6 git-rebase - Forward-port local commits to the updated upstream head
11 'git rebase' [-i | --interactive] [options] [--onto <newbase>]
13 'git rebase' [-i | --interactive] [options] --onto <newbase>
16 'git rebase' --continue | --skip | --abort
20 If <branch> is specified, 'git rebase' will perform an automatic
21 `git checkout <branch>` before doing anything else. Otherwise
22 it remains on the current branch.
24 All changes made by commits in the current branch but that are not
25 in <upstream> are saved to a temporary area. This is the same set
26 of commits that would be shown by `git log <upstream>..HEAD` (or
27 `git log HEAD`, if --root is specified).
29 The current branch is reset to <upstream>, or <newbase> if the
30 --onto option was supplied. This has the exact same effect as
31 `git reset --hard <upstream>` (or <newbase>). ORIG_HEAD is set
32 to point at the tip of the branch before the reset.
34 The commits that were previously saved into the temporary area are
35 then reapplied to the current branch, one by one, in order. Note that
36 any commits in HEAD which introduce the same textual changes as a commit
37 in HEAD..<upstream> are omitted (i.e., a patch already accepted upstream
38 with a different commit message or timestamp will be skipped).
40 It is possible that a merge failure will prevent this process from being
41 completely automatic. You will have to resolve any such merge failure
42 and run `git rebase --continue`. Another option is to bypass the commit
43 that caused the merge failure with `git rebase --skip`. To restore the
44 original <branch> and remove the .git/rebase-apply working files, use the
45 command `git rebase --abort` instead.
47 Assume the following history exists and the current branch is "topic":
55 From this point, the result of either of the following commands:
59 git rebase master topic
69 The latter form is just a short-hand of `git checkout topic`
70 followed by `git rebase master`.
72 If the upstream branch already contains a change you have made (e.g.,
73 because you mailed a patch which was applied upstream), then that commit
74 will be skipped. For example, running `git rebase master` on the
75 following history (in which A' and A introduce the same set of changes,
76 but have different committer information):
92 Here is how you would transplant a topic branch based on one
93 branch to another, to pretend that you forked the topic branch
94 from the latter branch, using `rebase --onto`.
96 First let's assume your 'topic' is based on branch 'next'.
97 For example, a feature developed in 'topic' depends on some
98 functionality which is found in 'next'.
101 o---o---o---o---o master
103 o---o---o---o---o next
108 We want to make 'topic' forked from branch 'master'; for example,
109 because the functionality on which 'topic' depends was merged into the
110 more stable 'master' branch. We want our tree to look like this:
113 o---o---o---o---o master
117 o---o---o---o---o next
120 We can get this using the following command:
122 git rebase --onto master next topic
125 Another example of --onto option is to rebase part of a
126 branch. If we have the following situation:
138 git rebase --onto master topicA topicB
150 This is useful when topicB does not depend on topicA.
152 A range of commits could also be removed with rebase. If we have
153 the following situation:
156 E---F---G---H---I---J topicA
161 git rebase --onto topicA~5 topicA~3 topicA
163 would result in the removal of commits F and G:
166 E---H'---I'---J' topicA
169 This is useful if F and G were flawed in some way, or should not be
170 part of topicA. Note that the argument to --onto and the <upstream>
171 parameter can be any valid commit-ish.
173 In case of conflict, 'git rebase' will stop at the first problematic commit
174 and leave conflict markers in the tree. You can use 'git diff' to locate
175 the markers (<<<<<<) and make edits to resolve the conflict. For each
176 file you edit, you need to tell git that the conflict has been resolved,
177 typically this would be done with
183 After resolving the conflict manually and updating the index with the
184 desired resolution, you can continue the rebasing process with
187 git rebase --continue
190 Alternatively, you can undo the 'git rebase' with
199 Whether to show a diffstat of what changed upstream since the last
200 rebase. False by default.
205 Starting point at which to create the new commits. If the
206 --onto option is not specified, the starting point is
207 <upstream>. May be any valid commit, and not just an
208 existing branch name.
211 Upstream branch to compare against. May be any valid commit,
212 not just an existing branch name.
215 Working branch; defaults to HEAD.
218 Restart the rebasing process after having resolved a merge conflict.
221 Restore the original branch and abort the rebase operation.
224 Restart the rebasing process by skipping the current patch.
228 Use merging strategies to rebase. When the recursive (default) merge
229 strategy is used, this allows rebase to be aware of renames on the
232 Note that a rebase merge works by replaying each commit from the working
233 branch on top of the <upstream> branch. Because of this, when a merge
234 conflict happens, the side reported as 'ours' is the so-far rebased
235 series, starting with <upstream>, and 'theirs' is the working branch. In
236 other words, the sides are swapped.
239 --strategy=<strategy>::
240 Use the given merge strategy.
241 If there is no `-s` option 'git merge-recursive' is used
242 instead. This implies --merge.
