1GIT-REBASE(1) Git Manual GIT-REBASE(1)
2
6 git-rebase - Forward-port local commits to the updated upstream head
7
9 git rebase [-i | --interactive] [options] [--exec <cmd>] [--onto <newbase>]
10 [<upstream>] [<branch>]
11 git rebase [-i | --interactive] [options] [--exec <cmd>] [--onto <newbase>]
12 --root [<branch>]
13 git rebase --continue | --skip | --abort | --edit-todo
14
17 If <branch> is specified, git rebase will perform an automatic git
18 checkout <branch> before doing anything else. Otherwise it remains on
19 the current branch.
20 If <upstream> is not specified, the upstream configured in
22 branch.<name>.remote and branch.<name>.merge options will be used; see
23 git-config(1) for details. If you are currently not on any branch or if
24 the current branch does not have a configured upstream, the rebase will
25 abort.
26 All changes made by commits in the current branch but that are not in
28 <upstream> are saved to a temporary area. This is the same set of
29 commits that would be shown by git log <upstream>..HEAD (or git log
30 HEAD, if --root is specified).
31 The current branch is reset to <upstream>, or <newbase> if the --onto
33 option was supplied. This has the exact same effect as git reset --hard
34 <upstream> (or <newbase>). ORIG_HEAD is set to point at the tip of the
35 branch before the reset.
36 The commits that were previously saved into the temporary area are then
38 reapplied to the current branch, one by one, in order. Note that any
39 commits in HEAD which introduce the same textual changes as a commit in
40 HEAD..<upstream> are omitted (i.e., a patch already accepted upstream
41 with a different commit message or timestamp will be skipped).
42 It is possible that a merge failure will prevent this process from
44 being completely automatic. You will have to resolve any such merge
45 failure and run git rebase --continue. Another option is to bypass the
46 commit that caused the merge failure with git rebase --skip. To check
47 out the original <branch> and remove the .git/rebase-apply working
48 files, use the command git rebase --abort instead.
49 Assume the following history exists and the current branch is "topic":
51 A---B---C topic
53 /
54 D---E---F---G master
55 From this point, the result of either of the following commands:
58 git rebase master
60 git rebase master topic
61 would be:
63 A'--B'--C' topic
65 /
66 D---E---F---G master
67 NOTE: The latter form is just a short-hand of git checkout topic
70 followed by git rebase master. When rebase exits topic will remain the
71 checked-out branch.
72 If the upstream branch already contains a change you have made (e.g.,
74 because you mailed a patch which was applied upstream), then that
75 commit will be skipped. For example, running ‘git rebase master` on the
76 following history (in which A’ and A introduce the same set of changes,
77 but have different committer information):
78 A---B---C topic
80 /
81 D---E---A'---F master
82 will result in:
85 B'---C' topic
87 /
88 D---E---A'---F master
89 Here is how you would transplant a topic branch based on one branch to
92 another, to pretend that you forked the topic branch from the latter
93 branch, using rebase --onto.
94 First let’s assume your topic is based on branch next. For example, a
96 feature developed in topic depends on some functionality which is found
97 in next.
98 o---o---o---o---o master
100 \
101 o---o---o---o---o next
102 \
103 o---o---o topic
104 We want to make topic forked from branch master; for example, because
107 the functionality on which topic depends was merged into the more
108 stable master branch. We want our tree to look like this:
109 o---o---o---o---o master
111 | \
112 | o'--o'--o' topic
113 \
114 o---o---o---o---o next
115 We can get this using the following command:
118 git rebase --onto master next topic
120 Another example of --onto option is to rebase part of a branch. If we
122 have the following situation:
123 H---I---J topicB
125 /
126 E---F---G topicA
127 /
128 A---B---C---D master
129 then the command
132 git rebase --onto master topicA topicB
134 would result in:
136 H'--I'--J' topicB
138 /
139 | E---F---G topicA
140 |/
141 A---B---C---D master
142 This is useful when topicB does not depend on topicA.
145 A range of commits could also be removed with rebase. If we have the
147 following situation:
148 E---F---G---H---I---J topicA
150 then the command
153 git rebase --onto topicA~5 topicA~3 topicA
155 would result in the removal of commits F and G:
157 E---H'---I'---J' topicA
159 This is useful if F and G were flawed in some way, or should not be
162 part of topicA. Note that the argument to --onto and the <upstream>
163 parameter can be any valid commit-ish.
