrefresh</command> the affected patch, and continue pushing until you have fixed your entire stack.</para> <para id="x_40b">This approach is easy to use and works well if you don't expect changes to the underlying code to affect how well your patches apply. If your patch stack touches code that is modified frequently or invasively in the underlying repository, however, fixing up rejected hunks by hand quickly becomes tiresome.</para> <para id="x_40c">It's possible to partially automate the rebasing process. If your patches apply cleanly against some revision of the underlying repo, MQ can use this information to help you to resolve conflicts between your patches and a different revision.</para> <para id="x_40d">The process is a little involved.</para> <orderedlist> <listitem><para id="x_40e">To begin, <command role="hg-cmd">hg qpush -a</command> all of your patches on top of the revision where you know that they apply cleanly.</para> </listitem> <listitem><para id="x_40f">Save a backup copy of your patch directory using <command role="hg-cmd">hg qsave <option role="hg-ext-mq-cmd-qsave-opt">hg -e</option> <option role="hg-ext-mq-cmd-qsave-opt">hg -c</option></command>. This prints the name of the directory that it has saved the patches in. It will save the patches to a directory called <filename role="special" class="directory">.hg/patches.N</filename>, where <literal>N</literal> is a small integer. It also commits a <quote>save changeset</quote> on top of your applied patches; this is for internal book-keeping, and records the states of the <filename role="special">series</filename> and <filename role="special">status</filename> files.</para> </listitem> <listitem><para id="x_410">Use <command role="hg-cmd">hg pull</command> to bring new changes into the underlying repository. (Don't run <command role="hg-cmd">hg pull -u</command>; see below for why.)</para> </listitem> <listitem><para id="x_411">Update to the new tip revision, using <command role="hg-cmd">hg update <option role="hg-opt-update">-C</option></command> to override the patches you have pushed.</para> </listitem> <listitem><para id="x_412">Merge all patches using <command>hg qpush -m -a</command>. The <option role="hg-ext-mq-cmd-qpush-opt">-m</option> option to <command role="hg-ext-mq">qpush</command> tells MQ to perform a three-way merge if the patch fails to apply.</para> </listitem></orderedlist> <para id="x_413">During the <command role="hg-cmd">hg qpush <option role="hg-ext-mq-cmd-qpush-opt">hg -m</option></command>, each patch in the <filename role="special">series</filename> file is applied normally. If a patch applies with fuzz or rejects, MQ looks at the queue you <command role="hg-ext-mq">qsave</command>d, and performs a three-way merge with the corresponding changeset. This merge uses Mercurial's normal merge machinery, so it may pop up a GUI merge tool to help you to resolve problems.</para> <para id="x_414">When you finish resolving the effects of a patch, MQ refreshes your patch based on the result of the merge.</para> <para id="x_415">At the end of this process, your repository will have one extra head from the old patch queue, and a copy of the old patch queue will be in <filename role="special" class="directory">.hg/patches.N</filename>. You can remove the extra head using <command role="hg-cmd">hg qpop -a -n patches.N</command> or <command role="hg-cmd">hg strip</command>. You can delete <filename role="special" class="directory">.hg/patches.N</filename> once you are sure that you no longer need it as a backup.</para> </sect1> <sect1> <title>Identifying patches</title> <para id="x_416">MQ commands that work with patches let you refer to a patch either by using its name or by a number. By name is obvious enough; pass the name <filename>foo.patch</filename> to <command role="hg-ext-mq">qpush</command>, for example, and it will push patches until <filename>foo.patch</filename> is applied.</para> <para id="x_417">As a shortcut, you can refer to a patch using both a name and a numeric offset; <literal>foo.patch-2</literal> means <quote>two patches before <literal>foo.patch</literal></quote>, while <literal>bar.patch+4</literal> means <quote>four patches after <literal>bar.patch</literal></quote>.</para> <para id="x_418">Referring to a patch by index isn't much different. The first patch printed in the output of <command role="hg-ext-mq">qseries</command> is patch zero (yes, it's one of those start-at-zero counting systems); the second is patch one; and so on.</para> <para id="x_419">MQ also makes it easy to work with patches when you are using normal Mercurial commands. Every command that accepts a changeset ID will also accept the name of an applied patch. MQ augments the tags normally in the repository with an eponymous one for each applied patch. In addition, the special tags <literal role="tag">qbase</literal> and <literal role="tag">qtip</literal> identify the <quote>bottom-most</quote> and topmost applied patches, respectively.</para> <para id="x_41a">These additions to Mercurial's normal tagging capabilities make dealing with patches even more of a breeze.</para> <itemizedlist> <listitem><para id="x_41b">Want to patchbomb a mailing list with your latest series of changes?