244 Because 'git rebase' replays each commit from the working branch
245 on top of the <upstream> branch using the given strategy, using
246 the 'ours' strategy simply discards all patches from the <branch>,
247 which makes little sense.
251 Be quiet. Implies --no-stat.
255 Be verbose. Implies --stat.
258 Show a diffstat of what changed upstream since the last rebase. The
259 diffstat is also controlled by the configuration option rebase.stat.
263 Do not show a diffstat as part of the rebase process.
266 This option bypasses the pre-rebase hook. See also linkgit:githooks[5].
269 Ensure at least <n> lines of surrounding context match before
270 and after each change. When fewer lines of surrounding
271 context exist they all must match. By default no context is
276 Force the rebase even if the current branch is a descendant
277 of the commit you are rebasing onto. Normally the command will
278 exit with the message "Current branch is up to date" in such a
281 --ignore-whitespace::
282 --whitespace=<option>::
283 These flag are passed to the 'git apply' program
284 (see linkgit:git-apply[1]) that applies the patch.
285 Incompatible with the --interactive option.
287 --committer-date-is-author-date::
289 These flags are passed to 'git am' to easily change the dates
290 of the rebased commits (see linkgit:git-am[1]).
294 Make a list of the commits which are about to be rebased. Let the
295 user edit that list before rebasing. This mode can also be used to
296 split commits (see SPLITTING COMMITS below).
300 Instead of ignoring merges, try to recreate them.
303 Rebase all commits reachable from <branch>, instead of
304 limiting them with an <upstream>. This allows you to rebase
305 the root commit(s) on a branch. Must be used with --onto, and
306 will skip changes already contained in <newbase> (instead of
307 <upstream>). When used together with --preserve-merges, 'all'
308 root commits will be rewritten to have <newbase> as parent
311 include::merge-strategies.txt[]
316 You should understand the implications of using 'git rebase' on a
317 repository that you share. See also RECOVERING FROM UPSTREAM REBASE
320 When the git-rebase command is run, it will first execute a "pre-rebase"
321 hook if one exists. You can use this hook to do sanity checks and
322 reject the rebase if it isn't appropriate. Please see the template
323 pre-rebase hook script for an example.
325 Upon completion, <branch> will be the current branch.
330 Rebasing interactively means that you have a chance to edit the commits
331 which are rebased. You can reorder the commits, and you can
332 remove them (weeding out bad or otherwise unwanted patches).
334 The interactive mode is meant for this type of workflow:
336 1. have a wonderful idea
338 3. prepare a series for submission
341 where point 2. consists of several instances of
344 1. finish something worthy of a commit
347 1. realize that something does not work
351 Sometimes the thing fixed in b.2. cannot be amended to the not-quite
352 perfect commit it fixes, because that commit is buried deeply in a
353 patch series. That is exactly what interactive rebase is for: use it
354 after plenty of "a"s and "b"s, by rearranging and editing
355 commits, and squashing multiple commits into one.
357 Start it with the last commit you want to retain as-is:
359 git rebase -i <after-this-commit>
361 An editor will be fired up with all the commits in your current branch
362 (ignoring merge commits), which come after the given commit. You can
363 reorder the commits in this list to your heart's content, and you can
364 remove them. The list looks more or less like this:
366 -------------------------------------------
367 pick deadbee The oneline of this commit
368 pick fa1afe1 The oneline of the next commit
370 -------------------------------------------
372 The oneline descriptions are purely for your pleasure; 'git rebase' will
373 not look at them but at the commit names ("deadbee" and "fa1afe1" in this
374 example), so do not delete or edit the names.
376 By replacing the command "pick" with the command "edit", you can tell
377 'git rebase' to stop after applying that commit, so that you can edit
378 the files and/or the commit message, amend the commit, and continue
381 If you just want to edit the commit message for a commit, replace the
382 command "pick" with the command "reword".
384 If you want to fold two or more commits into one, replace the command
385 "pick" with "squash" for the second and subsequent commit. If the
386 commits had different authors, it will attribute the squashed commit to
387 the author of the first commit.
389 'git rebase' will stop when "pick" has been replaced with "edit" or
390 when a command fails due to merge errors. When you are done editing
391 and/or resolving conflicts you can continue with `git rebase --continue`.
393 For example, if you want to reorder the last 5 commits, such that what
394 was HEAD~4 becomes the new HEAD. To achieve that, you would call
395 'git rebase' like this:
397 ----------------------
398 $ git rebase -i HEAD~5
399 ----------------------
401 And move the first patch to the end of the list.
403 You might want to preserve merges, if you have a history like this:
413 Suppose you want to rebase the side branch starting at "A" to "Q". Make
414 sure that the current HEAD is "B", and call
416 -----------------------------
417 $ git rebase -i -p --onto Q O
418 -----------------------------
424 In interactive mode, you can mark commits with the action "edit". However,
425 this does not necessarily mean that 'git rebase' expects the result of this
426 edit to be exactly one commit. Indeed, you can undo the commit, or you can
427 add other commits. This can be used to split a commit into two:
429 - Start an interactive rebase with `git rebase -i <commit>^`, where
430 <commit> is the commit you want to split. In fact, any commit range
431 will do, as long as it contains that commit.
