rebase: teach rebase --keep-base
[git/git.git] / contrib / subtree / git-subtree.txt
1 git-subtree(1)
2 ==============
3
4 NAME
5 ----
6 git-subtree - Merge subtrees together and split repository into subtrees
7
8
9 SYNOPSIS
10 --------
11 [verse]
12 'git subtree' add -P <prefix> <commit>
13 'git subtree' add -P <prefix> <repository> <ref>
14 'git subtree' pull -P <prefix> <repository> <ref>
15 'git subtree' push -P <prefix> <repository> <ref>
16 'git subtree' merge -P <prefix> <commit>
17 'git subtree' split -P <prefix> [OPTIONS] [<commit>]
18
19
20 DESCRIPTION
21 -----------
22 Subtrees allow subprojects to be included within a subdirectory
23 of the main project, optionally including the subproject's
24 entire history.
25
26 For example, you could include the source code for a library
27 as a subdirectory of your application.
28
29 Subtrees are not to be confused with submodules, which are meant for
30 the same task. Unlike submodules, subtrees do not need any special
31 constructions (like .gitmodules files or gitlinks) be present in
32 your repository, and do not force end-users of your
33 repository to do anything special or to understand how subtrees
34 work. A subtree is just a subdirectory that can be
35 committed to, branched, and merged along with your project in
36 any way you want.
37
38 They are also not to be confused with using the subtree merge
39 strategy. The main difference is that, besides merging
40 the other project as a subdirectory, you can also extract the
41 entire history of a subdirectory from your project and make it
42 into a standalone project. Unlike the subtree merge strategy
43 you can alternate back and forth between these
44 two operations. If the standalone library gets updated, you can
45 automatically merge the changes into your project; if you
46 update the library inside your project, you can "split" the
47 changes back out again and merge them back into the library
48 project.
49
50 For example, if a library you made for one application ends up being
51 useful elsewhere, you can extract its entire history and publish
52 that as its own git repository, without accidentally
53 intermingling the history of your application project.
54
55 [TIP]
56 In order to keep your commit messages clean, we recommend that
57 people split their commits between the subtrees and the main
58 project as much as possible. That is, if you make a change that
59 affects both the library and the main application, commit it in
60 two pieces. That way, when you split the library commits out
61 later, their descriptions will still make sense. But if this
62 isn't important to you, it's not *necessary*. git subtree will
63 simply leave out the non-library-related parts of the commit
64 when it splits it out into the subproject later.
65
66
67 COMMANDS
68 --------
69 add::
70 Create the <prefix> subtree by importing its contents
71 from the given <commit> or <repository> and remote <ref>.
72 A new commit is created automatically, joining the imported
73 project's history with your own. With '--squash', imports
74 only a single commit from the subproject, rather than its
75 entire history.
76
77 merge::
78 Merge recent changes up to <commit> into the <prefix>
79 subtree. As with normal 'git merge', this doesn't
80 remove your own local changes; it just merges those
81 changes into the latest <commit>. With '--squash',
82 creates only one commit that contains all the changes,
83 rather than merging in the entire history.
84 +
85 If you use '--squash', the merge direction doesn't always have to be
86 forward; you can use this command to go back in time from v2.5 to v2.4,
87 for example. If your merge introduces a conflict, you can resolve it in
88 the usual ways.
89
90 pull::
91 Exactly like 'merge', but parallels 'git pull' in that
92 it fetches the given ref from the specified remote
93 repository.
94
95 push::
96 Does a 'split' (see below) using the <prefix> supplied
97 and then does a 'git push' to push the result to the
98 repository and ref. This can be used to push your
99 subtree to different branches of the remote repository.
100
101 split::
102 Extract a new, synthetic project history from the
103 history of the <prefix> subtree. The new history
104 includes only the commits (including merges) that
105 affected <prefix>, and each of those commits now has the
106 contents of <prefix> at the root of the project instead
107 of in a subdirectory. Thus, the newly created history
108 is suitable for export as a separate git repository.
109 +
110 After splitting successfully, a single commit id is printed to stdout.
111 This corresponds to the HEAD of the newly created tree, which you can
112 manipulate however you want.
