hgbook
view en/ch08-branch.xml @ 634:d1a6ed326f87
Add package and upload script
| author | Dongsheng Song <dongsheng.song@gmail.com> |
|---|---|
| date | Fri Mar 13 16:01:21 2009 +0800 (2009-03-13) |
| parents | 13513d2a128d |
| children | d0160b0b1a9e |
line source
1 <!-- vim: set filetype=docbkxml shiftwidth=2 autoindent expandtab tw=77 : -->
3 <chapter id="chap.branch">
4 <?dbhtml filename="managing-releases-and-branchy-development.html"?>
5 <title>Managing releases and branchy development</title>
7 <para>Mercurial provides several mechanisms for you to manage a
8 project that is making progress on multiple fronts at once. To
9 understand these mechanisms, let's first take a brief look at a
10 fairly normal software project structure.</para>
12 <para>Many software projects issue periodic <quote>major</quote>
13 releases that contain substantial new features. In parallel, they
14 may issue <quote>minor</quote> releases. These are usually
15 identical to the major releases off which they're based, but with
16 a few bugs fixed.</para>
18 <para>In this chapter, we'll start by talking about how to keep
19 records of project milestones such as releases. We'll then
20 continue on to talk about the flow of work between different
21 phases of a project, and how Mercurial can help you to isolate and
22 manage this work.</para>
24 <sect1>
25 <title>Giving a persistent name to a revision</title>
27 <para>Once you decide that you'd like to call a particular
28 revision a <quote>release</quote>, it's a good idea to record
29 the identity of that revision. This will let you reproduce that
30 release at a later date, for whatever purpose you might need at
31 the time (reproducing a bug, porting to a new platform, etc).
32 &interaction.tag.init;</para>
34 <para>Mercurial lets you give a permanent name to any revision
35 using the <command role="hg-cmd">hg tag</command> command. Not
36 surprisingly, these names are called <quote>tags</quote>.</para>
38 &interaction.tag.tag;
40 <para>A tag is nothing more than a <quote>symbolic name</quote>
41 for a revision. Tags exist purely for your convenience, so that
42 you have a handy permanent way to refer to a revision; Mercurial
43 doesn't interpret the tag names you use in any way. Neither
44 does Mercurial place any restrictions on the name of a tag,
45 beyond a few that are necessary to ensure that a tag can be
46 parsed unambiguously. A tag name cannot contain any of the
47 following characters:</para>
48 <itemizedlist>
49 <listitem><para>Colon (ASCII 58,
50 <quote><literal>:</literal></quote>)</para>
51 </listitem>
52 <listitem><para>Carriage return (ASCII 13,
53 <quote><literal>\r</literal></quote>)</para>
54 </listitem>
55 <listitem><para>Newline (ASCII 10,
56 <quote><literal>\n</literal></quote>)</para>
57 </listitem></itemizedlist>
59 <para>You can use the <command role="hg-cmd">hg tags</command>
60 command to display the tags present in your repository. In the
61 output, each tagged revision is identified first by its name,
62 then by revision number, and finally by the unique hash of the
63 revision.</para>
65 &interaction.tag.tags;
67 <para>Notice that <literal>tip</literal> is listed in the output
68 of <command role="hg-cmd">hg tags</command>. The
69 <literal>tip</literal> tag is a special <quote>floating</quote>
70 tag, which always identifies the newest revision in the
71 repository.</para>
73 <para>In the output of the <command role="hg-cmd">hg
74 tags</command> command, tags are listed in reverse order, by
75 revision number. This usually means that recent tags are listed
76 before older tags. It also means that <literal>tip</literal> is
77 always going to be the first tag listed in the output of
78 <command role="hg-cmd">hg tags</command>.</para>
80 <para>When you run <command role="hg-cmd">hg log</command>, if it
81 displays a revision that has tags associated with it, it will
82 print those tags.</para>
84 &interaction.tag.log;
86 <para>Any time you need to provide a revision ID to a Mercurial
87 command, the command will accept a tag name in its place.