164 In case of conflict, git rebase will stop at the first problematic
166 commit and leave conflict markers in the tree. You can use git diff to
167 locate the markers (<<<<<<) and make edits to resolve the conflict. For
168 each file you edit, you need to tell Git that the conflict has been
169 resolved, typically this would be done with
170 git add <filename>
172 After resolving the conflict manually and updating the index with the
174 desired resolution, you can continue the rebasing process with
175 git rebase --continue
177 Alternatively, you can undo the git rebase with
179 git rebase --abort
181
183 rebase.stat
184 Whether to show a diffstat of what changed upstream since the last
185 rebase. False by default.
186 rebase.autosquash
188 If set to true enable --autosquash option by default.
189
191 --onto <newbase>
192 Starting point at which to create the new commits. If the --onto
193 option is not specified, the starting point is <upstream>. May be
194 any valid commit, and not just an existing branch name.
195 As a special case, you may use "A...B" as a shortcut for the merge
197 base of A and B if there is exactly one merge base. You can leave
198 out at most one of A and B, in which case it defaults to HEAD.
199 <upstream>
201 Upstream branch to compare against. May be any valid commit, not
202 just an existing branch name. Defaults to the configured upstream
203 for the current branch.
204 <branch>
206 Working branch; defaults to HEAD.
207 --continue
209 Restart the rebasing process after having resolved a merge
210 conflict.
211 --abort
213 Abort the rebase operation and reset HEAD to the original branch.
214 If <branch> was provided when the rebase operation was started,
215 then HEAD will be reset to <branch>. Otherwise HEAD will be reset
216 to where it was when the rebase operation was started.
217 --keep-empty
219 Keep the commits that do not change anything from its parents in
220 the result.
221 --skip
223 Restart the rebasing process by skipping the current patch.
224 --edit-todo
226 Edit the todo list during an interactive rebase.
227 -m, --merge
229 Use merging strategies to rebase. When the recursive (default)
230 merge strategy is used, this allows rebase to be aware of renames
231 on the upstream side.
232 Note that a rebase merge works by replaying each commit from the
234 working branch on top of the <upstream> branch. Because of this,
235 when a merge conflict happens, the side reported as ours is the
236 so-far rebased series, starting with <upstream>, and theirs is the
237 working branch. In other words, the sides are swapped.
238 -s <strategy>, --strategy=<strategy>
240 Use the given merge strategy. If there is no -s option git
241 merge-recursive is used instead. This implies --merge.
242 Because git rebase replays each commit from the working branch on
244 top of the <upstream> branch using the given strategy, using the
245 ours strategy simply discards all patches from the <branch>, which
246 makes little sense.
247 -X <strategy-option>, --strategy-option=<strategy-option>
249 Pass the <strategy-option> through to the merge strategy. This
250 implies --merge and, if no strategy has been specified, -s
251 recursive. Note the reversal of ours and theirs as noted above for
252 the -m option.
253 -q, --quiet
255 Be quiet. Implies --no-stat.
256 -v, --verbose
258 Be verbose. Implies --stat.
259 --stat
261 Show a diffstat of what changed upstream since the last rebase. The
262 diffstat is also controlled by the configuration option
263 rebase.stat.
264 -n, --no-stat
266 Do not show a diffstat as part of the rebase process.
267 --no-verify
269 This option bypasses the pre-rebase hook. See also githooks(5).
270 --verify
272 Allows the pre-rebase hook to run, which is the default. This
273 option can be used to override --no-verify. See also githooks(5).
274 -C<n>
276 Ensure at least <n> lines of surrounding context match before and
277 after each change. When fewer lines of surrounding context exist
278 they all must match. By default no context is ever ignored.
279 -f, --force-rebase
281 Force the rebase even if the current branch is a descendant of the
282 commit you are rebasing onto. Normally non-interactive rebase will
283 exit with the message "Current branch is up to date" in such a
284 situation. Incompatible with the --interactive option.
285 You may find this (or --no-ff with an interactive rebase) helpful
287 after reverting a topic branch merge, as this option recreates the
288 topic branch with fresh commits so it can be remerged successfully
289 without needing to "revert the reversion" (see the
290 revert-a-faulty-merge How-To[1] for details).