</para> <programlisting>hg email qbase:qtip</programlisting> <para id="x_41c"> (Don't know what <quote>patchbombing</quote> is? See <xref linkend="sec:hgext:patchbomb"/>.)</para> </listitem> <listitem><para id="x_41d">Need to see all of the patches since <literal>foo.patch</literal> that have touched files in a subdirectory of your tree?</para> <programlisting>hg log -r foo.patch:qtip subdir</programlisting> </listitem> </itemizedlist> <para id="x_41e">Because MQ makes the names of patches available to the rest of Mercurial through its normal internal tag machinery, you don't need to type in the entire name of a patch when you want to identify it by name.</para> <para id="x_41f">Another nice consequence of representing patch names as tags is that when you run the <command role="hg-cmd">hg log</command> command, it will display a patch's name as a tag, simply as part of its normal output. This makes it easy to visually distinguish applied patches from underlying <quote>normal</quote> revisions. The following example shows a few normal Mercurial commands in use with applied patches.</para> &interaction.mq.id.output; </sect1> <sect1> <title>Useful things to know about</title> <para id="x_420">There are a number of aspects of MQ usage that don't fit tidily into sections of their own, but that are good to know. Here they are, in one place.</para> <itemizedlist> <listitem><para id="x_421">Normally, when you <command role="hg-ext-mq">qpop</command> a patch and <command role="hg-ext-mq">qpush</command> it again, the changeset that represents the patch after the pop/push will have a <emphasis>different identity</emphasis> than the changeset that represented the hash beforehand. See <xref linkend="sec:mqref:cmd:qpush"/> for information as to why this is.</para> </listitem> <listitem><para id="x_422">It's not a good idea to <command role="hg-cmd">hg merge</command> changes from another branch with a patch changeset, at least if you want to maintain the <quote>patchiness</quote> of that changeset and changesets below it on the patch stack. If you try to do this, it will appear to succeed, but MQ will become confused.</para> </listitem></itemizedlist> </sect1> <sect1 id="sec:mq:repo"> <title>Managing patches in a repository</title> <para id="x_423">Because MQ's <filename role="special" class="directory">.hg/patches</filename> directory resides outside a Mercurial repository's working directory, the <quote>underlying</quote> Mercurial repository knows nothing about the management or presence of patches.</para> <para id="x_424">This presents the interesting possibility of managing the contents of the patch directory as a Mercurial repository in its own right. This can be a useful way to work. For example, you can work on a patch for a while, <command role="hg-ext-mq">qrefresh</command> it, then <command role="hg-cmd">hg commit</command> the current state of the patch. This lets you <quote>roll back</quote> to that version of the patch later on.</para> <para id="x_425">You can then share different versions of the same patch stack among multiple underlying repositories. I use this when I am developing a Linux kernel feature. I have a pristine copy of my kernel sources for each of several CPU architectures, and a cloned repository under each that contains the patches I am working on. When I want to test a change on a different architecture, I push my current patches to the patch repository associated with that kernel tree, pop and push all of my patches, and build and test that kernel.</para> <para id="x_426">Managing patches in a repository makes it possible for multiple developers to work on the same patch series without colliding with each other, all on top of an underlying source base that they may or may not control.</para> <sect2> <title>MQ support for patch repositories</title> <para id="x_427">MQ helps you to work with the <filename role="special" class="directory">.hg/patches</filename> directory as a repository; when you prepare a repository for working with patches using <command role="hg-ext-mq">qinit</command>, you can pass the <option role="hg-ext-mq-cmd-qinit-opt">hg -c</option> option to create the <filename role="special" class="directory">.hg/patches</filename> directory as a Mercurial repository.</para> <note> <para id="x_428"> If you forget to use the <option role="hg-ext-mq-cmd-qinit-opt">hg -c</option> option, you can simply go into the <filename role="special" class="directory">.hg/patches</filename> directory at any time and run <command role="hg-cmd">hg init</command>. Don't forget to add an entry for the <filename role="special">status</filename> file to the <filename role="special">.hgignore</filename> file, though</para> <para id="x_429"> (<command role="hg-cmd">hg qinit <option role="hg-ext-mq-cmd-qinit-opt">hg -c</option></command> does this for you automatically); you <emphasis>really</emphasis> don't want to manage the <filename role="special">status</filename> file.</para> </note> <para id="x_42a">As a convenience, if MQ notices that the <filename class="directory">.hg/patches</filename> directory is a repository, it will automatically <command role="hg-cmd">hg add</command> every patch that you create and import.