1GIT-REBASE(1) Git Manual GIT-REBASE(1)
2
6 git-rebase - Reapply commits on top of another base tip
7
9 git rebase [-i | --interactive] [<options>] [--exec <cmd>]
10 [--onto <newbase> | --keep-base] [<upstream> [<branch>]]
11 git rebase [-i | --interactive] [<options>] [--exec <cmd>] [--onto <newbase>]
12 --root [<branch>]
13 git rebase (--continue | --skip | --abort | --quit | --edit-todo | --show-current-patch)
14
16 If <branch> is specified, git rebase will perform an automatic git
17 switch <branch> before doing anything else. Otherwise it remains on the
18 current branch.
19 If <upstream> is not specified, the upstream configured in
21 branch.<name>.remote and branch.<name>.merge options will be used (see
22 git-config(1) for details) and the --fork-point option is assumed. If
23 you are currently not on any branch or if the current branch does not
24 have a configured upstream, the rebase will abort.
25 All changes made by commits in the current branch but that are not in
27 <upstream> are saved to a temporary area. This is the same set of
28 commits that would be shown by git log <upstream>..HEAD; or by git log
29 'fork_point'..HEAD, if --fork-point is active (see the description on
30 --fork-point below); or by git log HEAD, if the --root option is
31 specified.
32 The current branch is reset to <upstream>, or <newbase> if the --onto
34 option was supplied. This has the exact same effect as git reset --hard
35 <upstream> (or <newbase>). ORIG_HEAD is set to point at the tip of the
36 branch before the reset.
37 The commits that were previously saved into the temporary area are then
39 reapplied to the current branch, one by one, in order. Note that any
40 commits in HEAD which introduce the same textual changes as a commit in
41 HEAD..<upstream> are omitted (i.e., a patch already accepted upstream
42 with a different commit message or timestamp will be skipped).
43 It is possible that a merge failure will prevent this process from
45 being completely automatic. You will have to resolve any such merge
46 failure and run git rebase --continue. Another option is to bypass the
47 commit that caused the merge failure with git rebase --skip. To check
48 out the original <branch> and remove the .git/rebase-apply working
49 files, use the command git rebase --abort instead.
50 Assume the following history exists and the current branch is "topic":
52 A---B---C topic
54 /
55 D---E---F---G master
56 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
69 followed by git rebase master. When rebase exits topic will remain the
70 checked-out branch.
71 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
74 commit will be skipped and warnings will be issued (if the merge
75 backend is used). 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:
84 B'---C' topic
86 /
87 D---E---A'---F master
88 Here is how you would transplant a topic branch based on one branch to
90 another, to pretend that you forked the topic branch from the latter
91 branch, using rebase --onto.
92 First let’s assume your topic is based on branch next. For example, a
94 feature developed in topic depends on some functionality which is found
95 in next.
96 o---o---o---o---o master
98 \
99 o---o---o---o---o next
100 \
101 o---o---o topic
102 We want to make topic forked from branch master; for example, because
104 the functionality on which topic depends was merged into the more
105 stable master branch. We want our tree to look like this:
106 o---o---o---o---o master
108 | \
109 | o'--o'--o' topic
110 \
111 o---o---o---o---o next
112 We can get this using the following command:
114 git rebase --onto master next topic
116 Another example of --onto option is to rebase part of a branch. If we
118 have the following situation:
119 H---I---J topicB
121 /
122 E---F---G topicA
123 /
124 A---B---C---D master
125 then the command
127 git rebase --onto master topicA topicB
129 would result in:
131 H'--I'--J' topicB
133 /
134 | E---F---G topicA
135 |/
136 A---B---C---D master
137 This is useful when topicB does not depend on topicA.
139 A range of commits could also be removed with rebase. If we have the
141 following situation:
142 E---F---G---H---I---J topicA
144 then the command
146 git rebase --onto topicA~5 topicA~3 topicA
148 would result in the removal of commits F and G:
150 E---H'---I'---J' topicA
152 This is useful if F and G were flawed in some way, or should not be
154 part of topicA. Note that the argument to --onto and the <upstream>
155 parameter can be any valid commit-ish.
156 In case of conflict, git rebase will stop at the first problematic
158 commit and leave conflict markers in the tree. You can use git diff to
159 locate the markers (<<<<<<) and make edits to resolve the conflict. For
160 each file you edit, you need to tell Git that the conflict has been
161 resolved, typically this would be done with
162 git add <filename>
164 After resolving the conflict manually and updating the index with the
166 desired resolution, you can continue the rebasing process with
167 git rebase --continue
169 Alternatively, you can undo the git rebase with
171 git rebase --abort
173
175 --onto <newbase>
176 Starting point at which to create the new commits. If the --onto
177 option is not specified, the starting point is <upstream>. May be
178 any valid commit, and not just an existing branch name.
179 As a special case, you may use "A...B" as a shortcut for the merge
181 base of A and B if there is exactly one merge base. You can leave
182 out at most one of A and B, in which case it defaults to HEAD.
183 --keep-base
185 Set the starting point at which to create the new commits to the
186 merge base of <upstream> <branch>. Running git rebase --keep-base
187 <upstream> <branch> is equivalent to running git rebase --onto
188 <upstream>... <upstream>.
189 This option is useful in the case where one is developing a feature
191 on top of an upstream branch. While the feature is being worked on,
192 the upstream branch may advance and it may not be the best idea to
193 keep rebasing on top of the upstream but to keep the base commit
194 as-is.
195 Although both this option and --fork-point find the merge base
197 between <upstream> and <branch>, this option uses the merge base as
198 the starting point on which new commits will be created, whereas
199 --fork-point uses the merge base to determine the set of commits
200 which will be rebased.
201 See also INCOMPATIBLE OPTIONS below.
203 <upstream>
205 Upstream branch to compare against. May be any valid commit, not
206 just an existing branch name. Defaults to the configured upstream
207 for the current branch.
208 <branch>
210 Working branch; defaults to HEAD.
211 --continue
213 Restart the rebasing process after having resolved a merge
214 conflict.
