hgbook
view en/ch12-mq.xml @ 560:dbe91bb622d8
Small grammatical fix (does not requires -> does not require)
| author | Jon Parise <jon@indelible.org> |
|---|---|
| date | Thu Feb 19 20:49:46 2009 -0800 (2009-02-19) |
| parents | 8631da51309b |
| children | 8fcd44708f41 |
line source
1 <!-- vim: set filetype=docbkxml shiftwidth=2 autoindent expandtab tw=77 : -->
3 <chapter id="chap:mq">
4 <title>Managing change with Mercurial Queues</title>
6 <sect1 id="sec:mq:patch-mgmt">
7 <title>The patch management problem</title>
9 <para>Here is a common scenario: you need to install a software
10 package from source, but you find a bug that you must fix in the
11 source before you can start using the package. You make your
12 changes, forget about the package for a while, and a few months
13 later you need to upgrade to a newer version of the package. If
14 the newer version of the package still has the bug, you must
15 extract your fix from the older source tree and apply it against
16 the newer version. This is a tedious task, and it's easy to
17 make mistakes.</para>
19 <para>This is a simple case of the <quote>patch management</quote>
20 problem. You have an <quote>upstream</quote> source tree that
21 you can't change; you need to make some local changes on top of
22 the upstream tree; and you'd like to be able to keep those
23 changes separate, so that you can apply them to newer versions
24 of the upstream source.</para>
26 <para>The patch management problem arises in many situations.
27 Probably the most visible is that a user of an open source
28 software project will contribute a bug fix or new feature to the
29 project's maintainers in the form of a patch.</para>
31 <para>Distributors of operating systems that include open source
32 software often need to make changes to the packages they
33 distribute so that they will build properly in their
34 environments.</para>
36 <para>When you have few changes to maintain, it is easy to manage
37 a single patch using the standard <command>diff</command> and
38 <command>patch</command> programs (see section <xref
39 linkend="sec:mq:patch"/> for a discussion of these
40 tools). Once the number of changes grows, it starts to make
41 sense to maintain patches as discrete <quote>chunks of
42 work,</quote> so that for example a single patch will contain
43 only one bug fix (the patch might modify several files, but it's
44 doing <quote>only one thing</quote>), and you may have a number
45 of such patches for different bugs you need fixed and local
46 changes you require. In this situation, if you submit a bug fix
47 patch to the upstream maintainers of a package and they include
48 your fix in a subsequent release, you can simply drop that
49 single patch when you're updating to the newer release.</para>
51 <para>Maintaining a single patch against an upstream tree is a
52 little tedious and error-prone, but not difficult. However, the
53 complexity of the problem grows rapidly as the number of patches
54 you have to maintain increases. With more than a tiny number of
55 patches in hand, understanding which ones you have applied and
56 maintaining them moves from messy to overwhelming.</para>
58 <para>Fortunately, Mercurial includes a powerful extension,
59 Mercurial Queues (or simply <quote>MQ</quote>), that massively
60 simplifies the patch management problem.</para>
62 </sect1>
63 <sect1 id="sec:mq:history">
64 <title>The prehistory of Mercurial Queues</title>
66 <para>During the late 1990s, several Linux kernel developers
67 started to maintain <quote>patch series</quote> that modified
68 the behaviour of the Linux kernel. Some of these series were
69 focused on stability, some on feature coverage, and others were
70 more speculative.</para>
72 <para>The sizes of these patch series grew rapidly. In 2002,
73 Andrew Morton published some shell scripts he had been using to
74 automate the task of managing his patch queues. Andrew was
75 successfully using these scripts to manage hundreds (sometimes
76 thousands) of patches on top of the Linux kernel.</para>
78 <sect2 id="sec:mq:quilt">
79 <title>A patchwork quilt</title>
81 <para>In early 2003, Andreas Gruenbacher and Martin Quinson
82 borrowed the approach of Andrew's scripts and published a tool
83 called <quote>patchwork quilt</quote>
84 <citation>web:quilt</citation>, or simply <quote>quilt</quote>
85 (see <citation>gruenbacher:2005</citation> for a paper
86 describing it). Because quilt substantially automated patch
87 management, it rapidly gained a large following among open
88 source software developers.</para>
90 <para>Quilt manages a <emphasis>stack of patches</emphasis> on
91 top of a directory tree. To begin, you tell quilt to manage a
92 directory tree, and tell it which files you want to manage; it
93 stores away the names and contents of those files. To fix a
94 bug, you create a new patch (using a single command), edit the
95 files you need to fix, then <quote>refresh</quote> the
96 patch.</para>
98 <para>The refresh step causes quilt to scan the directory tree;
99 it updates the patch with all of the changes you have made.
100 You can create another patch on top of the first, which will
101 track the changes required to modify the tree from <quote>tree
102 with one patch applied</quote> to <quote>tree with two
103 patches applied</quote>.</para>
105 <para>You can <emphasis>change</emphasis> which patches are
106 applied to the tree. If you <quote>pop</quote> a patch, the
107 changes made by that patch will vanish from the directory
108 tree. Quilt remembers which patches you have popped, though,
109 so you can <quote>push</quote> a popped patch again, and the
110 directory tree will be restored to contain the modifications
111 in the patch. Most importantly, you can run the
112 <quote>refresh</quote> command at any time, and the topmost
113 applied patch will be updated. This means that you can, at
114 any time, change both which patches are applied and what
115 modifications those patches make.</para>
117 <para>Quilt knows nothing about revision control tools, so it
118 works equally well on top of an unpacked tarball or a
119 Subversion working copy.</para>
121 </sect2>
122 <sect2 id="sec:mq:quilt-mq">
123 <title>From patchwork quilt to Mercurial Queues</title>
125 <para>In mid-2005, Chris Mason took the features of quilt and
126 wrote an extension that he called Mercurial Queues, which
127 added quilt-like behaviour to Mercurial.</para>
129 <para>The key difference between quilt and MQ is that quilt
130 knows nothing about revision control systems, while MQ is
131 <emphasis>integrated</emphasis> into Mercurial. Each patch
132 that you push is represented as a Mercurial changeset. Pop a
133 patch, and the changeset goes away.</para>
135 <para>Because quilt does not care about revision control tools,
136 it is still a tremendously useful piece of software to know
137 about for situations where you cannot use Mercurial and
138 MQ.</para>
140 </sect2>
141 </sect1>
142 <sect1>
143 <title>The huge advantage of MQ</title>
145 <para>I cannot overstate the value that MQ offers through the
146 unification of patches and revision control.</para>
148 <para>A major reason that patches have persisted in the free
149 software and open source world&emdash;in spite of the
150 availability of increasingly capable revision control tools over
151 the years&emdash;is the <emphasis>agility</emphasis> they
152 offer.</para>
154 <para>Traditional revision control tools make a permanent,
155 irreversible record of everything that you do. While this has
156 great value, it's also somewhat stifling. If you want to
157 perform a wild-eyed experiment, you have to be careful in how
158 you go about it, or you risk leaving unneeded&emdash;or worse,
159 misleading or destabilising&emdash;traces of your missteps and
160 errors in the permanent revision record.</para>
162 <para>By contrast, MQ's marriage of distributed revision control
163 with patches makes it much easier to isolate your work. Your
164 patches live on top of normal revision history, and you can make
165 them disappear or reappear at will. If you don't like a patch,
166 you can drop it. If a patch isn't quite as you want it to be,
167 simply fix it&emdash;as many times as you need to, until you
168 have refined it into the form you desire.</para>
170 <para>As an example, the integration of patches with revision
171 control makes understanding patches and debugging their
172 effects&emdash;and their interplay with the code they're based
173 on&emdash;<emphasis>enormously</emphasis> easier. Since every
174 applied patch has an associated changeset, you can give <command
175 role="hg-cmd">hg log</command> a file name to see which
176 changesets and patches affected the file. You can use the
177 <command role="hg-cmd">hg bisect</command> command to
178 binary-search through all changesets and applied patches to see
179 where a bug got introduced or fixed. You can use the <command
180 role="hg-cmd">hg annotate</command> command to see which
181 changeset or patch modified a particular line of a source file.
