1GITWORKFLOWS(7) Git Manual GITWORKFLOWS(7)
2
6 gitworkflows - An overview of recommended workflows with git
7
9 git *
10
12 This document attempts to write down and motivate some of the workflow
13 elements used for git.git itself. Many ideas apply in general, though
14 the full workflow is rarely required for smaller projects with fewer
15 people involved.
16 We formulate a set of rules for quick reference, while the prose tries
18 to motivate each of them. Do not always take them literally; you should
19 value good reasons for your actions higher than manpages such as this
20 one.
21
23 As a general rule, you should try to split your changes into small
24 logical steps, and commit each of them. They should be consistent,
25 working independently of any later commits, pass the test suite, etc.
26 This makes the review process much easier, and the history much more
27 useful for later inspection and analysis, for example with git-blame(1)
28 and git-bisect(1).
29 To achieve this, try to split your work into small steps from the very
31 beginning. It is always easier to squash a few commits together than to
32 split one big commit into several. Don’t be afraid of making too small
33 or imperfect steps along the way. You can always go back later and edit
34 the commits with git rebase --interactive before you publish them. You
35 can use git stash save --keep-index to run the test suite independent
36 of other uncommitted changes; see the EXAMPLES section of git-stash(1).
37
39 There are two main tools that can be used to include changes from one
40 branch on another: git-merge(1) and git-cherry-pick(1).
41 Merges have many advantages, so we try to solve as many problems as
43 possible with merges alone. Cherry-picking is still occasionally
44 useful; see "Merging upwards" below for an example.
45 Most importantly, merging works at the branch level, while
47 cherry-picking works at the commit level. This means that a merge can
48 carry over the changes from 1, 10, or 1000 commits with equal ease,
49 which in turn means the workflow scales much better to a large number
50 of contributors (and contributions). Merges are also easier to
51 understand because a merge commit is a "promise" that all changes from
52 all its parents are now included.
53 There is a tradeoff of course: merges require a more careful branch
55 management. The following subsections discuss the important points.
56 Graduation
58 As a given feature goes from experimental to stable, it also
59 "graduates" between the corresponding branches of the software. git.git
60 uses the following integration branches:
61 · maint tracks the commits that should go into the next "maintenance
63 release", i.e., update of the last released stable version;
64 · master tracks the commits that should go into the next release;
66 · next is intended as a testing branch for topics being tested for
68 stability for master.
69 There is a fourth official branch that is used slightly differently:
71 · pu (proposed updates) is an integration branch for things that are
73 not quite ready for inclusion yet (see "Integration Branches"
74 below).
75 Each of the four branches is usually a direct descendant of the one
77 above it.
78 Conceptually, the feature enters at an unstable branch (usually next or
80 pu), and "graduates" to master for the next release once it is
81 considered stable enough.
82 Merging upwards
84 The "downwards graduation" discussed above cannot be done by actually
85 merging downwards, however, since that would merge all changes on the
86 unstable branch into the stable one. Hence the following:
87 Example 1. Merge upwards
89 Always commit your fixes to the oldest supported branch that require
91 them. Then (periodically) merge the integration branches upwards into
92 each other.
93 This gives a very controlled flow of fixes. If you notice that you have
95 applied a fix to e.g. master that is also required in maint, you will
96 need to cherry-pick it (using git-cherry-pick(1)) downwards. This will
97 happen a few times and is nothing to worry about unless you do it very
98 frequently.
99 Topic branches
101 Any nontrivial feature will require several patches to implement, and
102 may get extra bugfixes or improvements during its lifetime.
103 Committing everything directly on the integration branches leads to
105 many problems: Bad commits cannot be undone, so they must be reverted
106 one by one, which creates confusing histories and further error
107 potential when you forget to revert part of a group of changes. Working
108 in parallel mixes up the changes, creating further confusion.
109 Use of "topic branches" solves these problems. The name is pretty self
111 explanatory, with a caveat that comes from the "merge upwards" rule
112 above:
113 Example 2. Topic branches
115 Make a side branch for every topic (feature, bugfix, ...). Fork it off
117 at the oldest integration branch that you will eventually want to merge
118 it into.
119 Many things can then be done very naturally:
121 · To get the feature/bugfix into an integration branch, simply merge
123 it. If the topic has evolved further in the meantime, merge again.
124 (Note that you do not necessarily have to merge it to the oldest
125 integration branch first. For example, you can first merge a bugfix
126 to next, give it some testing time, and merge to maint when you
127 know it is stable.)