215 --abort
217 Abort the rebase operation and reset HEAD to the original branch.
218 If <branch> was provided when the rebase operation was started,
219 then HEAD will be reset to <branch>. Otherwise HEAD will be reset
220 to where it was when the rebase operation was started.
221 --quit
223 Abort the rebase operation but HEAD is not reset back to the
224 original branch. The index and working tree are also left unchanged
225 as a result. If a temporary stash entry was created using
226 --autostash, it will be saved to the stash list.
227 --apply
229 Use applying strategies to rebase (calling git-am internally). This
230 option may become a no-op in the future once the merge backend
231 handles everything the apply one does.
232 See also INCOMPATIBLE OPTIONS below.
234 --empty={drop,keep,ask}
236 How to handle commits that are not empty to start and are not clean
237 cherry-picks of any upstream commit, but which become empty after
238 rebasing (because they contain a subset of already upstream
239 changes). With drop (the default), commits that become empty are
240 dropped. With keep, such commits are kept. With ask (implied by
241 --interactive), the rebase will halt when an empty commit is
242 applied allowing you to choose whether to drop it, edit files more,
243 or just commit the empty changes. Other options, like --exec, will
244 use the default of drop unless -i/--interactive is explicitly
245 specified.
246 Note that commits which start empty are kept (unless
248 --no-keep-empty is specified), and commits which are clean
249 cherry-picks (as determined by git log --cherry-mark ...) are
250 detected and dropped as a preliminary step (unless
251 --reapply-cherry-picks is passed).
252 See also INCOMPATIBLE OPTIONS below.
254 --no-keep-empty, --keep-empty
256 Do not keep commits that start empty before the rebase (i.e. that
257 do not change anything from its parent) in the result. The default
258 is to keep commits which start empty, since creating such commits
259 requires passing the --allow-empty override flag to git commit,
260 signifying that a user is very intentionally creating such a commit
261 and thus wants to keep it.
262 Usage of this flag will probably be rare, since you can get rid of
264 commits that start empty by just firing up an interactive rebase
265 and removing the lines corresponding to the commits you don’t want.
266 This flag exists as a convenient shortcut, such as for cases where
267 external tools generate many empty commits and you want them all
268 removed.
269 For commits which do not start empty but become empty after
271 rebasing, see the --empty flag.
272 See also INCOMPATIBLE OPTIONS below.
274 --reapply-cherry-picks, --no-reapply-cherry-picks
276 Reapply all clean cherry-picks of any upstream commit instead of
277 preemptively dropping them. (If these commits then become empty
278 after rebasing, because they contain a subset of already upstream
279 changes, the behavior towards them is controlled by the --empty
280 flag.)
281 By default (or if --no-reapply-cherry-picks is given), these
283 commits will be automatically dropped. Because this necessitates
284 reading all upstream commits, this can be expensive in repos with a
285 large number of upstream commits that need to be read. When using
286 the merge backend, warnings will be issued for each dropped commit
287 (unless --quiet is given). Advice will also be issued unless
288 advice.skippedCherryPicks is set to false (see git-config(1)).
289 --reapply-cherry-picks allows rebase to forgo reading all upstream
291 commits, potentially improving performance.
292 See also INCOMPATIBLE OPTIONS below.
294 --allow-empty-message
296 No-op. Rebasing commits with an empty message used to fail and this
297 option would override that behavior, allowing commits with empty
298 messages to be rebased. Now commits with an empty message do not
299 cause rebasing to halt.
300 See also INCOMPATIBLE OPTIONS below.
302 --skip
304 Restart the rebasing process by skipping the current patch.
305 --edit-todo
307 Edit the todo list during an interactive rebase.
308 --show-current-patch
310 Show the current patch in an interactive rebase or when rebase is
311 stopped because of conflicts. This is the equivalent of git show
312 REBASE_HEAD.
313 -m, --merge
315 Using merging strategies to rebase (default).
316 Note that a rebase merge works by replaying each commit from the
318 working branch on top of the <upstream> branch. Because of this,
319 when a merge conflict happens, the side reported as ours is the
320 so-far rebased series, starting with <upstream>, and theirs is the
321 working branch. In other words, the sides are swapped.
322 See also INCOMPATIBLE OPTIONS below.
324 -s <strategy>, --strategy=<strategy>
326 Use the given merge strategy, instead of the default ort. This
327 implies --merge.
328 Because git rebase replays each commit from the working branch on
330 top of the <upstream> branch using the given strategy, using the
331 ours strategy simply empties all patches from the <branch>, which
332 makes little sense.
333 See also INCOMPATIBLE OPTIONS below.
335 -X <strategy-option>, --strategy-option=<strategy-option>
337 Pass the <strategy-option> through to the merge strategy. This
338 implies --merge and, if no strategy has been specified, -s ort.
339 Note the reversal of ours and theirs as noted above for the -m
340 option.
341 See also INCOMPATIBLE OPTIONS below.
343 --rerere-autoupdate, --no-rerere-autoupdate
345 Allow the rerere mechanism to update the index with the result of
346 auto-conflict resolution if possible.
347 -S[<keyid>], --gpg-sign[=<keyid>], --no-gpg-sign
349 GPG-sign commits. The keyid argument is optional and defaults to
350 the committer identity; if specified, it must be stuck to the
351 option without a space. --no-gpg-sign is useful to countermand
352 both commit.gpgSign configuration variable, and earlier --gpg-sign.
353 -q, --quiet
355 Be quiet. Implies --no-stat.
356 -v, --verbose
358 Be verbose. Implies --stat.
359 --stat
361 Show a diffstat of what changed upstream since the last rebase. The
362 diffstat is also controlled by the configuration option
363 rebase.stat.
364 -n, --no-stat
366 Do not show a diffstat as part of the rebase process.
367 --no-verify
369 This option bypasses the pre-rebase hook. See also githooks(5).
370 --verify
372 Allows the pre-rebase hook to run, which is the default. This
373 option can be used to override --no-verify. See also githooks(5).
374 -C<n>
376 Ensure at least <n> lines of surrounding context match before and
377 after each change. When fewer lines of surrounding context exist
378 they all must match. By default no context is ever ignored. Implies
379 --apply.