88 Internally, Mercurial will translate your tag name into the
89 corresponding revision ID, then use that.</para>
91 &interaction.tag.log.v1.0;
93 <para>There's no limit on the number of tags you can have in a
94 repository, or on the number of tags that a single revision can
95 have. As a practical matter, it's not a great idea to have
96 <quote>too many</quote> (a number which will vary from project
97 to project), simply because tags are supposed to help you to
98 find revisions. If you have lots of tags, the ease of using
99 them to identify revisions diminishes rapidly.</para>
101 <para>For example, if your project has milestones as frequent as
102 every few days, it's perfectly reasonable to tag each one of
103 those. But if you have a continuous build system that makes
104 sure every revision can be built cleanly, you'd be introducing a
105 lot of noise if you were to tag every clean build. Instead, you
106 could tag failed builds (on the assumption that they're rare!),
107 or simply not use tags to track buildability.</para>
109 <para>If you want to remove a tag that you no longer want, use
110 <command role="hg-cmd">hg tag --remove</command>.</para>
112 &interaction.tag.remove;
114 <para>You can also modify a tag at any time, so that it identifies
115 a different revision, by simply issuing a new <command
116 role="hg-cmd">hg tag</command> command. You'll have to use the
117 <option role="hg-opt-tag">-f</option> option to tell Mercurial
118 that you <emphasis>really</emphasis> want to update the
119 tag.</para>
121 &interaction.tag.replace;
123 <para>There will still be a permanent record of the previous
124 identity of the tag, but Mercurial will no longer use it.
125 There's thus no penalty to tagging the wrong revision; all you
126 have to do is turn around and tag the correct revision once you
127 discover your error.</para>
129 <para>Mercurial stores tags in a normal revision-controlled file
130 in your repository. If you've created any tags, you'll find
131 them in a file named <filename
132 role="special">.hgtags</filename>. When you run the <command
133 role="hg-cmd">hg tag</command> command, Mercurial modifies
134 this file, then automatically commits the change to it. This
135 means that every time you run <command role="hg-cmd">hg
136 tag</command>, you'll see a corresponding changeset in the
137 output of <command role="hg-cmd">hg log</command>.</para>
139 &interaction.tag.tip;
141 <sect2>
142 <title>Handling tag conflicts during a merge</title>
144 <para>You won't often need to care about the <filename
145 role="special">.hgtags</filename> file, but it sometimes
146 makes its presence known during a merge. The format of the
147 file is simple: it consists of a series of lines. Each line
148 starts with a changeset hash, followed by a space, followed by
149 the name of a tag.</para>
151 <para>If you're resolving a conflict in the <filename
152 role="special">.hgtags</filename> file during a merge,
153 there's one twist to modifying the <filename
154 role="special">.hgtags</filename> file: when Mercurial is
155 parsing the tags in a repository, it
156 <emphasis>never</emphasis> reads the working copy of the
157 <filename role="special">.hgtags</filename> file. Instead, it
158 reads the <emphasis>most recently committed</emphasis>
159 revision of the file.</para>
161 <para>An unfortunate consequence of this design is that you
162 can't actually verify that your merged <filename
163 role="special">.hgtags</filename> file is correct until
164 <emphasis>after</emphasis> you've committed a change. So if
165 you find yourself resolving a conflict on <filename
166 role="special">.hgtags</filename> during a merge, be sure to
167 run <command role="hg-cmd">hg tags</command> after you commit.
168 If it finds an error in the <filename
169 role="special">.hgtags</filename> file, it will report the
170 location of the error, which you can then fix and commit. You
171 should then run <command role="hg-cmd">hg tags</command>
172 again, just to be sure that your fix is correct.</para>
174 </sect2>
175 <sect2>
176 <title>Tags and cloning</title>
178 <para>You may have noticed that the <command role="hg-cmd">hg
179 clone</command> command has a <option
180 role="hg-opt-clone">-r</option> option that lets you clone
181 an exact copy of the repository as of a particular changeset.