113 +
114 Repeated splits of exactly the same history are guaranteed to be
115 identical (i.e. to produce the same commit ids). Because of this, if
116 you add new commits and then re-split, the new commits will be attached
117 as commits on top of the history you generated last time, so 'git merge'
118 and friends will work as expected.
119 +
120 Note that if you use '--squash' when you merge, you should usually not
121 just '--rejoin' when you split.
122
123
124 OPTIONS
125 -------
126 -q::
127 --quiet::
128 Suppress unnecessary output messages on stderr.
129
130 -d::
131 --debug::
132 Produce even more unnecessary output messages on stderr.
133
134 -P <prefix>::
135 --prefix=<prefix>::
136 Specify the path in the repository to the subtree you
137 want to manipulate. This option is mandatory
138 for all commands.
139
140 -m <message>::
141 --message=<message>::
142 This option is only valid for add, merge and pull (unsure).
143 Specify <message> as the commit message for the merge commit.
144
145
146 OPTIONS FOR add, merge, push, pull
147 ----------------------------------
148 --squash::
149 This option is only valid for add, merge, and pull
150 commands.
151 +
152 Instead of merging the entire history from the subtree project, produce
153 only a single commit that contains all the differences you want to
154 merge, and then merge that new commit into your project.
155 +
156 Using this option helps to reduce log clutter. People rarely want to see
157 every change that happened between v1.0 and v1.1 of the library they're
158 using, since none of the interim versions were ever included in their
159 application.
160 +
161 Using '--squash' also helps avoid problems when the same subproject is
162 included multiple times in the same project, or is removed and then
163 re-added. In such a case, it doesn't make sense to combine the
164 histories anyway, since it's unclear which part of the history belongs
165 to which subtree.
166 +
167 Furthermore, with '--squash', you can switch back and forth between
168 different versions of a subtree, rather than strictly forward. 'git
169 subtree merge --squash' always adjusts the subtree to match the exactly
170 specified commit, even if getting to that commit would require undoing
171 some changes that were added earlier.
172 +
173 Whether or not you use '--squash', changes made in your local repository
174 remain intact and can be later split and send upstream to the
175 subproject.
176
177
178 OPTIONS FOR split
179 -----------------
180 --annotate=<annotation>::
181 This option is only valid for the split command.
182 +
183 When generating synthetic history, add <annotation> as a prefix to each
184 commit message. Since we're creating new commits with the same commit
185 message, but possibly different content, from the original commits, this
186 can help to differentiate them and avoid confusion.
187 +
188 Whenever you split, you need to use the same <annotation>, or else you
189 don't have a guarantee that the new re-created history will be identical
190 to the old one. That will prevent merging from working correctly. git
191 subtree tries to make it work anyway, particularly if you use --rejoin,
192 but it may not always be effective.
193
194 -b <branch>::
195 --branch=<branch>::
196 This option is only valid for the split command.
197 +
198 After generating the synthetic history, create a new branch called
199 <branch> that contains the new history. This is suitable for immediate
200 pushing upstream. <branch> must not already exist.
201
202 --ignore-joins::
203 This option is only valid for the split command.
204 +
205 If you use '--rejoin', git subtree attempts to optimize its history
206 reconstruction to generate only the new commits since the last
207 '--rejoin'. '--ignore-join' disables this behaviour, forcing it to
208 regenerate the entire history. In a large project, this can take a long
209 time.
210
211 --onto=<onto>::
212 This option is only valid for the split command.
213 +
214 If your subtree was originally imported using something other than git
215 subtree, its history may not match what git subtree is expecting. In
216 that case, you can specify the commit id <onto> that corresponds to the
217 first revision of the subproject's history that was imported into your
218 project, and git subtree will attempt to build its history from there.
219 +
220 If you used 'git subtree add', you should never need this option.
221
222 --rejoin::
223 This option is only valid for the split command.
224 +
225 After splitting, merge the newly created synthetic history back into
226 your main project. That way, future splits can search only the part of
227 history that has been added since the most recent --rejoin.