380 See also INCOMPATIBLE OPTIONS below.
382 --no-ff, --force-rebase, -f
384 Individually replay all rebased commits instead of fast-forwarding
385 over the unchanged ones. This ensures that the entire history of
386 the rebased branch is composed of new commits.
387 You may find this helpful after reverting a topic branch merge, as
389 this option recreates the topic branch with fresh commits so it can
390 be remerged successfully without needing to "revert the reversion"
391 (see the revert-a-faulty-merge How-To[1] for details).
392 --fork-point, --no-fork-point
394 Use reflog to find a better common ancestor between <upstream> and
395 <branch> when calculating which commits have been introduced by
396 <branch>.
397 When --fork-point is active, fork_point will be used instead of
399 <upstream> to calculate the set of commits to rebase, where
400 fork_point is the result of git merge-base --fork-point <upstream>
401 <branch> command (see git-merge-base(1)). If fork_point ends up
402 being empty, the <upstream> will be used as a fallback.
403 If <upstream> is given on the command line, then the default is
405 --no-fork-point, otherwise the default is --fork-point. See also
406 rebase.forkpoint in git-config(1).
407 If your branch was based on <upstream> but <upstream> was rewound
409 and your branch contains commits which were dropped, this option
410 can be used with --keep-base in order to drop those commits from
411 your branch.
412 See also INCOMPATIBLE OPTIONS below.
414 --ignore-whitespace
416 Ignore whitespace differences when trying to reconcile differences.
417 Currently, each backend implements an approximation of this
418 behavior:
419 apply backend: When applying a patch, ignore changes in whitespace
421 in context lines. Unfortunately, this means that if the "old" lines
422 being replaced by the patch differ only in whitespace from the
423 existing file, you will get a merge conflict instead of a
424 successful patch application.
425 merge backend: Treat lines with only whitespace changes as
427 unchanged when merging. Unfortunately, this means that any patch
428 hunks that were intended to modify whitespace and nothing else will
429 be dropped, even if the other side had no changes that conflicted.
430 --whitespace=<option>
432 This flag is passed to the git apply program (see git-apply(1))
433 that applies the patch. Implies --apply.
434 See also INCOMPATIBLE OPTIONS below.
436 --committer-date-is-author-date
438 Instead of using the current time as the committer date, use the
439 author date of the commit being rebased as the committer date. This
440 option implies --force-rebase.
441 --ignore-date, --reset-author-date
443 Instead of using the author date of the original commit, use the
444 current time as the author date of the rebased commit. This option
445 implies --force-rebase.
446 See also INCOMPATIBLE OPTIONS below.
448 --signoff
450 Add a Signed-off-by trailer to all the rebased commits. Note that
451 if --interactive is given then only commits marked to be picked,
452 edited or reworded will have the trailer added.
453 See also INCOMPATIBLE OPTIONS below.
455 -i, --interactive
457 Make a list of the commits which are about to be rebased. Let the
458 user edit that list before rebasing. This mode can also be used to
459 split commits (see SPLITTING COMMITS below).
460 The commit list format can be changed by setting the configuration
462 option rebase.instructionFormat. A customized instruction format
463 will automatically have the long commit hash prepended to the
464 format.
465 See also INCOMPATIBLE OPTIONS below.
467 -r, --rebase-merges[=(rebase-cousins|no-rebase-cousins)]
469 By default, a rebase will simply drop merge commits from the todo
470 list, and put the rebased commits into a single, linear branch.
471 With --rebase-merges, the rebase will instead try to preserve the
472 branching structure within the commits that are to be rebased, by
473 recreating the merge commits. Any resolved merge conflicts or
474 manual amendments in these merge commits will have to be
475 resolved/re-applied manually.
476 By default, or when no-rebase-cousins was specified, commits which
478 do not have <upstream> as direct ancestor will keep their original
479 branch point, i.e. commits that would be excluded by git-log(1)'s
480 --ancestry-path option will keep their original ancestry by
481 default. If the rebase-cousins mode is turned on, such commits are
482 instead rebased onto <upstream> (or <onto>, if specified).
483 It is currently only possible to recreate the merge commits using
485 the ort merge strategy; different merge strategies can be used only
486 via explicit exec git merge -s <strategy> [...] commands.
487 See also REBASING MERGES and INCOMPATIBLE OPTIONS below.
489 -x <cmd>, --exec <cmd>
491 Append "exec <cmd>" after each line creating a commit in the final
492 history. <cmd> will be interpreted as one or more shell commands.
493 Any command that fails will interrupt the rebase, with exit code 1.
494 You may execute several commands by either using one instance of
496 --exec with several commands:
497 git rebase -i --exec "cmd1 && cmd2 && ..."
499 or by giving more than one --exec:
501 git rebase -i --exec "cmd1" --exec "cmd2" --exec ...
503 If --autosquash is used, "exec" lines will not be appended for the
505 intermediate commits, and will only appear at the end of each
506 squash/fixup series.
507 This uses the --interactive machinery internally, but it can be run
509 without an explicit --interactive.
510 See also INCOMPATIBLE OPTIONS below.
512 --root
514 Rebase all commits reachable from <branch>, instead of limiting
515 them with an <upstream>. This allows you to rebase the root
516 commit(s) on a branch. When used with --onto, it will skip changes
517 already contained in <newbase> (instead of <upstream>) whereas
518 without --onto it will operate on every change.
519 See also INCOMPATIBLE OPTIONS below.
521 --autosquash, --no-autosquash
523 When the commit log message begins with "squash! ..." or "fixup!
524 ..." or "amend! ...", and there is already a commit in the todo
525 list that matches the same ..., automatically modify the todo list
526 of rebase -i, so that the commit marked for squashing comes right
527 after the commit to be modified, and change the action of the moved
528 commit from pick to squash or fixup or fixup -C respectively. A
529 commit matches the ... if the commit subject matches, or if the
530 ... refers to the commit’s hash. As a fall-back, partial matches
531 of the commit subject work, too. The recommended way to create
532 fixup/amend/squash commits is by using the --fixup, --fixup=amend:
533 or --fixup=reword: and --squash options respectively of git-
534 commit(1).