128 · If you find you need new features from the branch other to continue
130 working on your topic, merge other to topic. (However, do not do
131 this "just habitually", see below.)
132 · If you find you forked off the wrong branch and want to move it
134 "back in time", use git-rebase(1).
135 Note that the last point clashes with the other two: a topic that has
137 been merged elsewhere should not be rebased. See the section on
138 RECOVERING FROM UPSTREAM REBASE in git-rebase(1).
139 We should point out that "habitually" (regularly for no real reason)
141 merging an integration branch into your topics — and by extension,
142 merging anything upstream into anything downstream on a regular basis —
143 is frowned upon:
144 Example 3. Merge to downstream only at well-defined points
146 Do not merge to downstream except with a good reason: upstream API
148 changes affect your branch; your branch no longer merges to upstream
149 cleanly; etc.
150 Otherwise, the topic that was merged to suddenly contains more than a
152 single (well-separated) change. The many resulting small merges will
153 greatly clutter up history. Anyone who later investigates the history
154 of a file will have to find out whether that merge affected the topic
155 in development. An upstream might even inadvertently be merged into a
156 "more stable" branch. And so on.
157 Throw-away integration
159 If you followed the last paragraph, you will now have many small topic
160 branches, and occasionally wonder how they interact. Perhaps the result
161 of merging them does not even work? But on the other hand, we want to
162 avoid merging them anywhere "stable" because such merges cannot easily
163 be undone.
164 The solution, of course, is to make a merge that we can undo: merge
166 into a throw-away branch.
167 Example 4. Throw-away integration branches
169 To test the interaction of several topics, merge them into a throw-away
171 branch. You must never base any work on such a branch!
172 If you make it (very) clear that this branch is going to be deleted
174 right after the testing, you can even publish this branch, for example
175 to give the testers a chance to work with it, or other developers a
176 chance to see if their in-progress work will be compatible. git.git has
177 such an official throw-away integration branch called pu.
178 Branch management for a release
180 Assuming you are using the merge approach discussed above, when you are
181 releasing your project you will need to do some additional branch
182 management work.
183 A feature release is created from the master branch, since master
185 tracks the commits that should go into the next feature release.
186 The master branch is supposed to be a superset of maint. If this
188 condition does not hold, then maint contains some commits that are not
189 included on master. The fixes represented by those commits will
190 therefore not be included in your feature release.
191 To verify that master is indeed a superset of maint, use git log:
193 Example 5. Verify master is a superset of maint
195 git log master..maint
197 This command should not list any commits. Otherwise, check out master
199 and merge maint into it.
200 Now you can proceed with the creation of the feature release. Apply a
202 tag to the tip of master indicating the release version:
203 Example 6. Release tagging
205 git tag -s -m "GIT X.Y.Z" vX.Y.Z master
207 You need to push the new tag to a public git server (see "DISTRIBUTED
209 WORKFLOWS" below). This makes the tag available to others tracking your
210 project. The push could also trigger a post-update hook to perform
211 release-related items such as building release tarballs and
212 preformatted documentation pages.
213 Similarly, for a maintenance release, maint is tracking the commits to
215 be released. Therefore, in the steps above simply tag and push maint
216 rather than master.
217 Maintenance branch management after a feature release
219 After a feature release, you need to manage your maintenance branches.
220 First, if you wish to continue to release maintenance fixes for the
222 feature release made before the recent one, then you must create
223 another branch to track commits for that previous release.
224 To do this, the current maintenance branch is copied to another branch
226 named with the previous release version number (e.g. maint-X.Y.(Z-1)
227 where X.Y.Z is the current release).
228 Example 7. Copy maint
230 git branch maint-X.Y.(Z-1) maint
232 The maint branch should now be fast-forwarded to the newly released
234 code so that maintenance fixes can be tracked for the current release:
235 Example 8. Update maint to new release
237 · git checkout maint
239 · git merge --ff-only master
241 If the merge fails because it is not a fast-forward, then it is
243 possible some fixes on maint were missed in the feature release. This
244 will not happen if the content of the branches was verified as
245 described in the previous section.
246 Branch management for next and pu after a feature release
248 After a feature release, the integration branch next may optionally be
249 rewound and rebuilt from the tip of master using the surviving topics
250 on next:
251 Example 9. Rewind and rebuild next
253 · git checkout next
255 · git reset --hard master
257 · git merge ai/topic_in_next1
259 · git merge ai/topic_in_next2
261 · ...