228 +
229 If your split commits end up merged into the upstream subproject, and
230 then you want to get the latest upstream version, this will allow git's
231 merge algorithm to more intelligently avoid conflicts (since it knows
232 these synthetic commits are already part of the upstream repository).
233 +
234 Unfortunately, using this option results in 'git log' showing an extra
235 copy of every new commit that was created (the original, and the
236 synthetic one).
237 +
238 If you do all your merges with '--squash', don't use '--rejoin' when you
239 split, because you don't want the subproject's history to be part of
240 your project anyway.
241
242
243 EXAMPLE 1. Add command
244 ----------------------
245 Let's assume that you have a local repository that you would like
246 to add an external vendor library to. In this case we will add the
247 git-subtree repository as a subdirectory of your already existing
248 git-extensions repository in ~/git-extensions/:
249
250 $ git subtree add --prefix=git-subtree --squash \
251 git://github.com/apenwarr/git-subtree.git master
252
253 'master' needs to be a valid remote ref and can be a different branch
254 name
255
256 You can omit the --squash flag, but doing so will increase the number
257 of commits that are included in your local repository.
258
259 We now have a ~/git-extensions/git-subtree directory containing code
260 from the master branch of git://github.com/apenwarr/git-subtree.git
261 in our git-extensions repository.
262
263 EXAMPLE 2. Extract a subtree using commit, merge and pull
264 ---------------------------------------------------------
265 Let's use the repository for the git source code as an example.
266 First, get your own copy of the git.git repository:
267
268 $ git clone git://git.kernel.org/pub/scm/git/git.git test-git
269 $ cd test-git
270
271 gitweb (commit 1130ef3) was merged into git as of commit
272 0a8f4f0, after which it was no longer maintained separately.
273 But imagine it had been maintained separately, and we wanted to
274 extract git's changes to gitweb since that time, to share with
275 the upstream. You could do this:
276
277 $ git subtree split --prefix=gitweb --annotate='(split) ' \
278 0a8f4f0^.. --onto=1130ef3 --rejoin \
279 --branch gitweb-latest
280 $ gitk gitweb-latest
281 $ git push git@github.com:whatever/gitweb.git gitweb-latest:master
282
283 (We use '0a8f4f0^..' because that means "all the changes from
284 0a8f4f0 to the current version, including 0a8f4f0 itself.")
285
286 If gitweb had originally been merged using 'git subtree add' (or
287 a previous split had already been done with --rejoin specified)
288 then you can do all your splits without having to remember any
289 weird commit ids:
290
291 $ git subtree split --prefix=gitweb --annotate='(split) ' --rejoin \
292 --branch gitweb-latest2
293
294 And you can merge changes back in from the upstream project just
295 as easily:
296
297 $ git subtree pull --prefix=gitweb \
298 git@github.com:whatever/gitweb.git master
299
300 Or, using '--squash', you can actually rewind to an earlier
301 version of gitweb:
302
303 $ git subtree merge --prefix=gitweb --squash gitweb-latest~10
304
305 Then make some changes:
306
307 $ date >gitweb/myfile
308 $ git add gitweb/myfile
309 $ git commit -m 'created myfile'
310
311 And fast forward again:
312
313 $ git subtree merge --prefix=gitweb --squash gitweb-latest
314
315 And notice that your change is still intact:
316
317 $ ls -l gitweb/myfile
318
319 And you can split it out and look at your changes versus
320 the standard gitweb:
321
322 git log gitweb-latest..$(git subtree split --prefix=gitweb)
323
324 EXAMPLE 3. Extract a subtree using branch
325 -----------------------------------------
326 Suppose you have a source directory with many files and
327 subdirectories, and you want to extract the lib directory to its own
328 git project. Here's a short way to do it:
329
330 First, make the new repository wherever you want:
331
332 $ <go to the new location>
333 $ git init --bare
334
335 Back in your original directory:
336
337 $ git subtree split --prefix=lib --annotate="(split)" -b split
338
339 Then push the new branch onto the new empty repository:
340
341 $ git push <new-repo> split:master
342
343
344 AUTHOR
345 ------
346 Written by Avery Pennarun <apenwarr@gmail.com>
347
348
349 GIT
350 ---
351 Part of the linkgit:git[1] suite