535 If the --autosquash option is enabled by default using the
537 configuration variable rebase.autoSquash, this option can be used
538 to override and disable this setting.
539 See also INCOMPATIBLE OPTIONS below.
541 --autostash, --no-autostash
543 Automatically create a temporary stash entry before the operation
544 begins, and apply it after the operation ends. This means that you
545 can run rebase on a dirty worktree. However, use with care: the
546 final stash application after a successful rebase might result in
547 non-trivial conflicts.
548 --reschedule-failed-exec, --no-reschedule-failed-exec
550 Automatically reschedule exec commands that failed. This only makes
551 sense in interactive mode (or when an --exec option was provided).
552 Even though this option applies once a rebase is started, it’s set
554 for the whole rebase at the start based on either the
555 rebase.rescheduleFailedExec configuration (see git-config(1) or
556 "CONFIGURATION" below) or whether this option is provided.
557 Otherwise an explicit --no-reschedule-failed-exec at the start
558 would be overridden by the presence of
559 rebase.rescheduleFailedExec=true configuration.
560
562 The following options:
563 • --apply
565 • --whitespace
567 • -C
569 are incompatible with the following options:
571 • --merge
573 • --strategy
575 • --strategy-option
577 • --allow-empty-message
579 • --[no-]autosquash
581 • --rebase-merges
583 • --interactive
585 • --exec
587 • --no-keep-empty
589 • --empty=
591 • --reapply-cherry-picks
593 • --edit-todo
595 • --root when used in combination with --onto
597 In addition, the following pairs of options are incompatible:
599 • --keep-base and --onto
601 • --keep-base and --root
603 • --fork-point and --root
605
607 git rebase has two primary backends: apply and merge. (The apply
608 backend used to be known as the am backend, but the name led to
609 confusion as it looks like a verb instead of a noun. Also, the merge
610 backend used to be known as the interactive backend, but it is now used
611 for non-interactive cases as well. Both were renamed based on
612 lower-level functionality that underpinned each.) There are some subtle
613 differences in how these two backends behave:
614 Empty commits
616 The apply backend unfortunately drops intentionally empty commits, i.e.
617 commits that started empty, though these are rare in practice. It also
618 drops commits that become empty and has no option for controlling this
619 behavior.
620 The merge backend keeps intentionally empty commits by default (though
622 with -i they are marked as empty in the todo list editor, or they can
623 be dropped automatically with --no-keep-empty).
624 Similar to the apply backend, by default the merge backend drops
626 commits that become empty unless -i/--interactive is specified (in
627 which case it stops and asks the user what to do). The merge backend
628 also has an --empty={drop,keep,ask} option for changing the behavior of
629 handling commits that become empty.
630 Directory rename detection
632 Due to the lack of accurate tree information (arising from constructing
633 fake ancestors with the limited information available in patches),
634 directory rename detection is disabled in the apply backend. Disabled
635 directory rename detection means that if one side of history renames a
636 directory and the other adds new files to the old directory, then the
637 new files will be left behind in the old directory without any warning
638 at the time of rebasing that you may want to move these files into the
639 new directory.
640 Directory rename detection works with the merge backend to provide you
642 warnings in such cases.
643 Context
645 The apply backend works by creating a sequence of patches (by calling
646 format-patch internally), and then applying the patches in sequence
647 (calling am internally). Patches are composed of multiple hunks, each
648 with line numbers, a context region, and the actual changes. The line
649 numbers have to be taken with some fuzz, since the other side will
650 likely have inserted or deleted lines earlier in the file. The context
651 region is meant to help find how to adjust the line numbers in order to
652 apply the changes to the right lines. However, if multiple areas of the
653 code have the same surrounding lines of context, the wrong one can be
654 picked. There are real-world cases where this has caused commits to be
655 reapplied incorrectly with no conflicts reported. Setting diff.context
656 to a larger value may prevent such types of problems, but increases the
657 chance of spurious conflicts (since it will require more lines of
658 matching context to apply).
659 The merge backend works with a full copy of each relevant file,
661 insulating it from these types of problems.
662 Labelling of conflicts markers
664 When there are content conflicts, the merge machinery tries to annotate
665 each side’s conflict markers with the commits where the content came
666 from. Since the apply backend drops the original information about the
667 rebased commits and their parents (and instead generates new fake
668 commits based off limited information in the generated patches), those
669 commits cannot be identified; instead it has to fall back to a commit
670 summary. Also, when merge.conflictStyle is set to diff3 or zdiff3, the
671 apply backend will use "constructed merge base" to label the content
672 from the merge base, and thus provide no information about the merge
673 base commit whatsoever.
674 The merge backend works with the full commits on both sides of history
676 and thus has no such limitations.
677 Hooks
679 The apply backend has not traditionally called the post-commit hook,
680 while the merge backend has. Both have called the post-checkout hook,
681 though the merge backend has squelched its output. Further, both
682 backends only call the post-checkout hook with the starting point
683 commit of the rebase, not the intermediate commits nor the final
684 commit. In each case, the calling of these hooks was by accident of
685 implementation rather than by design (both backends were originally
686 implemented as shell scripts and happened to invoke other commands like
687 git checkout or git commit that would call the hooks). Both backends
688 should have the same behavior, though it is not entirely clear which,
689 if any, is correct. We will likely make rebase stop calling either of
690 these hooks in the future.
691 Interruptability
693 The apply backend has safety problems with an ill-timed interrupt; if
694 the user presses Ctrl-C at the wrong time to try to abort the rebase,
695 the rebase can enter a state where it cannot be aborted with a
696 subsequent git rebase --abort. The merge backend does not appear to
697 suffer from the same shortcoming. (See
698 https://lore.kernel.org/git/20200207132152.GC2868@szeder.dev/ for
699 details.)
700 Commit Rewording
702 When a conflict occurs while rebasing, rebase stops and asks the user
703 to resolve. Since the user may need to make notable changes while
704 resolving conflicts, after conflicts are resolved and the user has run
705 git rebase --continue, the rebase should open an editor and ask the
706 user to update the commit message. The merge backend does this, while
707 the apply backend blindly applies the original commit message.