182 The new clone will not contain any project history that comes
183 after the revision you specified. This has an interaction
184 with tags that can surprise the unwary.</para>
186 <para>Recall that a tag is stored as a revision to the <filename
187 role="special">.hgtags</filename> file, so that when you
188 create a tag, the changeset in which it's recorded necessarily
189 refers to an older changeset. When you run <command
190 role="hg-cmd">hg clone -r foo</command> to clone a
191 repository as of tag <literal>foo</literal>, the new clone
192 <emphasis>will not contain the history that created the
193 tag</emphasis> that you used to clone the repository. The
194 result is that you'll get exactly the right subset of the
195 project's history in the new repository, but
196 <emphasis>not</emphasis> the tag you might have
197 expected.</para>
199 </sect2>
200 <sect2>
201 <title>When permanent tags are too much</title>
203 <para>Since Mercurial's tags are revision controlled and carried
204 around with a project's history, everyone you work with will
205 see the tags you create. But giving names to revisions has
206 uses beyond simply noting that revision
207 <literal>4237e45506ee</literal> is really
208 <literal>v2.0.2</literal>. If you're trying to track down a
209 subtle bug, you might want a tag to remind you of something
210 like <quote>Anne saw the symptoms with this
211 revision</quote>.</para>
213 <para>For cases like this, what you might want to use are
214 <emphasis>local</emphasis> tags. You can create a local tag
215 with the <option role="hg-opt-tag">-l</option> option to the
216 <command role="hg-cmd">hg tag</command> command. This will
217 store the tag in a file called <filename
218 role="special">.hg/localtags</filename>. Unlike <filename
219 role="special">.hgtags</filename>, <filename
220 role="special">.hg/localtags</filename> is not revision
221 controlled. Any tags you create using <option
222 role="hg-opt-tag">-l</option> remain strictly local to the
223 repository you're currently working in.</para>
225 </sect2>
226 </sect1>
227 <sect1>
228 <title>The flow of changes&emdash;big picture vs. little</title>
230 <para>To return to the outline I sketched at the beginning of a
231 chapter, let's think about a project that has multiple
232 concurrent pieces of work under development at once.</para>
234 <para>There might be a push for a new <quote>main</quote> release;
235 a new minor bugfix release to the last main release; and an
236 unexpected <quote>hot fix</quote> to an old release that is now
237 in maintenance mode.</para>
239 <para>The usual way people refer to these different concurrent
240 directions of development is as <quote>branches</quote>.
241 However, we've already seen numerous times that Mercurial treats
242 <emphasis>all of history</emphasis> as a series of branches and
243 merges. Really, what we have here is two ideas that are
244 peripherally related, but which happen to share a name.</para>
245 <itemizedlist>
246 <listitem><para><quote>Big picture</quote> branches represent
247 the sweep of a project's evolution; people give them names,
248 and talk about them in conversation.</para>
249 </listitem>
250 <listitem><para><quote>Little picture</quote> branches are
251 artefacts of the day-to-day activity of developing and
252 merging changes. They expose the narrative of how the code
253 was developed.</para>
254 </listitem></itemizedlist>
256 </sect1>
257 <sect1>
258 <title>Managing big-picture branches in repositories</title>
260 <para>The easiest way to isolate a <quote>big picture</quote>
261 branch in Mercurial is in a dedicated repository. If you have
262 an existing shared repository&emdash;let's call it
263 <literal>myproject</literal>&emdash;that reaches a
264 <quote>1.0</quote> milestone, you can start to prepare for
265 future maintenance releases on top of version 1.0 by tagging the
266 revision from which you prepared the 1.0 release.</para>
268 &interaction.branch-repo.tag;
270 <para>You can then clone a new shared
271 <literal>myproject-1.0.1</literal> repository as of that
272 tag.</para>
274 &interaction.branch-repo.clone;
276 <para>Afterwards, if someone needs to work on a bug fix that ought
277 to go into an upcoming 1.0.1 minor release, they clone the
278 <literal>myproject-1.0.1</literal> repository, make their
279 changes, and push them back.</para>
281 &interaction.branch-repo.bugfix;
283 <para>Meanwhile, development for
284 the next major release can continue, isolated and unabated, in
285 the <literal>myproject</literal> repository.</para>
287 &interaction.branch-repo.new;
289 </sect1>
290 <sect1>
291 <title>Don't repeat yourself: merging across branches</title>
293 <para>In many cases, if you have a bug to fix on a maintenance
294 branch, the chances are good that the bug exists on your
295 project's main branch (and possibly other maintenance branches,
296 too). It's a rare developer who wants to fix the same bug
297 multiple times, so let's look at a few ways that Mercurial can
298 help you to manage these bugfixes without duplicating your
299 work.</para>
301 <para>In the simplest instance, all you need to do is pull changes
302 from your maintenance branch into your local clone of the target
303 branch.</para>
305 &interaction.branch-repo.pull;
307 <para>You'll then need to merge the heads of the two branches, and
308 push back to the main branch.</para>
310 &interaction.branch-repo.merge;
312 </sect1>
313 <sect1>
314 <title>Naming branches within one repository</title>
316 <para>In most instances, isolating branches in repositories is the
317 right approach. Its simplicity makes it easy to understand; and
318 so it's hard to make mistakes. There's a one-to-one
319 relationship between branches you're working in and directories
320 on your system. This lets you use normal (non-Mercurial-aware)
321 tools to work on files within a branch/repository.</para>
323 <para>If you're more in the <quote>power user</quote> category
324 (<emphasis>and</emphasis> your collaborators are too), there is
325 an alternative way of handling branches that you can consider.