182 And so on.</para>
184 </sect1>
185 <sect1 id="sec:mq:patch">
186 <title>Understanding patches</title>
188 <para>Because MQ doesn't hide its patch-oriented nature, it is
189 helpful to understand what patches are, and a little about the
190 tools that work with them.</para>
192 <para>The traditional Unix <command>diff</command> command
193 compares two files, and prints a list of differences between
194 them. The <command>patch</command> command understands these
195 differences as <emphasis>modifications</emphasis> to make to a
196 file. Take a look below for a simple example of these commands
197 in action.</para>
199 <!-- &interaction.mq.dodiff.diff; -->
201 <para>The type of file that <command>diff</command> generates (and
202 <command>patch</command> takes as input) is called a
203 <quote>patch</quote> or a <quote>diff</quote>; there is no
204 difference between a patch and a diff. (We'll use the term
205 <quote>patch</quote>, since it's more commonly used.)</para>
207 <para>A patch file can start with arbitrary text; the
208 <command>patch</command> command ignores this text, but MQ uses
209 it as the commit message when creating changesets. To find the
210 beginning of the patch content, <command>patch</command>
211 searches for the first line that starts with the string
212 <quote><literal>diff -</literal></quote>.</para>
214 <para>MQ works with <emphasis>unified</emphasis> diffs
215 (<command>patch</command> can accept several other diff formats,
216 but MQ doesn't). A unified diff contains two kinds of header.
217 The <emphasis>file header</emphasis> describes the file being
218 modified; it contains the name of the file to modify. When
219 <command>patch</command> sees a new file header, it looks for a
220 file with that name to start modifying.</para>
222 <para>After the file header comes a series of
223 <emphasis>hunks</emphasis>. Each hunk starts with a header;
224 this identifies the range of line numbers within the file that
225 the hunk should modify. Following the header, a hunk starts and
226 ends with a few (usually three) lines of text from the
227 unmodified file; these are called the
228 <emphasis>context</emphasis> for the hunk. If there's only a
229 small amount of context between successive hunks,
230 <command>diff</command> doesn't print a new hunk header; it just
231 runs the hunks together, with a few lines of context between
232 modifications.</para>
234 <para>Each line of context begins with a space character. Within
235 the hunk, a line that begins with
236 <quote><literal>-</literal></quote> means <quote>remove this
237 line,</quote> while a line that begins with
238 <quote><literal>+</literal></quote> means <quote>insert this
239 line.</quote> For example, a line that is modified is
240 represented by one deletion and one insertion.</para>
242 <para>We will return to some of the more subtle aspects of patches
243 later (in section <xref linkend="sec:mq:adv-patch"/>), but you
244 should have
245 enough information now to use MQ.</para>
247 </sect1>
248 <sect1 id="sec:mq:start">
249 <title>Getting started with Mercurial Queues</title>
251 <para>Because MQ is implemented as an extension, you must
252 explicitly enable before you can use it. (You don't need to
253 download anything; MQ ships with the standard Mercurial
254 distribution.) To enable MQ, edit your <filename
255 role="home">~/.hgrc</filename> file, and add the lines
256 below.</para>
258 <programlisting>[extensions] hgext.mq =</programlisting>
260 <para>Once the extension is enabled, it will make a number of new
261 commands available. To verify that the extension is working,
262 you can use <command role="hg-cmd">hg help</command> to see if
263 the <command role="hg-ext-mq">qinit</command> command is now
264 available.</para>
266 <!-- &interaction.mq.qinit-help.help; -->
268 <para>You can use MQ with <emphasis>any</emphasis> Mercurial
269 repository, and its commands only operate within that
270 repository. To get started, simply prepare the repository using
271 the <command role="hg-ext-mq">qinit</command> command.</para>
273 <!-- &interaction.mq.tutorial.qinit; -->
275 <para>This command creates an empty directory called <filename
276 role="special" class="directory">.hg/patches</filename>, where
277 MQ will keep its metadata. As with many Mercurial commands, the
278 <command role="hg-ext-mq">qinit</command> command prints nothing
279 if it succeeds.</para>
281 <sect2>
282 <title>Creating a new patch</title>
284 <para>To begin work on a new patch, use the <command
285 role="hg-ext-mq">qnew</command> command. This command takes
286 one argument, the name of the patch to create.</para>
288 <para>MQ will use this as the name of an actual file in the
289 <filename role="special"
290 class="directory">.hg/patches</filename> directory, as you
291 can see below.</para>
293 <!-- &interaction.mq.tutorial.qnew; -->
295 <para>Also newly present in the <filename role="special"
296 class="directory">.hg/patches</filename> directory are two
297 other files, <filename role="special">series</filename> and
298 <filename role="special">status</filename>. The <filename
299 role="special">series</filename> file lists all of the
300 patches that MQ knows about for this repository, with one
301 patch per line. Mercurial uses the <filename
302 role="special">status</filename> file for internal
303 book-keeping; it tracks all of the patches that MQ has
304 <emphasis>applied</emphasis> in this repository.</para>
306 <note>
307 <para> You may sometimes want to edit the <filename
308 role="special">series</filename> file by hand; for
309 example, to change the sequence in which some patches are
310 applied. However, manually editing the <filename
311 role="special">status</filename> file is almost always a
312 bad idea, as it's easy to corrupt MQ's idea of what is
313 happening.</para>
314 </note>
316 <para>Once you have created your new patch, you can edit files
317 in the working directory as you usually would. All of the
318 normal Mercurial commands, such as <command role="hg-cmd">hg
319 diff</command> and <command role="hg-cmd">hg
320 annotate</command>, work exactly as they did before.</para>
322 </sect2>
323 <sect2>
324 <title>Refreshing a patch</title>
326 <para>When you reach a point where you want to save your work,
327 use the <command role="hg-ext-mq">qrefresh</command> command
328 to update the patch you are working on.</para>
330 <!-- &interaction.mq.tutorial.qrefresh; -->
332 <para>This command folds the changes you have made in the
333 working directory into your patch, and updates its
334 corresponding changeset to contain those changes.</para>
336 <para>You can run <command role="hg-ext-mq">qrefresh</command>
337 as often as you like, so it's a good way to