708 Miscellaneous differences
710 There are a few more behavioral differences that most folks would
711 probably consider inconsequential but which are mentioned for
712 completeness:
713 • Reflog: The two backends will use different wording when describing
715 the changes made in the reflog, though both will make use of the
716 word "rebase".
717 • Progress, informational, and error messages: The two backends
719 provide slightly different progress and informational messages.
720 Also, the apply backend writes error messages (such as "Your files
721 would be overwritten...") to stdout, while the merge backend writes
722 them to stderr.
723 • State directories: The two backends keep their state in different
725 directories under .git/
726
728 The merge mechanism (git merge and git pull commands) allows the
729 backend merge strategies to be chosen with -s option. Some strategies
730 can also take their own options, which can be passed by giving
731 -X<option> arguments to git merge and/or git pull.
732 ort
734 This is the default merge strategy when pulling or merging one
735 branch. This strategy can only resolve two heads using a 3-way
736 merge algorithm. When there is more than one common ancestor that
737 can be used for 3-way merge, it creates a merged tree of the common
738 ancestors and uses that as the reference tree for the 3-way merge.
739 This has been reported to result in fewer merge conflicts without
740 causing mismerges by tests done on actual merge commits taken from
741 Linux 2.6 kernel development history. Additionally this strategy
742 can detect and handle merges involving renames. It does not make
743 use of detected copies. The name for this algorithm is an acronym
744 ("Ostensibly Recursive’s Twin") and came from the fact that it was
745 written as a replacement for the previous default algorithm,
746 recursive.
747 The ort strategy can take the following options:
749 ours
751 This option forces conflicting hunks to be auto-resolved
752 cleanly by favoring our version. Changes from the other tree
753 that do not conflict with our side are reflected in the merge
754 result. For a binary file, the entire contents are taken from
755 our side.
756 This should not be confused with the ours merge strategy, which
758 does not even look at what the other tree contains at all. It
759 discards everything the other tree did, declaring our history
760 contains all that happened in it.
761 theirs
763 This is the opposite of ours; note that, unlike ours, there is
764 no theirs merge strategy to confuse this merge option with.
765 ignore-space-change, ignore-all-space, ignore-space-at-eol,
767 ignore-cr-at-eol
768 Treats lines with the indicated type of whitespace change as
769 unchanged for the sake of a three-way merge. Whitespace changes
770 mixed with other changes to a line are not ignored. See also
771 git-diff(1) -b, -w, --ignore-space-at-eol, and
772 --ignore-cr-at-eol.
773 • If their version only introduces whitespace changes to a
775 line, our version is used;
776 • If our version introduces whitespace changes but their
778 version includes a substantial change, their version is
779 used;
780 • Otherwise, the merge proceeds in the usual way.
782 renormalize
784 This runs a virtual check-out and check-in of all three stages
785 of a file when resolving a three-way merge. This option is
786 meant to be used when merging branches with different clean
787 filters or end-of-line normalization rules. See "Merging
788 branches with differing checkin/checkout attributes" in
789 gitattributes(5) for details.
790 no-renormalize
792 Disables the renormalize option. This overrides the
793 merge.renormalize configuration variable.
794 find-renames[=<n>]
796 Turn on rename detection, optionally setting the similarity
797 threshold. This is the default. This overrides the
798 merge.renames configuration variable. See also git-diff(1)
799 --find-renames.
800 rename-threshold=<n>
802 Deprecated synonym for find-renames=<n>.
803 subtree[=<path>]
805 This option is a more advanced form of subtree strategy, where
806 the strategy makes a guess on how two trees must be shifted to
807 match with each other when merging. Instead, the specified path
808 is prefixed (or stripped from the beginning) to make the shape
809 of two trees to match.
810 recursive
812 This can only resolve two heads using a 3-way merge algorithm. When
813 there is more than one common ancestor that can be used for 3-way
814 merge, it creates a merged tree of the common ancestors and uses
815 that as the reference tree for the 3-way merge. This has been
816 reported to result in fewer merge conflicts without causing
817 mismerges by tests done on actual merge commits taken from Linux
818 2.6 kernel development history. Additionally this can detect and
819 handle merges involving renames. It does not make use of detected
820 copies. This was the default strategy for resolving two heads from
821 Git v0.99.9k until v2.33.0.
822 The recursive strategy takes the same options as ort. However,
824 there are three additional options that ort ignores (not documented
825 above) that are potentially useful with the recursive strategy:
826 patience
828 Deprecated synonym for diff-algorithm=patience.
829 diff-algorithm=[patience|minimal|histogram|myers]
831 Use a different diff algorithm while merging, which can help
832 avoid mismerges that occur due to unimportant matching lines
833 (such as braces from distinct functions). See also git-diff(1)
834 --diff-algorithm. Note that ort specifically uses
835 diff-algorithm=histogram, while recursive defaults to the
836 diff.algorithm config setting.
837 no-renames
839 Turn off rename detection. This overrides the merge.renames
840 configuration variable. See also git-diff(1) --no-renames.
841 resolve
843 This can only resolve two heads (i.e. the current branch and
844 another branch you pulled from) using a 3-way merge algorithm. It
845 tries to carefully detect criss-cross merge ambiguities. It does
846 not handle renames.
847 octopus
849 This resolves cases with more than two heads, but refuses to do a
850 complex merge that needs manual resolution. It is primarily meant
851 to be used for bundling topic branch heads together. This is the
852 default merge strategy when pulling or merging more than one
853 branch.
854 ours
856 This resolves any number of heads, but the resulting tree of the
857 merge is always that of the current branch head, effectively
858 ignoring all changes from all other branches. It is meant to be
859 used to supersede old development history of side branches. Note
860 that this is different from the -Xours option to the recursive
861 merge strategy.
862 subtree
864 This is a modified ort strategy. When merging trees A and B, if B
865 corresponds to a subtree of A, B is first adjusted to match the
866 tree structure of A, instead of reading the trees at the same
867 level. This adjustment is also done to the common ancestor tree.
868 With the strategies that use 3-way merge (including the default, ort),
870 if a change is made on both branches, but later reverted on one of the
871 branches, that change will be present in the merged result; some people
872 find this behavior confusing. It occurs because only the heads and the
873 merge base are considered when performing a merge, not the individual
874 commits. The merge algorithm therefore considers the reverted change as
875 no change at all, and substitutes the changed version instead.