433 - Mark the commit you want to split with the action "edit".
435 - When it comes to editing that commit, execute `git reset HEAD^`. The
436 effect is that the HEAD is rewound by one, and the index follows suit.
437 However, the working tree stays the same.
439 - Now add the changes to the index that you want to have in the first
440 commit. You can use `git add` (possibly interactively) or
441 'git gui' (or both) to do that.
443 - Commit the now-current index with whatever commit message is appropriate
446 - Repeat the last two steps until your working tree is clean.
448 - Continue the rebase with `git rebase --continue`.
450 If you are not absolutely sure that the intermediate revisions are
451 consistent (they compile, pass the testsuite, etc.) you should use
452 'git stash' to stash away the not-yet-committed changes
453 after each commit, test, and amend the commit if fixes are necessary.
456 RECOVERING FROM UPSTREAM REBASE
457 -------------------------------
459 Rebasing (or any other form of rewriting) a branch that others have
460 based work on is a bad idea: anyone downstream of it is forced to
461 manually fix their history. This section explains how to do the fix
462 from the downstream's point of view. The real fix, however, would be
463 to avoid rebasing the upstream in the first place.
465 To illustrate, suppose you are in a situation where someone develops a
466 'subsystem' branch, and you are working on a 'topic' that is dependent
467 on this 'subsystem'. You might end up with a history like the
471 o---o---o---o---o---o---o---o---o master
473 o---o---o---o---o subsystem
478 If 'subsystem' is rebased against 'master', the following happens:
481 o---o---o---o---o---o---o---o master
483 o---o---o---o---o o'--o'--o'--o'--o' subsystem
488 If you now continue development as usual, and eventually merge 'topic'
489 to 'subsystem', the commits from 'subsystem' will remain duplicated forever:
492 o---o---o---o---o---o---o---o master
494 o---o---o---o---o o'--o'--o'--o'--o'--M subsystem
496 *---*---*-..........-*--* topic
499 Such duplicates are generally frowned upon because they clutter up
500 history, making it harder to follow. To clean things up, you need to
501 transplant the commits on 'topic' to the new 'subsystem' tip, i.e.,
502 rebase 'topic'. This becomes a ripple effect: anyone downstream from
503 'topic' is forced to rebase too, and so on!
505 There are two kinds of fixes, discussed in the following subsections:
507 Easy case: The changes are literally the same.::
509 This happens if the 'subsystem' rebase was a simple rebase and
512 Hard case: The changes are not the same.::
514 This happens if the 'subsystem' rebase had conflicts, or used
515 `\--interactive` to omit, edit, or squash commits; or if the
516 upstream used one of `commit \--amend`, `reset`, or
523 Only works if the changes (patch IDs based on the diff contents) on
524 'subsystem' are literally the same before and after the rebase
527 In that case, the fix is easy because 'git rebase' knows to skip
528 changes that are already present in the new upstream. So if you say
529 (assuming you're on 'topic')
531 $ git rebase subsystem
533 you will end up with the fixed history
535 o---o---o---o---o---o---o---o master
537 o'--o'--o'--o'--o' subsystem
546 Things get more complicated if the 'subsystem' changes do not exactly
547 correspond to the ones before the rebase.
549 NOTE: While an "easy case recovery" sometimes appears to be successful
550 even in the hard case, it may have unintended consequences. For
551 example, a commit that was removed via `git rebase
552 \--interactive` will be **resurrected**!
554 The idea is to manually tell 'git rebase' "where the old 'subsystem'
555 ended and your 'topic' began", that is, what the old merge-base
556 between them was. You will have to find a way to name the last commit
557 of the old 'subsystem', for example:
559 * With the 'subsystem' reflog: after 'git fetch', the old tip of
560 'subsystem' is at `subsystem@\{1}`. Subsequent fetches will
561 increase the number. (See linkgit:git-reflog[1].)
563 * Relative to the tip of 'topic': knowing that your 'topic' has three
564 commits, the old tip of 'subsystem' must be `topic~3`.
566 You can then transplant the old `subsystem..topic` to the new tip by
567 saying (for the reflog case, and assuming you are on 'topic' already):
569 $ git rebase --onto subsystem subsystem@{1}
572 The ripple effect of a "hard case" recovery is especially bad:
573 'everyone' downstream from 'topic' will now have to perform a "hard
579 Written by Junio C Hamano <gitster@pobox.com> and
580 Johannes E. Schindelin <johannes.schindelin@gmx.de>
584 Documentation by Junio C Hamano and the git-list <git@vger.kernel.org>.
588 Part of the linkgit:git[1] suite