338 <quote>checkpoint</quote> your work. Refresh your patch at an
339 opportune time; try an experiment; and if the experiment
340 doesn't work out, <command role="hg-cmd">hg revert</command>
341 your modifications back to the last time you refreshed.</para>
343 <!-- &interaction.mq.tutorial.qrefresh2; -->
345 </sect2>
346 <sect2>
347 <title>Stacking and tracking patches</title>
349 <para>Once you have finished working on a patch, or need to work
350 on another, you can use the <command
351 role="hg-ext-mq">qnew</command> command again to create a
352 new patch. Mercurial will apply this patch on top of your
353 existing patch.</para>
355 <!-- &interaction.mq.tutorial.qnew2; -->
356 <para>Notice that the patch contains the changes in our prior
357 patch as part of its context (you can see this more clearly in
358 the output of <command role="hg-cmd">hg
359 annotate</command>).</para>
361 <para>So far, with the exception of <command
362 role="hg-ext-mq">qnew</command> and <command
363 role="hg-ext-mq">qrefresh</command>, we've been careful to
364 only use regular Mercurial commands. However, MQ provides
365 many commands that are easier to use when you are thinking
366 about patches, as illustrated below.</para>
368 <!-- &interaction.mq.tutorial.qseries; -->
370 <itemizedlist>
371 <listitem><para>The <command
372 role="hg-ext-mq">qseries</command> command lists every
373 patch that MQ knows about in this repository, from oldest
374 to newest (most recently
375 <emphasis>created</emphasis>).</para>
376 </listitem>
377 <listitem><para>The <command
378 role="hg-ext-mq">qapplied</command> command lists every
379 patch that MQ has <emphasis>applied</emphasis> in this
380 repository, again from oldest to newest (most recently
381 applied).</para>
382 </listitem></itemizedlist>
384 </sect2>
385 <sect2>
386 <title>Manipulating the patch stack</title>
388 <para>The previous discussion implied that there must be a
389 difference between <quote>known</quote> and
390 <quote>applied</quote> patches, and there is. MQ can manage a
391 patch without it being applied in the repository.</para>
393 <para>An <emphasis>applied</emphasis> patch has a corresponding
394 changeset in the repository, and the effects of the patch and
395 changeset are visible in the working directory. You can undo
396 the application of a patch using the <command
397 role="hg-ext-mq">qpop</command> command. MQ still
398 <emphasis>knows about</emphasis>, or manages, a popped patch,
399 but the patch no longer has a corresponding changeset in the
400 repository, and the working directory does not contain the
401 changes made by the patch. Figure <xref
402 linkend="fig:mq:stack"/> illustrates
403 the difference between applied and tracked patches.</para>
405 <informalfigure id="fig:mq:stack">
406 <mediaobject><imageobject><imagedata
407 fileref="mq-stack"/></imageobject><textobject><phrase>XXX
408 add text</phrase></textobject><caption><para>Applied and
409 unapplied patches in the MQ patch
410 stack</para></caption></mediaobject>
411 </informalfigure>
413 <para>You can reapply an unapplied, or popped, patch using the
414 <command role="hg-ext-mq">qpush</command> command. This
415 creates a new changeset to correspond to the patch, and the
416 patch's changes once again become present in the working
417 directory. See below for examples of <command
418 role="hg-ext-mq">qpop</command> and <command
419 role="hg-ext-mq">qpush</command> in action.</para>
420 <!-- &interaction.mq.tutorial.qpop; -->
422 <para>Notice that once we have popped a patch or two patches,
423 the output of <command role="hg-ext-mq">qseries</command>
424 remains the same, while that of <command
425 role="hg-ext-mq">qapplied</command> has changed.</para>
428 </sect2>
429 <sect2>
430 <title>Pushing and popping many patches</title>
432 <para>While <command role="hg-ext-mq">qpush</command> and
433 <command role="hg-ext-mq">qpop</command> each operate on a
434 single patch at a time by default, you can push and pop many
435 patches in one go. The <option
436 role="hg-ext-mq-cmd-qpush-opt">hg -a</option> option to
437 <command role="hg-ext-mq">qpush</command> causes it to push
438 all unapplied patches, while the <option
439 role="hg-ext-mq-cmd-qpop-opt">-a</option> option to <command
440 role="hg-ext-mq">qpop</command> causes it to pop all applied
441 patches. (For some more ways to push and pop many patches,
442 see section <xref linkend="sec:mq:perf"/>
443 below.)</para>
445 <!-- &interaction.mq.tutorial.qpush-a; -->
447 </sect2>
448 <sect2>
449 <title>Safety checks, and overriding them</title>
451 <para>Several MQ commands check the working directory before
452 they do anything, and fail if they find any modifications.
453 They do this to ensure that you won't lose any changes that
454 you have made, but not yet incorporated into a patch. The
455 example below illustrates this; the <command
456 role="hg-ext-mq">qnew</command> command will not create a
457 new patch if there are outstanding changes, caused in this
458 case by the <command role="hg-cmd">hg add</command> of
459 <filename>file3</filename>.</para>
461 <!-- &interaction.mq.tutorial.add; -->
463 <para>Commands that check the working directory all take an
464 <quote>I know what I'm doing</quote> option, which is always
465 named <option>-f</option>. The exact meaning of
466 <option>-f</option> depends on the command. For example,
467 <command role="hg-cmd">hg qnew <option
468 role="hg-ext-mq-cmd-qnew-opt">hg -f</option></command>
469 will incorporate any outstanding changes into the new patch it
470 creates, but <command role="hg-cmd">hg qpop <option
471 role="hg-ext-mq-cmd-qpop-opt">hg -f</option></command>
472 will revert modifications to any files affected by the patch
473 that it is popping. Be sure to read the documentation for a
474 command's <option>-f</option> option before you use it!</para>
476 </sect2>
477 <sect2>
478 <title>Working on several patches at once</title>
480 <para>The <command role="hg-ext-mq">qrefresh</command> command
481 always refreshes the <emphasis>topmost</emphasis> applied
482 patch. This means that you can suspend work on one patch (by
483 refreshing it), pop or push to make a different patch the top,
484 and work on <emphasis>that</emphasis> patch for a
485 while.</para>
487 <para>Here's an example that illustrates how you can use this
488 ability. Let's say you're developing a new feature as two
489 patches. The first is a change to the core of your software,
490 and the second&emdash;layered on top of the
491 first&emdash;changes the user interface to use the code you
492 just added to the core. If you notice a bug in the core while
493 you're working on the UI patch, it's easy to fix the core.