291 --ignore-whitespace, --whitespace=<option>
293 These flag are passed to the git apply program (see git-apply(1))
294 that applies the patch. Incompatible with the --interactive option.
295 --committer-date-is-author-date, --ignore-date
297 These flags are passed to git am to easily change the dates of the
298 rebased commits (see git-am(1)). Incompatible with the
299 --interactive option.
300 -i, --interactive
302 Make a list of the commits which are about to be rebased. Let the
303 user edit that list before rebasing. This mode can also be used to
304 split commits (see SPLITTING COMMITS below).
305 -p, --preserve-merges
307 Instead of ignoring merges, try to recreate them.
308 This uses the --interactive machinery internally, but combining it
310 with the --interactive option explicitly is generally not a good
311 idea unless you know what you are doing (see BUGS below).
312 -x <cmd>, --exec <cmd>
314 Append "exec <cmd>" after each line creating a commit in the final
315 history. <cmd> will be interpreted as one or more shell commands.
316 This option can only be used with the --interactive option (see
318 INTERACTIVE MODE below).
319 You may execute several commands by either using one instance of
321 --exec with several commands:
322 git rebase -i --exec "cmd1 && cmd2 && ..."
324 or by giving more than one --exec:
326 git rebase -i --exec "cmd1" --exec "cmd2" --exec ...
328 If --autosquash is used, "exec" lines will not be appended for the
330 intermediate commits, and will only appear at the end of each
331 squash/fixup series.
332 --root
334 Rebase all commits reachable from <branch>, instead of limiting
335 them with an <upstream>. This allows you to rebase the root
336 commit(s) on a branch. When used with --onto, it will skip changes
337 already contained in <newbase> (instead of <upstream>) whereas
338 without --onto it will operate on every change. When used together
339 with both --onto and --preserve-merges, all root commits will be
340 rewritten to have <newbase> as parent instead.
341 --autosquash, --no-autosquash
343 When the commit log message begins with "squash! ..." (or "fixup!
344 ..."), and there is a commit whose title begins with the same ...,
345 automatically modify the todo list of rebase -i so that the commit
346 marked for squashing comes right after the commit to be modified,
347 and change the action of the moved commit from pick to squash (or
348 fixup).
349 This option is only valid when the --interactive option is used.
351 If the --autosquash option is enabled by default using the
353 configuration variable rebase.autosquash, this option can be used
354 to override and disable this setting.
355 --no-ff
357 With --interactive, cherry-pick all rebased commits instead of
358 fast-forwarding over the unchanged ones. This ensures that the
359 entire history of the rebased branch is composed of new commits.
360 Without --interactive, this is a synonym for --force-rebase.
362 You may find this helpful after reverting a topic branch merge, as
364 this option recreates the topic branch with fresh commits so it can
365 be remerged successfully without needing to "revert the reversion"
366 (see the revert-a-faulty-merge How-To[1] for details).
367
369 The merge mechanism (git-merge and git-pull commands) allows the
370 backend merge strategies to be chosen with -s option. Some strategies
371 can also take their own options, which can be passed by giving
372 -X<option> arguments to git-merge and/or git-pull.
373 resolve
375 This can only resolve two heads (i.e. the current branch and
376 another branch you pulled from) using a 3-way merge algorithm. It
377 tries to carefully detect criss-cross merge ambiguities and is
378 considered generally safe and fast.
379 recursive
381 This can only resolve two heads using a 3-way merge algorithm. When
382 there is more than one common ancestor that can be used for 3-way
383 merge, it creates a merged tree of the common ancestors and uses
384 that as the reference tree for the 3-way merge. This has been
385 reported to result in fewer merge conflicts without causing
386 mis-merges by tests done on actual merge commits taken from Linux
387 2.6 kernel development history. Additionally this can detect and
388 handle merges involving renames. This is the default merge strategy
389 when pulling or merging one branch.
390 The recursive strategy can take the following options:
392 ours
394 This option forces conflicting hunks to be auto-resolved
395 cleanly by favoring our version. Changes from the other tree
396 that do not conflict with our side are reflected to the merge
397 result. For a binary file, the entire contents are taken from
398 our side.