876
878 You should understand the implications of using git rebase on a
879 repository that you share. See also RECOVERING FROM UPSTREAM REBASE
880 below.
881 When the git-rebase command is run, it will first execute a
883 "pre-rebase" hook if one exists. You can use this hook to do sanity
884 checks and reject the rebase if it isn’t appropriate. Please see the
885 template pre-rebase hook script for an example.
886 Upon completion, <branch> will be the current branch.
888
890 Rebasing interactively means that you have a chance to edit the commits
891 which are rebased. You can reorder the commits, and you can remove them
892 (weeding out bad or otherwise unwanted patches).
893 The interactive mode is meant for this type of workflow:
895 1. have a wonderful idea
897 2. hack on the code
899 3. prepare a series for submission
901 4. submit
903 where point 2. consists of several instances of
905 a) regular use
907 1. finish something worthy of a commit
909 2. commit
911 b) independent fixup
913 1. realize that something does not work
915 2. fix that
917 3. commit it
919 Sometimes the thing fixed in b.2. cannot be amended to the not-quite
921 perfect commit it fixes, because that commit is buried deeply in a
922 patch series. That is exactly what interactive rebase is for: use it
923 after plenty of "a"s and "b"s, by rearranging and editing commits, and
924 squashing multiple commits into one.
925 Start it with the last commit you want to retain as-is:
927 git rebase -i <after-this-commit>
929 An editor will be fired up with all the commits in your current branch
931 (ignoring merge commits), which come after the given commit. You can
932 reorder the commits in this list to your heart’s content, and you can
933 remove them. The list looks more or less like this:
934 pick deadbee The oneline of this commit
936 pick fa1afe1 The oneline of the next commit
937 ...
938 The oneline descriptions are purely for your pleasure; git rebase will
940 not look at them but at the commit names ("deadbee" and "fa1afe1" in
941 this example), so do not delete or edit the names.
942 By replacing the command "pick" with the command "edit", you can tell
944 git rebase to stop after applying that commit, so that you can edit the
945 files and/or the commit message, amend the commit, and continue
946 rebasing.
947 To interrupt the rebase (just like an "edit" command would do, but
949 without cherry-picking any commit first), use the "break" command.
950 If you just want to edit the commit message for a commit, replace the
952 command "pick" with the command "reword".
953 To drop a commit, replace the command "pick" with "drop", or just
955 delete the matching line.
956 If you want to fold two or more commits into one, replace the command
958 "pick" for the second and subsequent commits with "squash" or "fixup".
959 If the commits had different authors, the folded commit will be
960 attributed to the author of the first commit. The suggested commit
961 message for the folded commit is the concatenation of the first
962 commit’s message with those identified by "squash" commands, omitting
963 the messages of commits identified by "fixup" commands, unless "fixup
964 -c" is used. In that case the suggested commit message is only the
965 message of the "fixup -c" commit, and an editor is opened allowing you
966 to edit the message. The contents (patch) of the "fixup -c" commit are
967 still incorporated into the folded commit. If there is more than one
968 "fixup -c" commit, the message from the final one is used. You can also
969 use "fixup -C" to get the same behavior as "fixup -c" except without
970 opening an editor.
971 git rebase will stop when "pick" has been replaced with "edit" or when
973 a command fails due to merge errors. When you are done editing and/or
974 resolving conflicts you can continue with git rebase --continue.
975 For example, if you want to reorder the last 5 commits, such that what
977 was HEAD~4 becomes the new HEAD. To achieve that, you would call git
978 rebase like this:
979 $ git rebase -i HEAD~5
981 And move the first patch to the end of the list.
983 You might want to recreate merge commits, e.g. if you have a history
985 like this:
986 X
988 \
989 A---M---B
990 /
991 ---o---O---P---Q
992 Suppose you want to rebase the side branch starting at "A" to "Q". Make
994 sure that the current HEAD is "B", and call
995 $ git rebase -i -r --onto Q O
997 Reordering and editing commits usually creates untested intermediate
999 steps. You may want to check that your history editing did not break
1000 anything by running a test, or at least recompiling at intermediate
1001 points in history by using the "exec" command (shortcut "x"). You may
1002 do so by creating a todo list like this one:
1003 pick deadbee Implement feature XXX
1005 fixup f1a5c00 Fix to feature XXX
1006 exec make
1007 pick c0ffeee The oneline of the next commit
1008 edit deadbab The oneline of the commit after
1009 exec cd subdir; make test
1010 ...
1011 The interactive rebase will stop when a command fails (i.e. exits with
1013 non-0 status) to give you an opportunity to fix the problem. You can
1014 continue with git rebase --continue.
1015 The "exec" command launches the command in a shell (the one specified
1017 in $SHELL, or the default shell if $SHELL is not set), so you can use
1018 shell features (like "cd", ">", ";" ...). The command is run from the
1019 root of the working tree.
1020 $ git rebase -i --exec "make test"
1022 This command lets you check that intermediate commits are compilable.
1024 The todo list becomes like that:
1025 pick 5928aea one
1027 exec make test
1028 pick 04d0fda two
1029 exec make test
1030 pick ba46169 three
1031 exec make test
1032 pick f4593f9 four
1033 exec make test
1034
1036 In interactive mode, you can mark commits with the action "edit".
1037 However, this does not necessarily mean that git rebase expects the
1038 result of this edit to be exactly one commit. Indeed, you can undo the
1039 commit, or you can add other commits. This can be used to split a
1040 commit into two:
1041 • Start an interactive rebase with git rebase -i <commit>^, where
1043 <commit> is the commit you want to split. In fact, any commit range
1044 will do, as long as it contains that commit.
1045 • Mark the commit you want to split with the action "edit".
1047 • When it comes to editing that commit, execute git reset HEAD^. The
1049 effect is that the HEAD is rewound by one, and the index follows
1050 suit. However, the working tree stays the same.
1051 • Now add the changes to the index that you want to have in the first
1053 commit. You can use git add (possibly interactively) or git gui (or
1054 both) to do that.