494 Simply <command role="hg-ext-mq">qrefresh</command> the UI
495 patch to save your in-progress changes, and <command
496 role="hg-ext-mq">qpop</command> down to the core patch. Fix
497 the core bug, <command role="hg-ext-mq">qrefresh</command> the
498 core patch, and <command role="hg-ext-mq">qpush</command> back
499 to the UI patch to continue where you left off.</para>
501 </sect2>
502 </sect1>
503 <sect1 id="sec:mq:adv-patch">
504 <title>More about patches</title>
506 <para>MQ uses the GNU <command>patch</command> command to apply
507 patches, so it's helpful to know a few more detailed aspects of
508 how <command>patch</command> works, and about patches
509 themselves.</para>
511 <sect2>
512 <title>The strip count</title>
514 <para>If you look at the file headers in a patch, you will
515 notice that the pathnames usually have an extra component on
516 the front that isn't present in the actual path name. This is
517 a holdover from the way that people used to generate patches
518 (people still do this, but it's somewhat rare with modern
519 revision control tools).</para>
521 <para>Alice would unpack a tarball, edit her files, then decide
522 that she wanted to create a patch. So she'd rename her
523 working directory, unpack the tarball again (hence the need
524 for the rename), and use the <option
525 role="cmd-opt-diff">-r</option> and <option
526 role="cmd-opt-diff">-N</option> options to
527 <command>diff</command> to recursively generate a patch
528 between the unmodified directory and the modified one. The
529 result would be that the name of the unmodified directory
530 would be at the front of the left-hand path in every file
531 header, and the name of the modified directory would be at the
532 front of the right-hand path.</para>
534 <para>Since someone receiving a patch from the Alices of the net
535 would be unlikely to have unmodified and modified directories
536 with exactly the same names, the <command>patch</command>
537 command has a <option role="cmd-opt-patch">-p</option> option
538 that indicates the number of leading path name components to
539 strip when trying to apply a patch. This number is called the
540 <emphasis>strip count</emphasis>.</para>
542 <para>An option of <quote><literal>-p1</literal></quote> means
543 <quote>use a strip count of one</quote>. If
544 <command>patch</command> sees a file name
545 <filename>foo/bar/baz</filename> in a file header, it will
546 strip <filename>foo</filename> and try to patch a file named
547 <filename>bar/baz</filename>. (Strictly speaking, the strip
548 count refers to the number of <emphasis>path
549 separators</emphasis> (and the components that go with them
550 ) to strip. A strip count of one will turn
551 <filename>foo/bar</filename> into <filename>bar</filename>,
552 but <filename>/foo/bar</filename> (notice the extra leading
553 slash) into <filename>foo/bar</filename>.)</para>
555 <para>The <quote>standard</quote> strip count for patches is
556 one; almost all patches contain one leading path name
557 component that needs to be stripped. Mercurial's <command
558 role="hg-cmd">hg diff</command> command generates path names
559 in this form, and the <command role="hg-cmd">hg
560 import</command> command and MQ expect patches to have a
561 strip count of one.</para>
563 <para>If you receive a patch from someone that you want to add
564 to your patch queue, and the patch needs a strip count other
565 than one, you cannot just <command
566 role="hg-ext-mq">qimport</command> the patch, because
567 <command role="hg-ext-mq">qimport</command> does not yet have
568 a <literal>-p</literal> option (see <ulink role="hg-bug"
569 url="http://www.selenic.com/mercurial/bts/issue311">issue
570 311</ulink>). Your best bet is to <command
571 role="hg-ext-mq">qnew</command> a patch of your own, then
572 use <command>patch -pN</command> to apply their patch,
573 followed by <command role="hg-cmd">hg addremove</command> to
574 pick up any files added or removed by the patch, followed by
575 <command role="hg-ext-mq">hg qrefresh</command>. This
576 complexity may become unnecessary; see <ulink role="hg-bug"
577 url="http://www.selenic.com/mercurial/bts/issue311">issue
578 311</ulink> for details.
579 </para>
580 </sect2>
581 <sect2>
582 <title>Strategies for applying a patch</title>
584 <para>When <command>patch</command> applies a hunk, it tries a
585 handful of successively less accurate strategies to try to
586 make the hunk apply. This falling-back technique often makes
587 it possible to take a patch that was generated against an old
588 version of a file, and apply it against a newer version of
589 that file.</para>
591 <para>First, <command>patch</command> tries an exact match,
592 where the line numbers, the context, and the text to be
593 modified must apply exactly. If it cannot make an exact
594 match, it tries to find an exact match for the context,
595 without honouring the line numbering information. If this
596 succeeds, it prints a line of output saying that the hunk was
597 applied, but at some <emphasis>offset</emphasis> from the
598 original line number.</para>
600 <para>If a context-only match fails, <command>patch</command>
601 removes the first and last lines of the context, and tries a
602 <emphasis>reduced</emphasis> context-only match. If the hunk
603 with reduced context succeeds, it prints a message saying that
604 it applied the hunk with a <emphasis>fuzz factor</emphasis>
605 (the number after the fuzz factor indicates how many lines of
606 context <command>patch</command> had to trim before the patch
607 applied).</para>
609 <para>When neither of these techniques works,
610 <command>patch</command> prints a message saying that the hunk
611 in question was rejected. It saves rejected hunks (also
612 simply called <quote>rejects</quote>) to a file with the same
613 name, and an added <filename role="special">.rej</filename>
614 extension. It also saves an unmodified copy of the file with
615 a <filename role="special">.orig</filename> extension; the
616 copy of the file without any extensions will contain any
617 changes made by hunks that <emphasis>did</emphasis> apply
618 cleanly. If you have a patch that modifies
619 <filename>foo</filename> with six hunks, and one of them fails
620 to apply, you will have: an unmodified
621 <filename>foo.orig</filename>, a <filename>foo.rej</filename>
622 containing one hunk, and <filename>foo</filename>, containing
623 the changes made by the five successful hunks.</para>
625 </sect2>
626 <sect2>
627 <title>Some quirks of patch representation</title>
629 <para>There are a few useful things to know about how
630 <command>patch</command> works with files.</para>
631 <itemizedlist>
632 <listitem><para>This should already be obvious, but
633 <command>patch</command> cannot handle binary
634 files.</para>
635 </listitem>
636 <listitem><para>Neither does it care about the executable bit;
637 it creates new files as readable, but not
638 executable.</para>
639 </listitem>
640 <listitem><para><command>patch</command> treats the removal of
641 a file as a diff between the file to be removed and the
642 empty file. So your idea of <quote>I deleted this
643 file</quote> looks like <quote>every line of this file
644 was deleted</quote> in a patch.</para>
645 </listitem>
646 <listitem><para>It treats the addition of a file as a diff
647 between the empty file and the file to be added. So in a
648 patch, your idea of <quote>I added this file</quote> looks
649 like <quote>every line of this file was
650 added</quote>.</para>
651 </listitem>
652 <listitem><para>It treats a renamed file as the removal of the
653 old name, and the addition of the new name. This means
654 that renamed files have a big footprint in patches. (Note
655 also that Mercurial does not currently try to infer when
656 files have been renamed or copied in a patch.)</para>
657 </listitem>
658 <listitem><para><command>patch</command> cannot represent
659 empty files, so you cannot use a patch to represent the
660 notion <quote>I added this empty file to the
661 tree</quote>.</para>