399 This should not be confused with the ours merge strategy, which
401 does not even look at what the other tree contains at all. It
402 discards everything the other tree did, declaring our history
403 contains all that happened in it.
404 theirs
406 This is the opposite of ours.
407 patience
409 With this option, merge-recursive spends a little extra time to
410 avoid mismerges that sometimes occur due to unimportant
411 matching lines (e.g., braces from distinct functions). Use this
412 when the branches to be merged have diverged wildly. See also
413 git-diff(1)--patience.
414 diff-algorithm=[patience|minimal|histogram|myers]
416 Tells merge-recursive to use a different diff algorithm, which
417 can help avoid mismerges that occur due to unimportant matching
418 lines (such as braces from distinct functions). See also git-
419 diff(1)--diff-algorithm.
420 ignore-space-change, ignore-all-space, ignore-space-at-eol
422 Treats lines with the indicated type of whitespace change as
423 unchanged for the sake of a three-way merge. Whitespace changes
424 mixed with other changes to a line are not ignored. See also
425 git-diff(1)-b, -w, and --ignore-space-at-eol.
426 · If their version only introduces whitespace changes to a
428 line, our version is used;
429 · If our version introduces whitespace changes but their
431 version includes a substantial change, their version is
432 used;
433 · Otherwise, the merge proceeds in the usual way.
435 renormalize
437 This runs a virtual check-out and check-in of all three stages
438 of a file when resolving a three-way merge. This option is
439 meant to be used when merging branches with different clean
440 filters or end-of-line normalization rules. See "Merging
441 branches with differing checkin/checkout attributes" in
442 gitattributes(5) for details.
443 no-renormalize
445 Disables the renormalize option. This overrides the
446 merge.renormalize configuration variable.
447 rename-threshold=<n>
449 Controls the similarity threshold used for rename detection.
450 See also git-diff(1)-M.
451 subtree[=<path>]
453 This option is a more advanced form of subtree strategy, where
454 the strategy makes a guess on how two trees must be shifted to
455 match with each other when merging. Instead, the specified path
456 is prefixed (or stripped from the beginning) to make the shape
457 of two trees to match.
458 octopus
460 This resolves cases with more than two heads, but refuses to do a
461 complex merge that needs manual resolution. It is primarily meant
462 to be used for bundling topic branch heads together. This is the
463 default merge strategy when pulling or merging more than one
464 branch.
465 ours
467 This resolves any number of heads, but the resulting tree of the
468 merge is always that of the current branch head, effectively
469 ignoring all changes from all other branches. It is meant to be
470 used to supersede old development history of side branches. Note
471 that this is different from the -Xours option to the recursive
472 merge strategy.
473 subtree
475 This is a modified recursive strategy. When merging trees A and B,
476 if B corresponds to a subtree of A, B is first adjusted to match
477 the tree structure of A, instead of reading the trees at the same
478 level. This adjustment is also done to the common ancestor tree.
479
481 You should understand the implications of using git rebase on a
482 repository that you share. See also RECOVERING FROM UPSTREAM REBASE
483 below.
484 When the git-rebase command is run, it will first execute a
486 "pre-rebase" hook if one exists. You can use this hook to do sanity
487 checks and reject the rebase if it isn’t appropriate. Please see the
488 template pre-rebase hook script for an example.
489 Upon completion, <branch> will be the current branch.
491
493 Rebasing interactively means that you have a chance to edit the commits
494 which are rebased. You can reorder the commits, and you can remove them
495 (weeding out bad or otherwise unwanted patches).
496 The interactive mode is meant for this type of workflow:
498 1. have a wonderful idea
500 2. hack on the code
502 3. prepare a series for submission
504 4. submit
506 where point 2. consists of several instances of
508 a) regular use
510 1. finish something worthy of a commit
512 2. commit
514 b) independent fixup
516 1. realize that something does not work
518 2. fix that
520 3. commit it
522 Sometimes the thing fixed in b.2. cannot be amended to the not-quite
524 perfect commit it fixes, because that commit is buried deeply in a
525 patch series. That is exactly what interactive rebase is for: use it
526 after plenty of "a"s and "b"s, by rearranging and editing commits, and
527 squashing multiple commits into one.