1055 • Commit the now-current index with whatever commit message is
1057 appropriate now.
1058 • Repeat the last two steps until your working tree is clean.
1060 • Continue the rebase with git rebase --continue.
1062 If you are not absolutely sure that the intermediate revisions are
1064 consistent (they compile, pass the testsuite, etc.) you should use git
1065 stash to stash away the not-yet-committed changes after each commit,
1066 test, and amend the commit if fixes are necessary.
1067
1069 Rebasing (or any other form of rewriting) a branch that others have
1070 based work on is a bad idea: anyone downstream of it is forced to
1071 manually fix their history. This section explains how to do the fix
1072 from the downstream’s point of view. The real fix, however, would be to
1073 avoid rebasing the upstream in the first place.
1074 To illustrate, suppose you are in a situation where someone develops a
1076 subsystem branch, and you are working on a topic that is dependent on
1077 this subsystem. You might end up with a history like the following:
1078 o---o---o---o---o---o---o---o master
1080 \
1081 o---o---o---o---o subsystem
1082 \
1083 *---*---* topic
1084 If subsystem is rebased against master, the following happens:
1086 o---o---o---o---o---o---o---o master
1088 \ \
1089 o---o---o---o---o o'--o'--o'--o'--o' subsystem
1090 \
1091 *---*---* topic
1092 If you now continue development as usual, and eventually merge topic to
1094 subsystem, the commits from subsystem will remain duplicated forever:
1095 o---o---o---o---o---o---o---o master
1097 \ \
1098 o---o---o---o---o o'--o'--o'--o'--o'--M subsystem
1099 \ /
1100 *---*---*-..........-*--* topic
1101 Such duplicates are generally frowned upon because they clutter up
1103 history, making it harder to follow. To clean things up, you need to
1104 transplant the commits on topic to the new subsystem tip, i.e., rebase
1105 topic. This becomes a ripple effect: anyone downstream from topic is
1106 forced to rebase too, and so on!
1107 There are two kinds of fixes, discussed in the following subsections:
1109 Easy case: The changes are literally the same.
1111 This happens if the subsystem rebase was a simple rebase and had no
1112 conflicts.
1113 Hard case: The changes are not the same.
1115 This happens if the subsystem rebase had conflicts, or used
1116 --interactive to omit, edit, squash, or fixup commits; or if the
1117 upstream used one of commit --amend, reset, or a full history
1118 rewriting command like filter-repo[2].
1119 The easy case
1121 Only works if the changes (patch IDs based on the diff contents) on
1122 subsystem are literally the same before and after the rebase subsystem
1123 did.
1124 In that case, the fix is easy because git rebase knows to skip changes
1126 that are already present in the new upstream (unless
1127 --reapply-cherry-picks is given). So if you say (assuming you’re on
1128 topic)
1129 $ git rebase subsystem
1131 you will end up with the fixed history
1133 o---o---o---o---o---o---o---o master
1135 \
1136 o'--o'--o'--o'--o' subsystem
1137 \
1138 *---*---* topic
1139 The hard case
1141 Things get more complicated if the subsystem changes do not exactly
1142 correspond to the ones before the rebase.
1143 Note
1145 While an "easy case recovery" sometimes appears to be successful
1146 even in the hard case, it may have unintended consequences. For
1147 example, a commit that was removed via git rebase --interactive
1148 will be resurrected!
1149 The idea is to manually tell git rebase "where the old subsystem ended
1151 and your topic began", that is, what the old merge base between them
1152 was. You will have to find a way to name the last commit of the old
1153 subsystem, for example:
1154 • With the subsystem reflog: after git fetch, the old tip of
1156 subsystem is at subsystem@{1}. Subsequent fetches will increase the
1157 number. (See git-reflog(1).)
1158 • Relative to the tip of topic: knowing that your topic has three
1160 commits, the old tip of subsystem must be topic~3.
1161 You can then transplant the old subsystem..topic to the new tip by
1163 saying (for the reflog case, and assuming you are on topic already):
1164 $ git rebase --onto subsystem subsystem@{1}
1166 The ripple effect of a "hard case" recovery is especially bad: everyone
1168 downstream from topic will now have to perform a "hard case" recovery
1169 too!
1170
1172 The interactive rebase command was originally designed to handle
1173 individual patch series. As such, it makes sense to exclude merge
1174 commits from the todo list, as the developer may have merged the
1175 then-current master while working on the branch, only to rebase all the
1176 commits onto master eventually (skipping the merge commits).
1177 However, there are legitimate reasons why a developer may want to
1179 recreate merge commits: to keep the branch structure (or "commit
1180 topology") when working on multiple, inter-related branches.
1181 In the following example, the developer works on a topic branch that
1183 refactors the way buttons are defined, and on another topic branch that
1184 uses that refactoring to implement a "Report a bug" button. The output
1185 of git log --graph --format=%s -5 may look like this:
1186 * Merge branch 'report-a-bug'
1188 |\
1189 | * Add the feedback button
1190 * | Merge branch 'refactor-button'
1191 |\ \
1192 | |/
1193 | * Use the Button class for all buttons
1194 | * Extract a generic Button class from the DownloadButton one
1195 The developer might want to rebase those commits to a newer master
1197 while keeping the branch topology, for example when the first topic
1198 branch is expected to be integrated into master much earlier than the
1199 second one, say, to resolve merge conflicts with changes to the
1200 DownloadButton class that made it into master.
1201 This rebase can be performed using the --rebase-merges option. It will
1203 generate a todo list looking like this:
1204 label onto
1206 # Branch: refactor-button
1208 reset onto
1209 pick 123456 Extract a generic Button class from the DownloadButton one
1210 pick 654321 Use the Button class for all buttons
1211 label refactor-button
1212 # Branch: report-a-bug
1214 reset refactor-button # Use the Button class for all buttons
1215 pick abcdef Add the feedback button
1216 label report-a-bug
1217 reset onto
1219 merge -C a1b2c3 refactor-button # Merge 'refactor-button'
1220 merge -C 6f5e4d report-a-bug # Merge 'report-a-bug'
1221 In contrast to a regular interactive rebase, there are label, reset and
1223 merge commands in addition to pick ones.