662 </listitem></itemizedlist>
663 </sect2>
664 <sect2>
665 <title>Beware the fuzz</title>
667 <para>While applying a hunk at an offset, or with a fuzz factor,
668 will often be completely successful, these inexact techniques
669 naturally leave open the possibility of corrupting the patched
670 file. The most common cases typically involve applying a
671 patch twice, or at an incorrect location in the file. If
672 <command>patch</command> or <command
673 role="hg-ext-mq">qpush</command> ever mentions an offset or
674 fuzz factor, you should make sure that the modified files are
675 correct afterwards.</para>
677 <para>It's often a good idea to refresh a patch that has applied
678 with an offset or fuzz factor; refreshing the patch generates
679 new context information that will make it apply cleanly. I
680 say <quote>often,</quote> not <quote>always,</quote> because
681 sometimes refreshing a patch will make it fail to apply
682 against a different revision of the underlying files. In some
683 cases, such as when you're maintaining a patch that must sit
684 on top of multiple versions of a source tree, it's acceptable
685 to have a patch apply with some fuzz, provided you've verified
686 the results of the patching process in such cases.</para>
688 </sect2>
689 <sect2>
690 <title>Handling rejection</title>
692 <para>If <command role="hg-ext-mq">qpush</command> fails to
693 apply a patch, it will print an error message and exit. If it
694 has left <filename role="special">.rej</filename> files
695 behind, it is usually best to fix up the rejected hunks before
696 you push more patches or do any further work.</para>
698 <para>If your patch <emphasis>used to</emphasis> apply cleanly,
699 and no longer does because you've changed the underlying code
700 that your patches are based on, Mercurial Queues can help; see
701 section <xref
702 linkend="sec:mq:merge"/> for details.</para>
704 <para>Unfortunately, there aren't any great techniques for
705 dealing with rejected hunks. Most often, you'll need to view
706 the <filename role="special">.rej</filename> file and edit the
707 target file, applying the rejected hunks by hand.</para>
709 <para>If you're feeling adventurous, Neil Brown, a Linux kernel
710 hacker, wrote a tool called <command>wiggle</command>
711 <citation>web:wiggle</citation>, which is more vigorous than
712 <command>patch</command> in its attempts to make a patch
713 apply.</para>
715 <para>Another Linux kernel hacker, Chris Mason (the author of
716 Mercurial Queues), wrote a similar tool called
717 <command>mpatch</command> <citation>web:mpatch</citation>,
718 which takes a simple approach to automating the application of
719 hunks rejected by <command>patch</command>. The
720 <command>mpatch</command> command can help with four common
721 reasons that a hunk may be rejected:</para>
723 <itemizedlist>
724 <listitem><para>The context in the middle of a hunk has
725 changed.</para>
726 </listitem>
727 <listitem><para>A hunk is missing some context at the
728 beginning or end.</para>
729 </listitem>
730 <listitem><para>A large hunk might apply better&emdash;either
731 entirely or in part&emdash;if it was broken up into
732 smaller hunks.</para>
733 </listitem>
734 <listitem><para>A hunk removes lines with slightly different
735 content than those currently present in the file.</para>
736 </listitem></itemizedlist>
738 <para>If you use <command>wiggle</command> or
739 <command>mpatch</command>, you should be doubly careful to
740 check your results when you're done. In fact,
741 <command>mpatch</command> enforces this method of
742 double-checking the tool's output, by automatically dropping
743 you into a merge program when it has done its job, so that you
744 can verify its work and finish off any remaining
745 merges.</para>
747 </sect2>
748 </sect1>
749 <sect1 id="sec:mq:perf">
750 <title>Getting the best performance out of MQ</title>
752 <para>MQ is very efficient at handling a large number of patches.
753 I ran some performance experiments in mid-2006 for a talk that I
754 gave at the 2006 EuroPython conference
755 <citation>web:europython</citation>. I used as my data set the
756 Linux 2.6.17-mm1 patch series, which consists of 1,738 patches.
757 I applied these on top of a Linux kernel repository containing
758 all 27,472 revisions between Linux 2.6.12-rc2 and Linux
759 2.6.17.</para>
761 <para>On my old, slow laptop, I was able to <command
762 role="hg-cmd">hg qpush <option
763 role="hg-ext-mq-cmd-qpush-opt">hg -a</option></command> all
764 1,738 patches in 3.5 minutes, and <command role="hg-cmd">hg qpop
765 <option role="hg-ext-mq-cmd-qpop-opt">hg -a</option></command>
766 them all in 30 seconds. (On a newer laptop, the time to push
767 all patches dropped to two minutes.) I could <command
768 role="hg-ext-mq">qrefresh</command> one of the biggest patches
769 (which made 22,779 lines of changes to 287 files) in 6.6
770 seconds.</para>
772 <para>Clearly, MQ is well suited to working in large trees, but
773 there are a few tricks you can use to get the best performance
774 of it.</para>
776 <para>First of all, try to <quote>batch</quote> operations
777 together. Every time you run <command
778 role="hg-ext-mq">qpush</command> or <command
779 role="hg-ext-mq">qpop</command>, these commands scan the
780 working directory once to make sure you haven't made some
781 changes and then forgotten to run <command
782 role="hg-ext-mq">qrefresh</command>. On a small tree, the
783 time that this scan takes is unnoticeable. However, on a
784 medium-sized tree (containing tens of thousands of files), it
785 can take a second or more.</para>
787 <para>The <command role="hg-ext-mq">qpush</command> and <command
788 role="hg-ext-mq">qpop</command> commands allow you to push and
789 pop multiple patches at a time. You can identify the
790 <quote>destination patch</quote> that you want to end up at.
791 When you <command role="hg-ext-mq">qpush</command> with a
792 destination specified, it will push patches until that patch is
793 at the top of the applied stack. When you <command
794 role="hg-ext-mq">qpop</command> to a destination, MQ will pop
795 patches until the destination patch is at the top.</para>
797 <para>You can identify a destination patch using either the name
798 of the patch, or by number. If you use numeric addressing,
799 patches are counted from zero; this means that the first patch
800 is zero, the second is one, and so on.</para>
802 </sect1>
803 <sect1 id="sec:mq:merge">
804 <title>Updating your patches when the underlying code
805 changes</title>
807 <para>It's common to have a stack of patches on top of an
808 underlying repository that you don't modify directly. If you're
809 working on changes to third-party code, or on a feature that is
810 taking longer to develop than the rate of change of the code
811 beneath, you will often need to sync up with the underlying
812 code, and fix up any hunks in your patches that no longer apply.
813 This is called <emphasis>rebasing</emphasis> your patch
814 series.</para>
816 <para>The simplest way to do this is to <command role="hg-cmd">hg
817 qpop <option role="hg-ext-mq-cmd-qpop-opt">hg
818 -a</option></command> your patches, then <command
819 role="hg-cmd">hg pull</command> changes into the underlying
820 repository, and finally <command role="hg-cmd">hg qpush <option
821 role="hg-ext-mq-cmd-qpop-opt">hg -a</option></command> your
822 patches again. MQ will stop pushing any time it runs across a
823 patch that fails to apply during conflicts, allowing you to fix
824 your conflicts, <command role="hg-ext-mq">qrefresh</command> the
825 affected patch, and continue pushing until you have fixed your
826 entire stack.</para>
828 <para>This approach is easy to use and works well if you don't
829 expect changes to the underlying code to affect how well your
830 patches apply. If your patch stack touches code that is modified
831 frequently or invasively in the underlying repository, however,
832 fixing up rejected hunks by hand quickly becomes
833 tiresome.</para>
835 <para>It's possible to partially automate the rebasing process.