528 Start it with the last commit you want to retain as-is:
530 git rebase -i <after-this-commit>
532 An editor will be fired up with all the commits in your current branch
534 (ignoring merge commits), which come after the given commit. You can
535 reorder the commits in this list to your heart’s content, and you can
536 remove them. The list looks more or less like this:
537 pick deadbee The oneline of this commit
539 pick fa1afe1 The oneline of the next commit
540 ...
541 The oneline descriptions are purely for your pleasure; git rebase will
544 not look at them but at the commit names ("deadbee" and "fa1afe1" in
545 this example), so do not delete or edit the names.
546 By replacing the command "pick" with the command "edit", you can tell
548 git rebase to stop after applying that commit, so that you can edit the
549 files and/or the commit message, amend the commit, and continue
550 rebasing.
551 If you just want to edit the commit message for a commit, replace the
553 command "pick" with the command "reword".
554 If you want to fold two or more commits into one, replace the command
556 "pick" for the second and subsequent commits with "squash" or "fixup".
557 If the commits had different authors, the folded commit will be
558 attributed to the author of the first commit. The suggested commit
559 message for the folded commit is the concatenation of the commit
560 messages of the first commit and of those with the "squash" command,
561 but omits the commit messages of commits with the "fixup" command.
562 git rebase will stop when "pick" has been replaced with "edit" or when
564 a command fails due to merge errors. When you are done editing and/or
565 resolving conflicts you can continue with git rebase --continue.
566 For example, if you want to reorder the last 5 commits, such that what
568 was HEAD~4 becomes the new HEAD. To achieve that, you would call git
569 rebase like this:
570 $ git rebase -i HEAD~5
572 And move the first patch to the end of the list.
575 You might want to preserve merges, if you have a history like this:
577 X
579 \
580 A---M---B
581 /
582 ---o---O---P---Q
583 Suppose you want to rebase the side branch starting at "A" to "Q". Make
586 sure that the current HEAD is "B", and call
587 $ git rebase -i -p --onto Q O
589 Reordering and editing commits usually creates untested intermediate
592 steps. You may want to check that your history editing did not break
593 anything by running a test, or at least recompiling at intermediate
594 points in history by using the "exec" command (shortcut "x"). You may
595 do so by creating a todo list like this one:
596 pick deadbee Implement feature XXX
598 fixup f1a5c00 Fix to feature XXX
599 exec make
600 pick c0ffeee The oneline of the next commit
601 edit deadbab The oneline of the commit after
602 exec cd subdir; make test
603 ...
604 The interactive rebase will stop when a command fails (i.e. exits with
607 non-0 status) to give you an opportunity to fix the problem. You can
608 continue with git rebase --continue.
609 The "exec" command launches the command in a shell (the one specified
611 in $SHELL, or the default shell if $SHELL is not set), so you can use
612 shell features (like "cd", ">", ";" ...). The command is run from the
613 root of the working tree.
614 $ git rebase -i --exec "make test"
616 This command lets you check that intermediate commits are compilable.
619 The todo list becomes like that:
620 pick 5928aea one
622 exec make test
623 pick 04d0fda two
624 exec make test
625 pick ba46169 three
626 exec make test
627 pick f4593f9 four
628 exec make test
629
632 In interactive mode, you can mark commits with the action "edit".
633 However, this does not necessarily mean that git rebase expects the
634 result of this edit to be exactly one commit. Indeed, you can undo the
635 commit, or you can add other commits. This can be used to split a
636 commit into two:
637 · Start an interactive rebase with git rebase -i <commit>^, where
639 <commit> is the commit you want to split. In fact, any commit range
640 will do, as long as it contains that commit.
641 · Mark the commit you want to split with the action "edit".
643 · When it comes to editing that commit, execute git reset HEAD^. The
645 effect is that the HEAD is rewound by one, and the index follows
646 suit. However, the working tree stays the same.
647 · Now add the changes to the index that you want to have in the first
649 commit. You can use git add (possibly interactively) or git gui (or
650 both) to do that.
651 · Commit the now-current index with whatever commit message is
653 appropriate now.
654 · Repeat the last two steps until your working tree is clean.
656 · Continue the rebase with git rebase --continue.
658 If you are not absolutely sure that the intermediate revisions are
660 consistent (they compile, pass the testsuite, etc.) you should use git
661 stash to stash away the not-yet-committed changes after each commit,
662 test, and amend the commit if fixes are necessary.