263 The advantage of doing this is that the history of next will be clean.
265 For example, some topics merged into next may have initially looked
266 promising, but were later found to be undesirable or premature. In such
267 a case, the topic is reverted out of next but the fact remains in the
268 history that it was once merged and reverted. By recreating next, you
269 give another incarnation of such topics a clean slate to retry, and a
270 feature release is a good point in history to do so.
271 If you do this, then you should make a public announcement indicating
273 that next was rewound and rebuilt.
274 The same rewind and rebuild process may be followed for pu. A public
276 announcement is not necessary since pu is a throw-away branch, as
277 described above.
278
280 After the last section, you should know how to manage topics. In
281 general, you will not be the only person working on the project, so you
282 will have to share your work.
283 Roughly speaking, there are two important workflows: merge and patch.
285 The important difference is that the merge workflow can propagate full
286 history, including merges, while patches cannot. Both workflows can be
287 used in parallel: in git.git, only subsystem maintainers use the merge
288 workflow, while everyone else sends patches.
289 Note that the maintainer(s) may impose restrictions, such as
291 "Signed-off-by" requirements, that all commits/patches submitted for
292 inclusion must adhere to. Consult your project’s documentation for more
293 information.
294 Merge workflow
296 The merge workflow works by copying branches between upstream and
297 downstream. Upstream can merge contributions into the official history;
298 downstream base their work on the official history.
299 There are three main tools that can be used for this:
301 · git-push(1) copies your branches to a remote repository, usually
303 to one that can be read by all involved parties;
304 · git-fetch(1) that copies remote branches to your repository; and
306 · git-pull(1) that does fetch and merge in one go.
308 Note the last point. Do not use git pull unless you actually want to
310 merge the remote branch.
311 Getting changes out is easy:
313 Example 10. Push/pull: Publishing branches/topics
315 git push <remote> <branch> and tell everyone where they can fetch from.
317 You will still have to tell people by other means, such as mail. (Git
319 provides the git-request-pull(1) to send preformatted pull requests to
320 upstream maintainers to simplify this task.)
321 If you just want to get the newest copies of the integration branches,
323 staying up to date is easy too:
324 Example 11. Push/pull: Staying up to date
326 Use git fetch <remote> or git remote update to stay up to date.
328 Then simply fork your topic branches from the stable remotes as
330 explained earlier.
331 If you are a maintainer and would like to merge other people’s topic
333 branches to the integration branches, they will typically send a
334 request to do so by mail. Such a request looks like
335 Please pull from
337 <url> <branch>
338 In that case, git pull can do the fetch and merge in one go, as
341 follows.
342 Example 12. Push/pull: Merging remote topics
344 git pull <url> <branch>
346 Occasionally, the maintainer may get merge conflicts when he tries to
348 pull changes from downstream. In this case, he can ask downstream to do
349 the merge and resolve the conflicts themselves (perhaps they will know
350 better how to resolve them). It is one of the rare cases where
351 downstream should merge from upstream.
352 Patch workflow
354 If you are a contributor that sends changes upstream in the form of
355 emails, you should use topic branches as usual (see above). Then use
356 git-format-patch(1) to generate the corresponding emails (highly
357 recommended over manually formatting them because it makes the
358 maintainer’s life easier).
359 Example 13. format-patch/am: Publishing branches/topics
361 · git format-patch -M upstream..topic to turn them into preformatted
363 patch files
364 · git send-email --to=<recipient> <patches>
366 See the git-format-patch(1) and git-send-email(1) manpages for further
368 usage notes.
369 If the maintainer tells you that your patch no longer applies to the
371 current upstream, you will have to rebase your topic (you cannot use a
372 merge because you cannot format-patch merges):
373 Example 14. format-patch/am: Keeping topics up to date
375 git pull --rebase <url> <branch>
377 You can then fix the conflicts during the rebase. Presumably you have
379 not published your topic other than by mail, so rebasing it is not a
380 problem.
381 If you receive such a patch series (as maintainer, or perhaps as a
383 reader of the mailing list it was sent to), save the mails to files,
384 create a new topic branch and use git am to import the commits:
385 Example 15. format-patch/am: Importing patches
387 git am < patch
389 One feature worth pointing out is the three-way merge, which can help
391 if you get conflicts: git am -3 will use index information contained in
392 patches to figure out the merge base. See git-am(1) for other options.
393
395 gittutorial(7), git-push(1), git-pull(1), git-merge(1), git-rebase(1),
396 git-format-patch(1), git-send-email(1), git-am(1)
397
399 Part of the git(1) suite.
400Git 1.7.1 08/16/2017 GITWORKFLOWS(7)