</para> <para id="x_42b">MQ provides a shortcut command, <command role="hg-ext-mq">qcommit</command>, that runs <command role="hg-cmd">hg commit</command> in the <filename role="special" class="directory">.hg/patches</filename> directory. This saves some bothersome typing.</para> <para id="x_42c">Finally, as a convenience to manage the patch directory, you can define the alias <command>mq</command> on Unix systems. For example, on Linux systems using the <command>bash</command> shell, you can include the following snippet in your <filename role="home">~/.bashrc</filename>.</para> <programlisting>alias mq=`hg -R $(hg root)/.hg/patches'</programlisting> <para id="x_42d">You can then issue commands of the form <command>mq pull</command> from the main repository.</para> </sect2> <sect2> <title>A few things to watch out for</title> <para id="x_42e">MQ's support for working with a repository full of patches is limited in a few small respects.</para> <para id="x_42f">MQ cannot automatically detect changes that you make to the patch directory. If you <command role="hg-cmd">hg pull</command>, manually edit, or <command role="hg-cmd">hg update</command> changes to patches or the <filename role="special">series</filename> file, you will have to <command role="hg-cmd">hg qpop <option role="hg-ext-mq-cmd-qpop-opt">hg -a</option></command> and then <command role="hg-cmd">hg qpush <option role="hg-ext-mq-cmd-qpush-opt">hg -a</option></command> in the underlying repository to see those changes show up there. If you forget to do this, you can confuse MQ's idea of which patches are applied.</para> </sect2> </sect1> <sect1 id="sec:mq:tools"> <title>Third party tools for working with patches</title> <para id="x_430">Once you've been working with patches for a while, you'll find yourself hungry for tools that will help you to understand and manipulate the patches you're dealing with.</para> <para id="x_431">The <command>diffstat</command> command <citation>web:diffstat</citation> generates a histogram of the modifications made to each file in a patch. It provides a good way to <quote>get a sense of</quote> a patch&emdash;which files it affects, and how much change it introduces to each file and as a whole. (I find that it's a good idea to use <command>diffstat</command>'s <option role="cmd-opt-diffstat">-p</option> option as a matter of course, as otherwise it will try to do clever things with prefixes of file names that inevitably confuse at least me.)</para> &interaction.mq.tools.tools; <para id="x_432">The <literal role="package">patchutils</literal> package <citation>web:patchutils</citation> is invaluable. It provides a set of small utilities that follow the <quote>Unix philosophy;</quote> each does one useful thing with a patch. The <literal role="package">patchutils</literal> command I use most is <command>filterdiff</command>, which extracts subsets from a patch file. For example, given a patch that modifies hundreds of files across dozens of directories, a single invocation of <command>filterdiff</command> can generate a smaller patch that only touches files whose names match a particular glob pattern. See <xref linkend="mq-collab:tips:interdiff"/> for another example.</para> </sect1> <sect1> <title>Good ways to work with patches</title> <para id="x_433">Whether you are working on a patch series to submit to a free software or open source project, or a series that you intend to treat as a sequence of regular changesets when you're done, you can use some simple techniques to keep your work well organized.</para> <para id="x_434">Give your patches descriptive names. A good name for a patch might be <filename>rework-device-alloc.patch</filename>, because it will immediately give you a hint what the purpose of the patch is. Long names shouldn't be a problem; you won't be typing the names often, but you <emphasis>will</emphasis> be running commands like <command role="hg-ext-mq">qapplied</command> and <command role="hg-ext-mq">qtop</command> over and over. Good naming becomes especially important when you have a number of patches to work with, or if you are juggling a number of different tasks and your patches only get a fraction of your attention.</para> <para id="x_435">Be aware of what patch you're working on. Use the <command role="hg-ext-mq">qtop</command> command and skim over the text of your patches frequently&emdash;for example, using <command role="hg-cmd">hg tip <option role="hg-opt-tip">-p</option></command>)&emdash;to be sure of where you stand. I have several times worked on and <command role="hg-ext-mq">qrefresh</command>ed a patch other than the one I intended, and it's often tricky to migrate changes into the right patch after making them in the wrong one.</para> <para id="x_436">For this reason, it is very much worth investing a little time to learn how to use some of the third-party tools I described in <xref linkend="sec:mq:tools"/>, particularly <command>diffstat</command> and <command>filterdiff</command>. The former will give you a quick idea of what changes your patch is making, while the latter makes it easy to splice hunks selectively out of one patch and into another.