663
665 Rebasing (or any other form of rewriting) a branch that others have
666 based work on is a bad idea: anyone downstream of it is forced to
667 manually fix their history. This section explains how to do the fix
668 from the downstream’s point of view. The real fix, however, would be to
669 avoid rebasing the upstream in the first place.
670 To illustrate, suppose you are in a situation where someone develops a
672 subsystem branch, and you are working on a topic that is dependent on
673 this subsystem. You might end up with a history like the following:
674 o---o---o---o---o---o---o---o---o master
676 \
677 o---o---o---o---o subsystem
678 \
679 *---*---* topic
680 If subsystem is rebased against master, the following happens:
683 o---o---o---o---o---o---o---o master
685 \ \
686 o---o---o---o---o o'--o'--o'--o'--o' subsystem
687 \
688 *---*---* topic
689 If you now continue development as usual, and eventually merge topic to
692 subsystem, the commits from subsystem will remain duplicated forever:
693 o---o---o---o---o---o---o---o master
695 \ \
696 o---o---o---o---o o'--o'--o'--o'--o'--M subsystem
697 \ /
698 *---*---*-..........-*--* topic
699 Such duplicates are generally frowned upon because they clutter up
702 history, making it harder to follow. To clean things up, you need to
703 transplant the commits on topic to the new subsystem tip, i.e., rebase
704 topic. This becomes a ripple effect: anyone downstream from topic is
705 forced to rebase too, and so on!
706 There are two kinds of fixes, discussed in the following subsections:
708 Easy case: The changes are literally the same.
710 This happens if the subsystem rebase was a simple rebase and had no
711 conflicts.
712 Hard case: The changes are not the same.
714 This happens if the subsystem rebase had conflicts, or used
715 --interactive to omit, edit, squash, or fixup commits; or if the
716 upstream used one of commit --amend, reset, or filter-branch.
717 The easy case
719 Only works if the changes (patch IDs based on the diff contents) on
720 subsystem are literally the same before and after the rebase subsystem
721 did.
722 In that case, the fix is easy because git rebase knows to skip changes
724 that are already present in the new upstream. So if you say (assuming
725 you’re on topic)
726 $ git rebase subsystem
728 you will end up with the fixed history
731 o---o---o---o---o---o---o---o master
733 \
734 o'--o'--o'--o'--o' subsystem
735 \
736 *---*---* topic
737 The hard case
740 Things get more complicated if the subsystem changes do not exactly
741 correspond to the ones before the rebase.
742 Note
744 While an "easy case recovery" sometimes appears to be successful
745 even in the hard case, it may have unintended consequences. For
746 example, a commit that was removed via git rebase --interactive
747 will be resurrected!
748 The idea is to manually tell git rebase "where the old subsystem ended
750 and your topic began", that is, what the old merge-base between them
751 was. You will have to find a way to name the last commit of the old
752 subsystem, for example:
753 · With the subsystem reflog: after git fetch, the old tip of
755 subsystem is at subsystem@{1}. Subsequent fetches will increase the
756 number. (See git-reflog(1).)
757 · Relative to the tip of topic: knowing that your topic has three
759 commits, the old tip of subsystem must be topic~3.
760 You can then transplant the old subsystem..topic to the new tip by
762 saying (for the reflog case, and assuming you are on topic already):
763 $ git rebase --onto subsystem subsystem@{1}
765 The ripple effect of a "hard case" recovery is especially bad: everyone
768 downstream from topic will now have to perform a "hard case" recovery
769 too!
770
772 The todo list presented by --preserve-merges --interactive does not
773 represent the topology of the revision graph. Editing commits and
774 rewording their commit messages should work fine, but attempts to
775 reorder commits tend to produce counterintuitive results.
776 For example, an attempt to rearrange
778 1 --- 2 --- 3 --- 4 --- 5
780 to
783 1 --- 2 --- 4 --- 3 --- 5
785 by moving the "pick 4" line will result in the following history:
788 3
790 /
791 1 --- 2 --- 4 --- 5
792
795 Part of the git(1) suite
796
798 1. revert-a-faulty-merge How-To
799 file:///usr/share/doc/git-1.8.3.1/howto/revert-a-faulty-merge.txt
800Git 1.8.3.1 11/19/2018 GIT-REBASE(1)