326 I've already mentioned the human-level distinction between
327 <quote>small picture</quote> and <quote>big picture</quote>
328 branches. While Mercurial works with multiple <quote>small
329 picture</quote> branches in a repository all the time (for
330 example after you pull changes in, but before you merge them),
331 it can <emphasis>also</emphasis> work with multiple <quote>big
332 picture</quote> branches.</para>
334 <para>The key to working this way is that Mercurial lets you
335 assign a persistent <emphasis>name</emphasis> to a branch.
336 There always exists a branch named <literal>default</literal>.
337 Even before you start naming branches yourself, you can find
338 traces of the <literal>default</literal> branch if you look for
339 them.</para>
341 <para>As an example, when you run the <command role="hg-cmd">hg
342 commit</command> command, and it pops up your editor so that
343 you can enter a commit message, look for a line that contains
344 the text <quote><literal>HG: branch default</literal></quote> at
345 the bottom. This is telling you that your commit will occur on
346 the branch named <literal>default</literal>.</para>
348 <para>To start working with named branches, use the <command
349 role="hg-cmd">hg branches</command> command. This command
350 lists the named branches already present in your repository,
351 telling you which changeset is the tip of each.</para>
353 &interaction.branch-named.branches;
355 <para>Since you haven't created any named branches yet, the only
356 one that exists is <literal>default</literal>.</para>
358 <para>To find out what the <quote>current</quote> branch is, run
359 the <command role="hg-cmd">hg branch</command> command, giving
360 it no arguments. This tells you what branch the parent of the
361 current changeset is on.</para>
363 &interaction.branch-named.branch;
365 <para>To create a new branch, run the <command role="hg-cmd">hg
366 branch</command> command again. This time, give it one
367 argument: the name of the branch you want to create.</para>
369 &interaction.branch-named.create;
371 <para>After you've created a branch, you might wonder what effect
372 the <command role="hg-cmd">hg branch</command> command has had.
373 What do the <command role="hg-cmd">hg status</command> and
374 <command role="hg-cmd">hg tip</command> commands report?</para>
376 &interaction.branch-named.status;
378 <para>Nothing has changed in the
379 working directory, and there's been no new history created. As
380 this suggests, running the <command role="hg-cmd">hg
381 branch</command> command has no permanent effect; it only
382 tells Mercurial what branch name to use the
383 <emphasis>next</emphasis> time you commit a changeset.</para>
385 <para>When you commit a change, Mercurial records the name of the
386 branch on which you committed. Once you've switched from the
387 <literal>default</literal> branch to another and committed,
388 you'll see the name of the new branch show up in the output of
389 <command role="hg-cmd">hg log</command>, <command
390 role="hg-cmd">hg tip</command>, and other commands that
391 display the same kind of output.</para>
393 &interaction.branch-named.commit;
395 <para>The <command role="hg-cmd">hg log</command>-like commands
396 will print the branch name of every changeset that's not on the
397 <literal>default</literal> branch. As a result, if you never
398 use named branches, you'll never see this information.</para>
400 <para>Once you've named a branch and committed a change with that
401 name, every subsequent commit that descends from that change
402 will inherit the same branch name. You can change the name of a
403 branch at any time, using the <command role="hg-cmd">hg
404 branch</command> command.</para>
406 &interaction.branch-named.rebranch;
408 <para>In practice, this is something you won't do very often, as
409 branch names tend to have fairly long lifetimes. (This isn't a
410 rule, just an observation.)</para>
412 </sect1>
413 <sect1>
414 <title>Dealing with multiple named branches in a
415 repository</title>
417 <para>If you have more than one named branch in a repository,
418 Mercurial will remember the branch that your working directory
419 on when you start a command like <command role="hg-cmd">hg
420 update</command> or <command role="hg-cmd">hg pull
421 -u</command>. It will update the working directory to the tip
422 of this branch, no matter what the <quote>repo-wide</quote> tip
423 is. To update to a revision that's on a different named branch,
424 you may need to use the <option role="hg-opt-update">-C</option>
425 option to <command role="hg-cmd">hg update</command>.</para>