</para> </sect1> <sect1> <title>MQ cookbook</title> <sect2> <title>Manage <quote>trivial</quote> patches</title> <para id="x_437">Because the overhead of dropping files into a new Mercurial repository is so low, it makes a lot of sense to manage patches this way even if you simply want to make a few changes to a source tarball that you downloaded.</para> <para id="x_438">Begin by downloading and unpacking the source tarball, and turning it into a Mercurial repository.</para> &interaction.mq.tarball.download; <para id="x_439">Continue by creating a patch stack and making your changes.</para> &interaction.mq.tarball.qinit; <para id="x_43a">Let's say a few weeks or months pass, and your package author releases a new version. First, bring their changes into the repository.</para> &interaction.mq.tarball.newsource; <para id="x_43b">The pipeline starting with <command role="hg-cmd">hg locate</command> above deletes all files in the working directory, so that <command role="hg-cmd">hg commit</command>'s <option role="hg-opt-commit">--addremove</option> option can actually tell which files have really been removed in the newer version of the source.</para> <para id="x_43c">Finally, you can apply your patches on top of the new tree.</para> &interaction.mq.tarball.repush; </sect2> <sect2 id="sec:mq:combine"> <title>Combining entire patches</title> <para id="x_43d">MQ provides a command, <command role="hg-ext-mq">qfold</command> that lets you combine entire patches. This <quote>folds</quote> the patches you name, in the order you name them, into the topmost applied patch, and concatenates their descriptions onto the end of its description. The patches that you fold must be unapplied before you fold them.</para> <para id="x_43e">The order in which you fold patches matters. If your topmost applied patch is <literal>foo</literal>, and you <command role="hg-ext-mq">qfold</command> <literal>bar</literal> and <literal>quux</literal> into it, you will end up with a patch that has the same effect as if you applied first <literal>foo</literal>, then <literal>bar</literal>, followed by <literal>quux</literal>.</para> </sect2> <sect2> <title>Merging part of one patch into another</title> <para id="x_43f">Merging <emphasis>part</emphasis> of one patch into another is more difficult than combining entire patches.</para> <para id="x_440">If you want to move changes to entire files, you can use <command>filterdiff</command>'s <option role="cmd-opt-filterdiff">-i</option> and <option role="cmd-opt-filterdiff">-x</option> options to choose the modifications to snip out of one patch, concatenating its output onto the end of the patch you want to merge into. You usually won't need to modify the patch you've merged the changes from. Instead, MQ will report some rejected hunks when you <command role="hg-ext-mq">qpush</command> it (from the hunks you moved into the other patch), and you can simply <command role="hg-ext-mq">qrefresh</command> the patch to drop the duplicate hunks.</para> <para id="x_441">If you have a patch that has multiple hunks modifying a file, and you only want to move a few of those hunks, the job becomes more messy, but you can still partly automate it. Use <command>lsdiff -nvv</command> to print some metadata about the patch.</para> &interaction.mq.tools.lsdiff; <para id="x_442">This command prints three different kinds of number:</para> <itemizedlist> <listitem><para id="x_443">(in the first column) a <emphasis>file number</emphasis> to identify each file modified in the patch;</para> </listitem> <listitem><para id="x_444">(on the next line, indented) the line number within a modified file where a hunk starts; and</para> </listitem> <listitem><para id="x_445">(on the same line) a <emphasis>hunk number</emphasis> to identify that hunk.</para> </listitem></itemizedlist> <para id="x_446">You'll have to use some visual inspection, and reading of the patch, to identify the file and hunk numbers you'll want, but you can then pass them to to <command>filterdiff</command>'s <option role="cmd-opt-filterdiff">--files</option> and <option role="cmd-opt-filterdiff">--hunks</option> options, to select exactly the file and hunk you want to extract.</para> <para id="x_447">Once you have this hunk, you can concatenate it onto the end of your destination patch and continue with the remainder of <xref linkend="sec:mq:combine"/>.</para> </sect2> </sect1> <sect1> <title>Differences between quilt and MQ</title> <para id="x_448">If you are already familiar with quilt, MQ provides a similar command set. There are a few differences in the way that it works.</para> <para id="x_449">You will already have noticed that most quilt commands have MQ counterparts that simply begin with a <quote><literal>q</literal></quote>. The exceptions are quilt's <literal>add</literal> and <literal>remove</literal> commands, the counterparts for which are the normal Mercurial <command role="hg-cmd">hg add</command> and <command role="hg-cmd">hg remove</command> commands. There is no MQ equivalent of the quilt <literal>edit</literal> command.</para> </sect1> </chapter> <!-- local variables: sgml-parent-document: ("00book.xml" "book" "chapter") end: -->