836 If your patches apply cleanly against some revision of the
837 underlying repo, MQ can use this information to help you to
838 resolve conflicts between your patches and a different
839 revision.</para>
841 <para>The process is a little involved.</para>
842 <orderedlist>
843 <listitem><para>To begin, <command role="hg-cmd">hg qpush
844 -a</command> all of your patches on top of the revision
845 where you know that they apply cleanly.</para>
846 </listitem>
847 <listitem><para>Save a backup copy of your patch directory using
848 <command role="hg-cmd">hg qsave <option
849 role="hg-ext-mq-cmd-qsave-opt">hg -e</option> <option
850 role="hg-ext-mq-cmd-qsave-opt">hg -c</option></command>.
851 This prints the name of the directory that it has saved the
852 patches in. It will save the patches to a directory called
853 <filename role="special"
854 class="directory">.hg/patches.N</filename>, where
855 <literal>N</literal> is a small integer. It also commits a
856 <quote>save changeset</quote> on top of your applied
857 patches; this is for internal book-keeping, and records the
858 states of the <filename role="special">series</filename> and
859 <filename role="special">status</filename> files.</para>
860 </listitem>
861 <listitem><para>Use <command role="hg-cmd">hg pull</command> to
862 bring new changes into the underlying repository. (Don't
863 run <command role="hg-cmd">hg pull -u</command>; see below
864 for why.)</para>
865 </listitem>
866 <listitem><para>Update to the new tip revision, using <command
867 role="hg-cmd">hg update <option
868 role="hg-opt-update">-C</option></command> to override
869 the patches you have pushed.</para>
870 </listitem>
871 <listitem><para>Merge all patches using
872 \hgcmdargs{qpush}{<option role="hg-ext-mq-cmd-qpush-opt">hg
873 -m</option> <option role="hg-ext-mq-cmd-qpush-opt">hg
874 -a</option>}. The <option
875 role="hg-ext-mq-cmd-qpush-opt">hg -m</option> option to
876 <command role="hg-ext-mq">qpush</command> tells MQ to
877 perform a three-way merge if the patch fails to
878 apply.</para>
879 </listitem></orderedlist>
881 <para>During the <command role="hg-cmd">hg qpush <option
882 role="hg-ext-mq-cmd-qpush-opt">hg -m</option></command>,
883 each patch in the <filename role="special">series</filename>
884 file is applied normally. If a patch applies with fuzz or
885 rejects, MQ looks at the queue you <command
886 role="hg-ext-mq">qsave</command>d, and performs a three-way
887 merge with the corresponding changeset. This merge uses
888 Mercurial's normal merge machinery, so it may pop up a GUI merge
889 tool to help you to resolve problems.</para>
891 <para>When you finish resolving the effects of a patch, MQ
892 refreshes your patch based on the result of the merge.</para>
894 <para>At the end of this process, your repository will have one
895 extra head from the old patch queue, and a copy of the old patch
896 queue will be in <filename role="special"
897 class="directory">.hg/patches.N</filename>. You can remove the
898 extra head using <command role="hg-cmd">hg qpop -a -n
899 patches.N</command> or <command role="hg-cmd">hg
900 strip</command>. You can delete <filename role="special"
901 class="directory">.hg/patches.N</filename> once you are sure
902 that you no longer need it as a backup.</para>
904 </sect1>
905 <sect1>
906 <title>Identifying patches</title>
908 <para>MQ commands that work with patches let you refer to a patch
909 either by using its name or by a number. By name is obvious
910 enough; pass the name <filename>foo.patch</filename> to <command
911 role="hg-ext-mq">qpush</command>, for example, and it will
912 push patches until <filename>foo.patch</filename> is
913 applied.</para>
915 <para>As a shortcut, you can refer to a patch using both a name
916 and a numeric offset; <literal>foo.patch-2</literal> means
917 <quote>two patches before <literal>foo.patch</literal></quote>,
918 while <literal>bar.patch+4</literal> means <quote>four patches
919 after <literal>bar.patch</literal></quote>.</para>
921 <para>Referring to a patch by index isn't much different. The
922 first patch printed in the output of <command
923 role="hg-ext-mq">qseries</command> is patch zero (yes, it's
924 one of those start-at-zero counting systems); the second is
925 patch one; and so on.</para>
927 <para>MQ also makes it easy to work with patches when you are
928 using normal Mercurial commands. Every command that accepts a
929 changeset ID will also accept the name of an applied patch. MQ
930 augments the tags normally in the repository with an eponymous
931 one for each applied patch. In addition, the special tags
932 \index{tags!special tag
933 names!<literal>qbase</literal>}<literal>qbase</literal> and
934 \index{tags!special tag
935 names!<literal>qtip</literal>}<literal>qtip</literal> identify
936 the <quote>bottom-most</quote> and topmost applied patches,
937 respectively.</para>
939 <para>These additions to Mercurial's normal tagging capabilities
940 make dealing with patches even more of a breeze.</para>
941 <itemizedlist>
942 <listitem><para>Want to patchbomb a mailing list with your
943 latest series of changes?</para>
944 <programlisting>hg email qbase:qtip
945 </programlisting>
946 <para> (Don't know what <quote>patchbombing</quote> is? See
947 section <xref linkend="sec:hgext:patchbomb"/>.)</para>
948 </listitem>
949 <listitem><para>Need to see all of the patches since
950 <literal>foo.patch</literal> that have touched files in a
951 subdirectory of your tree?</para>
952 <programlisting>
953 hg log -r foo.patch:qtip <emphasis>subdir</emphasis>
954 </programlisting>
955 </listitem>
956 </itemizedlist>
958 <para>Because MQ makes the names of patches available to the rest
959 of Mercurial through its normal internal tag machinery, you
960 don't need to type in the entire name of a patch when you want
961 to identify it by name.</para>
963 <para>Another nice consequence of representing patch names as tags
964 is that when you run the <command role="hg-cmd">hg log</command>
965 command, it will display a patch's name as a tag, simply as part
966 of its normal output. This makes it easy to visually
967 distinguish applied patches from underlying
968 <quote>normal</quote> revisions. The following example shows a
969 few normal Mercurial commands in use with applied
970 patches.</para>
972 <!-- &interaction.mq.id.output; -->
974 </sect1>
975 <sect1>
976 <title>Useful things to know about</title>
978 <para>There are a number of aspects of MQ usage that don't fit
979 tidily into sections of their own, but that are good to know.