427 <para>This behaviour is a little subtle, so let's see it in
428 action. First, let's remind ourselves what branch we're
429 currently on, and what branches are in our repository.</para>
431 &interaction.branch-named.parents;
433 <para>We're on the <literal>bar</literal> branch, but there also
434 exists an older <command role="hg-cmd">hg foo</command>
435 branch.</para>
437 <para>We can <command role="hg-cmd">hg update</command> back and
438 forth between the tips of the <literal>foo</literal> and
439 <literal>bar</literal> branches without needing to use the
440 <option role="hg-opt-update">-C</option> option, because this
441 only involves going backwards and forwards linearly through our
442 change history.</para>
444 &interaction.branch-named.update-switchy;
446 <para>If we go back to the <literal>foo</literal> branch and then
447 run <command role="hg-cmd">hg update</command>, it will keep us
448 on <literal>foo</literal>, not move us to the tip of
449 <literal>bar</literal>.</para>
451 &interaction.branch-named.update-nothing;
453 <para>Committing a new change on the <literal>foo</literal> branch
454 introduces a new head.</para>
456 &interaction.branch-named.foo-commit;
458 </sect1>
459 <sect1>
460 <title>Branch names and merging</title>
462 <para>As you've probably noticed, merges in Mercurial are not
463 symmetrical. Let's say our repository has two heads, 17 and 23.
464 If I <command role="hg-cmd">hg update</command> to 17 and then
465 <command role="hg-cmd">hg merge</command> with 23, Mercurial
466 records 17 as the first parent of the merge, and 23 as the
467 second. Whereas if I <command role="hg-cmd">hg update</command>
468 to 23 and then <command role="hg-cmd">hg merge</command> with
469 17, it records 23 as the first parent, and 17 as the
470 second.</para>
472 <para>This affects Mercurial's choice of branch name when you
473 merge. After a merge, Mercurial will retain the branch name of
474 the first parent when you commit the result of the merge. If
475 your first parent's branch name is <literal>foo</literal>, and
476 you merge with <literal>bar</literal>, the branch name will
477 still be <literal>foo</literal> after you merge.</para>
479 <para>It's not unusual for a repository to contain multiple heads,
480 each with the same branch name. Let's say I'm working on the
481 <literal>foo</literal> branch, and so are you. We commit
482 different changes; I pull your changes; I now have two heads,
483 each claiming to be on the <literal>foo</literal> branch. The
484 result of a merge will be a single head on the
485 <literal>foo</literal> branch, as you might hope.</para>
487 <para>But if I'm working on the <literal>bar</literal> branch, and
488 I merge work from the <literal>foo</literal> branch, the result
489 will remain on the <literal>bar</literal> branch.</para>
491 &interaction.branch-named.merge;
493 <para>To give a more concrete example, if I'm working on the
494 <literal>bleeding-edge</literal> branch, and I want to bring in
495 the latest fixes from the <literal>stable</literal> branch,
496 Mercurial will choose the <quote>right</quote>
497 (<literal>bleeding-edge</literal>) branch name when I pull and
498 merge from <literal>stable</literal>.</para>
500 </sect1>
501 <sect1>
502 <title>Branch naming is generally useful</title>
504 <para>You shouldn't think of named branches as applicable only to
505 situations where you have multiple long-lived branches
506 cohabiting in a single repository. They're very useful even in
507 the one-branch-per-repository case.</para>
509 <para>In the simplest case, giving a name to each branch gives you
510 a permanent record of which branch a changeset originated on.
511 This gives you more context when you're trying to follow the
512 history of a long-lived branchy project.</para>
514 <para>If you're working with shared repositories, you can set up a
515 <literal role="hook">pretxnchangegroup</literal> hook on each
516 that will block incoming changes that have the
517 <quote>wrong</quote> branch name. This provides a simple, but
518 effective, defence against people accidentally pushing changes
519 from a <quote>bleeding edge</quote> branch to a
520 <quote>stable</quote> branch. Such a hook might look like this
521 inside the shared repo's <filename role="special">
522 /.hgrc</filename>.</para>
523 <programlisting>[hooks] pretxnchangegroup.branch = hg heads
524 --template '{branches} ' | grep mybranch</programlisting>
526 </sect1>
527 </chapter>
529 <!--
530 local variables:
531 sgml-parent-document: ("00book.xml" "book" "chapter")
532 end:
533 -->