1224 The label command associates a label with the current HEAD when that
1226 command is executed. These labels are created as worktree-local refs
1227 (refs/rewritten/<label>) that will be deleted when the rebase finishes.
1228 That way, rebase operations in multiple worktrees linked to the same
1229 repository do not interfere with one another. If the label command
1230 fails, it is rescheduled immediately, with a helpful message how to
1231 proceed.
1232 The reset command resets the HEAD, index and worktree to the specified
1234 revision. It is similar to an exec git reset --hard <label>, but
1235 refuses to overwrite untracked files. If the reset command fails, it is
1236 rescheduled immediately, with a helpful message how to edit the todo
1237 list (this typically happens when a reset command was inserted into the
1238 todo list manually and contains a typo).
1239 The merge command will merge the specified revision(s) into whatever is
1241 HEAD at that time. With -C <original-commit>, the commit message of the
1242 specified merge commit will be used. When the -C is changed to a
1243 lower-case -c, the message will be opened in an editor after a
1244 successful merge so that the user can edit the message.
1245 If a merge command fails for any reason other than merge conflicts
1247 (i.e. when the merge operation did not even start), it is rescheduled
1248 immediately.
1249 By default, the merge command will use the ort merge strategy for
1251 regular merges, and octopus for octopus merges. One can specify a
1252 default strategy for all merges using the --strategy argument when
1253 invoking rebase, or can override specific merges in the interactive
1254 list of commands by using an exec command to call git merge explicitly
1255 with a --strategy argument. Note that when calling git merge explicitly
1256 like this, you can make use of the fact that the labels are
1257 worktree-local refs (the ref refs/rewritten/onto would correspond to
1258 the label onto, for example) in order to refer to the branches you want
1259 to merge.
1260 Note: the first command (label onto) labels the revision onto which the
1262 commits are rebased; The name onto is just a convention, as a nod to
1263 the --onto option.
1264 It is also possible to introduce completely new merge commits from
1266 scratch by adding a command of the form merge <merge-head>. This form
1267 will generate a tentative commit message and always open an editor to
1268 let the user edit it. This can be useful e.g. when a topic branch turns
1269 out to address more than a single concern and wants to be split into
1270 two or even more topic branches. Consider this todo list:
1271 pick 192837 Switch from GNU Makefiles to CMake
1273 pick 5a6c7e Document the switch to CMake
1274 pick 918273 Fix detection of OpenSSL in CMake
1275 pick afbecd http: add support for TLS v1.3
1276 pick fdbaec Fix detection of cURL in CMake on Windows
1277 The one commit in this list that is not related to CMake may very well
1279 have been motivated by working on fixing all those bugs introduced by
1280 switching to CMake, but it addresses a different concern. To split this
1281 branch into two topic branches, the todo list could be edited like
1282 this:
1283 label onto
1285 pick afbecd http: add support for TLS v1.3
1287 label tlsv1.3
1288 reset onto
1290 pick 192837 Switch from GNU Makefiles to CMake
1291 pick 918273 Fix detection of OpenSSL in CMake
1292 pick fdbaec Fix detection of cURL in CMake on Windows
1293 pick 5a6c7e Document the switch to CMake
1294 label cmake
1295 reset onto
1297 merge tlsv1.3
1298 merge cmake
1299
1301 rebase.backend
1302 Default backend to use for rebasing. Possible choices are apply or
1303 merge. In the future, if the merge backend gains all remaining
1304 capabilities of the apply backend, this setting may become unused.
1305 rebase.stat
1307 Whether to show a diffstat of what changed upstream since the last
1308 rebase. False by default.
1309 rebase.autoSquash
1311 If set to true enable --autosquash option by default.
1312 rebase.autoStash
1314 When set to true, automatically create a temporary stash entry
1315 before the operation begins, and apply it after the operation ends.
1316 This means that you can run rebase on a dirty worktree. However,
1317 use with care: the final stash application after a successful
1318 rebase might result in non-trivial conflicts. This option can be
1319 overridden by the --no-autostash and --autostash options of git-
1320 rebase(1). Defaults to false.
1321 rebase.missingCommitsCheck
1323 If set to "warn", git rebase -i will print a warning if some
1324 commits are removed (e.g. a line was deleted), however the rebase
1325 will still proceed. If set to "error", it will print the previous
1326 warning and stop the rebase, git rebase --edit-todo can then be
1327 used to correct the error. If set to "ignore", no checking is done.
1328 To drop a commit without warning or error, use the drop command in
1329 the todo list. Defaults to "ignore".
1330 rebase.instructionFormat
1332 A format string, as specified in git-log(1), to be used for the
1333 todo list during an interactive rebase. The format will
1334 automatically have the long commit hash prepended to the format.
1335 rebase.abbreviateCommands
1337 If set to true, git rebase will use abbreviated command names in
1338 the todo list resulting in something like this:
1339 p deadbee The oneline of the commit
1341 p fa1afe1 The oneline of the next commit
1342 ...
1343 instead of:
1345 pick deadbee The oneline of the commit
1347 pick fa1afe1 The oneline of the next commit
1348 ...
1349 Defaults to false.
1351 rebase.rescheduleFailedExec
1353 Automatically reschedule exec commands that failed. This only makes
1354 sense in interactive mode (or when an --exec option was provided).
1355 This is the same as specifying the --reschedule-failed-exec option.
1356 rebase.forkPoint
1358 If set to false set --no-fork-point option by default.
1359 sequence.editor
1361 Text editor used by git rebase -i for editing the rebase
1362 instruction file. The value is meant to be interpreted by the shell
1363 when it is used. It can be overridden by the GIT_SEQUENCE_EDITOR
1364 environment variable. When not configured the default commit
1365 message editor is used instead.
1366
1368 Part of the git(1) suite
1369
1371 1. revert-a-faulty-merge How-To
1372 file:///usr/share/doc/git/howto/revert-a-faulty-merge.html
1373 2. filter-repo
1375 https://github.com/newren/git-filter-repo
1376Git 2.36.1 2022-05-05 GIT-REBASE(1)