980 Here they are, in one place.</para>
982 <itemizedlist>
983 <listitem><para>Normally, when you <command
984 role="hg-ext-mq">qpop</command> a patch and <command
985 role="hg-ext-mq">qpush</command> it again, the changeset
986 that represents the patch after the pop/push will have a
987 <emphasis>different identity</emphasis> than the changeset
988 that represented the hash beforehand. See section <xref
989 linkend="sec:mqref:cmd:qpush"/> for
990 information as to why this is.</para>
991 </listitem>
992 <listitem><para>It's not a good idea to <command
993 role="hg-cmd">hg merge</command> changes from another
994 branch with a patch changeset, at least if you want to
995 maintain the <quote>patchiness</quote> of that changeset and
996 changesets below it on the patch stack. If you try to do
997 this, it will appear to succeed, but MQ will become
998 confused.</para>
999 </listitem></itemizedlist>
1001 </sect1>
1002 <sect1 id="sec:mq:repo">
1003 <title>Managing patches in a repository</title>
1005 <para>Because MQ's <filename role="special"
1006 class="directory">.hg/patches</filename> directory resides
1007 outside a Mercurial repository's working directory, the
1008 <quote>underlying</quote> Mercurial repository knows nothing
1009 about the management or presence of patches.</para>
1011 <para>This presents the interesting possibility of managing the
1012 contents of the patch directory as a Mercurial repository in its
1013 own right. This can be a useful way to work. For example, you
1014 can work on a patch for a while, <command
1015 role="hg-ext-mq">qrefresh</command> it, then <command
1016 role="hg-cmd">hg commit</command> the current state of the
1017 patch. This lets you <quote>roll back</quote> to that version
1018 of the patch later on.</para>
1020 <para>You can then share different versions of the same patch
1021 stack among multiple underlying repositories. I use this when I
1022 am developing a Linux kernel feature. I have a pristine copy of
1023 my kernel sources for each of several CPU architectures, and a
1024 cloned repository under each that contains the patches I am
1025 working on. When I want to test a change on a different
1026 architecture, I push my current patches to the patch repository
1027 associated with that kernel tree, pop and push all of my
1028 patches, and build and test that kernel.</para>
1030 <para>Managing patches in a repository makes it possible for
1031 multiple developers to work on the same patch series without
1032 colliding with each other, all on top of an underlying source
1033 base that they may or may not control.</para>
1035 <sect2>
1036 <title>MQ support for patch repositories</title>
1038 <para>MQ helps you to work with the <filename role="special"
1039 class="directory">.hg/patches</filename> directory as a
1040 repository; when you prepare a repository for working with
1041 patches using <command role="hg-ext-mq">qinit</command>, you
1042 can pass the <option role="hg-ext-mq-cmd-qinit-opt">hg
1043 -c</option> option to create the <filename role="special"
1044 class="directory">.hg/patches</filename> directory as a
1045 Mercurial repository.</para>
1047 <note>
1048 <para> If you forget to use the <option
1049 role="hg-ext-mq-cmd-qinit-opt">hg -c</option> option, you
1050 can simply go into the <filename role="special"
1051 class="directory">.hg/patches</filename> directory at any
1052 time and run <command role="hg-cmd">hg init</command>.
1053 Don't forget to add an entry for the <filename
1054 role="special">status</filename> file to the <filename
1055 role="special">.hgignore</filename> file, though</para>
1057 <para> (<command role="hg-cmd">hg qinit <option
1058 role="hg-ext-mq-cmd-qinit-opt">hg -c</option></command>
1059 does this for you automatically); you
1060 <emphasis>really</emphasis> don't want to manage the
1061 <filename role="special">status</filename> file.</para>
1062 </note>
1064 <para>As a convenience, if MQ notices that the <filename
1065 class="directory">.hg/patches</filename> directory is a
1066 repository, it will automatically <command role="hg-cmd">hg
1067 add</command> every patch that you create and import.</para>
1069 <para>MQ provides a shortcut command, <command
1070 role="hg-ext-mq">qcommit</command>, that runs <command
1071 role="hg-cmd">hg commit</command> in the <filename
1072 role="special" class="directory">.hg/patches</filename>
1073 directory. This saves some bothersome typing.</para>
1075 <para>Finally, as a convenience to manage the patch directory,
1076 you can define the alias <command>mq</command> on Unix
1077 systems. For example, on Linux systems using the
1078 <command>bash</command> shell, you can include the following
1079 snippet in your <filename
1080 role="home">~/.bashrc</filename>.</para>
1082 <programlisting>alias mq=`hg -R $(hg
1083 root)/.hg/patches'</programlisting>
1085 <para>You can then issue commands of the form <command>mq
1086 pull</command> from the main repository.</para>
1088 </sect2>
1089 <sect2>
1090 <title>A few things to watch out for</title>
1092 <para>MQ's support for working with a repository full of patches
1093 is limited in a few small respects.</para>
1095 <para>MQ cannot automatically detect changes that you make to
1096 the patch directory. If you <command role="hg-cmd">hg
1097 pull</command>, manually edit, or <command role="hg-cmd">hg
1098 update</command> changes to patches or the <filename
1099 role="special">series</filename> file, you will have to
1100 <command role="hg-cmd">hg qpop <option
1101 role="hg-ext-mq-cmd-qpop-opt">hg -a</option></command> and
1102 then <command role="hg-cmd">hg qpush <option
1103 role="hg-ext-mq-cmd-qpush-opt">hg -a</option></command> in
1104 the underlying repository to see those changes show up there.
1105 If you forget to do this, you can confuse MQ's idea of which
1106 patches are applied.</para>
1108 </sect2>
1109 </sect1>
1110 <sect1 id="sec:mq:tools">
1111 <title>Third party tools for working with patches</title>
1113 <para>Once you've been working with patches for a while, you'll
1114 find yourself hungry for tools that will help you to understand
1115 and manipulate the patches you're dealing with.</para>
1117 <para>The <command>diffstat</command> command
1118 <citation>web:diffstat</citation> generates a histogram of the
1119 modifications made to each file in a patch. It provides a good
1120 way to <quote>get a sense of</quote> a patch&emdash;which files
1121 it affects, and how much change it introduces to each file and
1122 as a whole. (I find that it's a good idea to use
1123 <command>diffstat</command>'s <option
1124 role="cmd-opt-diffstat">-p</option> option as a matter of
1125 course, as otherwise it will try to do clever things with
1126 prefixes of file names that inevitably confuse at least
1127 me.)</para>
1129 <!-- &interaction.mq.tools.tools; -->
1131 <para>The <literal role="package">patchutils</literal> package
1132 <citation>web:patchutils</citation> is invaluable. It provides a
1133 set of small utilities that follow the <quote>Unix
1134 philosophy;</quote> each does one useful thing with a patch.
