6 git-rebase - Reapply commits on top of another base tip
11 'git rebase' [-i | --interactive] [<options>] [--exec <cmd>]
12 [--onto <newbase> | --keep-base] [<upstream> [<branch>]]
13 'git rebase' [-i | --interactive] [<options>] [--exec <cmd>] [--onto <newbase>]
15 'git rebase' (--continue | --skip | --abort | --quit | --edit-todo | --show-current-patch)
19 If <branch> is specified, 'git rebase' will perform an automatic
20 `git switch <branch>` before doing anything else. Otherwise
21 it remains on the current branch.
23 If <upstream> is not specified, the upstream configured in
24 branch.<name>.remote and branch.<name>.merge options will be used (see
25 linkgit:git-config[1] for details) and the `--fork-point` option is
26 assumed. If you are currently not on any branch or if the current
27 branch does not have a configured upstream, the rebase will abort.
29 All changes made by commits in the current branch but that are not
30 in <upstream> are saved to a temporary area. This is the same set
31 of commits that would be shown by `git log <upstream>..HEAD`; or by
32 `git log 'fork_point'..HEAD`, if `--fork-point` is active (see the
33 description on `--fork-point` below); or by `git log HEAD`, if the
34 `--root` option is specified.
36 The current branch is reset to <upstream>, or <newbase> if the
37 --onto option was supplied. This has the exact same effect as
38 `git reset --hard <upstream>` (or <newbase>). ORIG_HEAD is set
39 to point at the tip of the branch before the reset.
41 The commits that were previously saved into the temporary area are
42 then reapplied to the current branch, one by one, in order. Note that
43 any commits in HEAD which introduce the same textual changes as a commit
44 in HEAD..<upstream> are omitted (i.e., a patch already accepted upstream
45 with a different commit message or timestamp will be skipped).
47 It is possible that a merge failure will prevent this process from being
48 completely automatic. You will have to resolve any such merge failure
49 and run `git rebase --continue`. Another option is to bypass the commit
50 that caused the merge failure with `git rebase --skip`. To check out the
51 original <branch> and remove the .git/rebase-apply working files, use the
52 command `git rebase --abort` instead.
54 Assume the following history exists and the current branch is "topic":
62 From this point, the result of either of the following commands:
66 git rebase master topic
76 *NOTE:* The latter form is just a short-hand of `git checkout topic`
77 followed by `git rebase master`. When rebase exits `topic` will
78 remain the checked-out branch.
80 If the upstream branch already contains a change you have made (e.g.,
81 because you mailed a patch which was applied upstream), then that commit
82 will be skipped. For example, running `git rebase master` on the
83 following history (in which `A'` and `A` introduce the same set of changes,
84 but have different committer information):
100 Here is how you would transplant a topic branch based on one
101 branch to another, to pretend that you forked the topic branch
102 from the latter branch, using `rebase --onto`.
104 First let's assume your 'topic' is based on branch 'next'.
105 For example, a feature developed in 'topic' depends on some
106 functionality which is found in 'next'.
109 o---o---o---o---o master
111 o---o---o---o---o next
116 We want to make 'topic' forked from branch 'master'; for example,
117 because the functionality on which 'topic' depends was merged into the
118 more stable 'master' branch. We want our tree to look like this:
121 o---o---o---o---o master
125 o---o---o---o---o next
128 We can get this using the following command:
130 git rebase --onto master next topic
133 Another example of --onto option is to rebase part of a
134 branch. If we have the following situation:
146 git rebase --onto master topicA topicB
158 This is useful when topicB does not depend on topicA.
160 A range of commits could also be removed with rebase. If we have
161 the following situation:
164 E---F---G---H---I---J topicA
169 git rebase --onto topicA~5 topicA~3 topicA
171 would result in the removal of commits F and G:
174 E---H'---I'---J' topicA
177 This is useful if F and G were flawed in some way, or should not be
178 part of topicA. Note that the argument to --onto and the <upstream>
179 parameter can be any valid commit-ish.
181 In case of conflict, 'git rebase' will stop at the first problematic commit
182 and leave conflict markers in the tree. You can use 'git diff' to locate
183 the markers (<<<<<<) and make edits to resolve the conflict. For each
184 file you edit, you need to tell Git that the conflict has been resolved,
185 typically this would be done with
191 After resolving the conflict manually and updating the index with the
192 desired resolution, you can continue the rebasing process with
195 git rebase --continue
198 Alternatively, you can undo the 'git rebase' with
206 include::config/rebase.txt[]
211 Starting point at which to create the new commits. If the
212 --onto option is not specified, the starting point is
213 <upstream>. May be any valid commit, and not just an
214 existing branch name.
216 As a special case, you may use "A\...B" as a shortcut for the
217 merge base of A and B if there is exactly one merge base. You can
218 leave out at most one of A and B, in which case it defaults to HEAD.