1135 The <literal role="package">patchutils</literal> command I use
1136 most is <command>filterdiff</command>, which extracts subsets
1137 from a patch file. For example, given a patch that modifies
1138 hundreds of files across dozens of directories, a single
1139 invocation of <command>filterdiff</command> can generate a
1140 smaller patch that only touches files whose names match a
1141 particular glob pattern. See section <xref
1142 linkend="mq-collab:tips:interdiff"/> for another
1143 example.</para>
1145 </sect1>
1146 <sect1>
1147 <title>Good ways to work with patches</title>
1149 <para>Whether you are working on a patch series to submit to a
1150 free software or open source project, or a series that you
1151 intend to treat as a sequence of regular changesets when you're
1152 done, you can use some simple techniques to keep your work well
1153 organised.</para>
1155 <para>Give your patches descriptive names. A good name for a
1156 patch might be <filename>rework-device-alloc.patch</filename>,
1157 because it will immediately give you a hint what the purpose of
1158 the patch is. Long names shouldn't be a problem; you won't be
1159 typing the names often, but you <emphasis>will</emphasis> be
1160 running commands like <command
1161 role="hg-ext-mq">qapplied</command> and <command
1162 role="hg-ext-mq">qtop</command> over and over. Good naming
1163 becomes especially important when you have a number of patches
1164 to work with, or if you are juggling a number of different tasks
1165 and your patches only get a fraction of your attention.</para>
1167 <para>Be aware of what patch you're working on. Use the <command
1168 role="hg-ext-mq">qtop</command> command and skim over the text
1169 of your patches frequently&emdash;for example, using <command
1170 role="hg-cmd">hg tip <option
1171 role="hg-opt-tip">-p</option></command>)&emdash;to be sure
1172 of where you stand. I have several times worked on and <command
1173 role="hg-ext-mq">qrefresh</command>ed a patch other than the
1174 one I intended, and it's often tricky to migrate changes into
1175 the right patch after making them in the wrong one.</para>
1177 <para>For this reason, it is very much worth investing a little
1178 time to learn how to use some of the third-party tools I
1179 described in section <xref linkend="sec:mq:tools"/>,
1180 particularly
1181 <command>diffstat</command> and <command>filterdiff</command>.
1182 The former will give you a quick idea of what changes your patch
1183 is making, while the latter makes it easy to splice hunks
1184 selectively out of one patch and into another.</para>
1186 </sect1>
1187 <sect1>
1188 <title>MQ cookbook</title>
1190 <sect2>
1191 <title>Manage <quote>trivial</quote> patches</title>
1193 <para>Because the overhead of dropping files into a new
1194 Mercurial repository is so low, it makes a lot of sense to
1195 manage patches this way even if you simply want to make a few
1196 changes to a source tarball that you downloaded.</para>
1198 <para>Begin by downloading and unpacking the source tarball, and
1199 turning it into a Mercurial repository. <!--
1200 &interaction.mq.tarball.download; --></para>
1202 <para>Continue by creating a patch stack and making your
1203 changes. <!-- &interaction.mq.tarball.qinit; --></para>
1205 <para>Let's say a few weeks or months pass, and your package
1206 author releases a new version. First, bring their changes
1207 into the repository. <!-- &interaction.mq.tarball.newsource;
1208 --> The pipeline starting with <command role="hg-cmd">hg
1209 locate</command> above deletes all files in the working
1210 directory, so that <command role="hg-cmd">hg
1211 commit</command>'s <option
1212 role="hg-opt-commit">--addremove</option> option can
1213 actually tell which files have really been removed in the
1214 newer version of the source.</para>
1216 <para>Finally, you can apply your patches on top of the new
1217 tree. <!-- &interaction.mq.tarball.repush; --></para>
1219 </sect2>
1220 <sect2 id="sec:mq:combine">
1221 <title>Combining entire patches</title>
1223 <para>MQ provides a command, <command
1224 role="hg-ext-mq">qfold</command> that lets you combine
1225 entire patches. This <quote>folds</quote> the patches you
1226 name, in the order you name them, into the topmost applied
1227 patch, and concatenates their descriptions onto the end of its
1228 description. The patches that you fold must be unapplied
1229 before you fold them.</para>
1231 <para>The order in which you fold patches matters. If your
1232 topmost applied patch is <literal>foo</literal>, and you
1233 <command role="hg-ext-mq">qfold</command>
1234 <literal>bar</literal> and <literal>quux</literal> into it,
1235 you will end up with a patch that has the same effect as if
1236 you applied first <literal>foo</literal>, then
1237 <literal>bar</literal>, followed by
1238 <literal>quux</literal>.</para>
1240 </sect2>
1241 <sect2>
1242 <title>Merging part of one patch into another</title>
1244 <para>Merging <emphasis>part</emphasis> of one patch into
1245 another is more difficult than combining entire
1246 patches.</para>
1248 <para>If you want to move changes to entire files, you can use
1249 <command>filterdiff</command>'s <option
1250 role="cmd-opt-filterdiff">-i</option> and <option
1251 role="cmd-opt-filterdiff">-x</option> options to choose the
1252 modifications to snip out of one patch, concatenating its
1253 output onto the end of the patch you want to merge into. You
1254 usually won't need to modify the patch you've merged the
1255 changes from. Instead, MQ will report some rejected hunks
1256 when you <command role="hg-ext-mq">qpush</command> it (from
1257 the hunks you moved into the other patch), and you can simply
1258 <command role="hg-ext-mq">qrefresh</command> the patch to drop
1259 the duplicate hunks.</para>
1261 <para>If you have a patch that has multiple hunks modifying a
1262 file, and you only want to move a few of those hunks, the job
1263 becomes more messy, but you can still partly automate it. Use
1264 <command>lsdiff -nvv</command> to print some metadata about
1265 the patch. <!-- &interaction.mq.tools.lsdiff; --></para>
1267 <para>This command prints three different kinds of
1268 number:</para>
1269 <itemizedlist>
1270 <listitem><para>(in the first column) a <emphasis>file
1271 number</emphasis> to identify each file modified in the
1272 patch;</para>
1273 </listitem>
1274 <listitem><para>(on the next line, indented) the line number
1275 within a modified file where a hunk starts; and</para>
1276 </listitem>
1277 <listitem><para>(on the same line) a <emphasis>hunk
1278 number</emphasis> to identify that hunk.</para>
1279 </listitem></itemizedlist>
1281 <para>You'll have to use some visual inspection, and reading of
1282 the patch, to identify the file and hunk numbers you'll want,
1283 but you can then pass them to to
1284 <command>filterdiff</command>'s <option
1285 role="cmd-opt-filterdiff">--files</option> and <option
1286 role="cmd-opt-filterdiff">--hunks</option> options, to
1287 select exactly the file and hunk you want to extract.</para>
1289 <para>Once you have this hunk, you can concatenate it onto the
1290 end of your destination patch and continue with the remainder
1291 of section <xref linkend="sec:mq:combine"/>.</para>
1293 </sect2>
1294 </sect1>
1295 <sect1>
1296 <title>Differences between quilt and MQ</title>
1298 <para>If you are already familiar with quilt, MQ provides a
1299 similar command set. There are a few differences in the way
1300 that it works.</para>
1302 <para>You will already have noticed that most quilt commands have
1303 MQ counterparts that simply begin with a
1304 <quote><literal>q</literal></quote>. The exceptions are quilt's
1305 <literal>add</literal> and <literal>remove</literal> commands,
1306 the counterparts for which are the normal Mercurial <command
1307 role="hg-cmd">hg add</command> and <command role="hg-cmd">hg
1308 remove</command> commands. There is no MQ equivalent of the
1309 quilt <literal>edit</literal> command.</para>
1311 </sect1>
1312 </chapter>
1314 <!--
1315 local variables:
1316 sgml-parent-document: ("00book.xml" "book" "chapter")
1317 end:
1318 -->