221 Set the starting point at which to create the new commits to the
222 merge base of <upstream> <branch>. Running
223 'git rebase --keep-base <upstream> <branch>' is equivalent to
224 running 'git rebase --onto <upstream>... <upstream>'.
226 This option is useful in the case where one is developing a feature on
227 top of an upstream branch. While the feature is being worked on, the
228 upstream branch may advance and it may not be the best idea to keep
229 rebasing on top of the upstream but to keep the base commit as-is.
231 Although both this option and --fork-point find the merge base between
232 <upstream> and <branch>, this option uses the merge base as the _starting
233 point_ on which new commits will be created, whereas --fork-point uses
234 the merge base to determine the _set of commits_ which will be rebased.
236 See also INCOMPATIBLE OPTIONS below.
239 Upstream branch to compare against. May be any valid commit,
240 not just an existing branch name. Defaults to the configured
241 upstream for the current branch.
244 Working branch; defaults to HEAD.
247 Restart the rebasing process after having resolved a merge conflict.
250 Abort the rebase operation and reset HEAD to the original
251 branch. If <branch> was provided when the rebase operation was
252 started, then HEAD will be reset to <branch>. Otherwise HEAD
253 will be reset to where it was when the rebase operation was
257 Abort the rebase operation but HEAD is not reset back to the
258 original branch. The index and working tree are also left
259 unchanged as a result.
262 Use applying strategies to rebase (calling `git-am`
263 internally). This option may become a no-op in the future
264 once the merge backend handles everything the apply one does.
266 See also INCOMPATIBLE OPTIONS below.
268 --empty={drop,keep,ask}::
269 How to handle commits that are not empty to start and are not
270 clean cherry-picks of any upstream commit, but which become
271 empty after rebasing (because they contain a subset of already
272 upstream changes). With drop (the default), commits that
273 become empty are dropped. With keep, such commits are kept.
274 With ask (implied by --interactive), the rebase will halt when
275 an empty commit is applied allowing you to choose whether to
276 drop it, edit files more, or just commit the empty changes.
277 Other options, like --exec, will use the default of drop unless
278 -i/--interactive is explicitly specified.
280 Note that commits which start empty are kept (unless --no-keep-empty
281 is specified), and commits which are clean cherry-picks (as determined
282 by `git log --cherry-mark ...`) are detected and dropped as a
283 preliminary step (unless --reapply-cherry-picks is passed).
285 See also INCOMPATIBLE OPTIONS below.
289 Do not keep commits that start empty before the rebase
290 (i.e. that do not change anything from its parent) in the
291 result. The default is to keep commits which start empty,
292 since creating such commits requires passing the --allow-empty
293 override flag to `git commit`, signifying that a user is very
294 intentionally creating such a commit and thus wants to keep
297 Usage of this flag will probably be rare, since you can get rid of
298 commits that start empty by just firing up an interactive rebase and
299 removing the lines corresponding to the commits you don't want. This
300 flag exists as a convenient shortcut, such as for cases where external
301 tools generate many empty commits and you want them all removed.
303 For commits which do not start empty but become empty after rebasing,
304 see the --empty flag.
306 See also INCOMPATIBLE OPTIONS below.
308 --reapply-cherry-picks::
309 --no-reapply-cherry-picks::
310 Reapply all clean cherry-picks of any upstream commit instead
311 of preemptively dropping them. (If these commits then become
312 empty after rebasing, because they contain a subset of already
313 upstream changes, the behavior towards them is controlled by
316 By default (or if `--no-reapply-cherry-picks` is given), these commits
317 will be automatically dropped. Because this necessitates reading all
318 upstream commits, this can be expensive in repos with a large number
319 of upstream commits that need to be read.
321 `--reapply-cherry-picks` allows rebase to forgo reading all upstream
322 commits, potentially improving performance.
324 See also INCOMPATIBLE OPTIONS below.
326 --allow-empty-message::
327 No-op. Rebasing commits with an empty message used to fail
328 and this option would override that behavior, allowing commits
329 with empty messages to be rebased. Now commits with an empty
330 message do not cause rebasing to halt.
332 See also INCOMPATIBLE OPTIONS below.
335 Restart the rebasing process by skipping the current patch.
338 Edit the todo list during an interactive rebase.
340 --show-current-patch::
341 Show the current patch in an interactive rebase or when rebase
342 is stopped because of conflicts. This is the equivalent of
343 `git show REBASE_HEAD`.
347 Use merging strategies to rebase. When the recursive (default) merge
348 strategy is used, this allows rebase to be aware of renames on the
349 upstream side. This is the default.
351 Note that a rebase merge works by replaying each commit from the working
352 branch on top of the <upstream> branch. Because of this, when a merge
353 conflict happens, the side reported as 'ours' is the so-far rebased
354 series, starting with <upstream>, and 'theirs' is the working branch. In
355 other words, the sides are swapped.
357 See also INCOMPATIBLE OPTIONS below.
360 --strategy=<strategy>::
361 Use the given merge strategy.
362 If there is no `-s` option 'git merge-recursive' is used
363 instead. This implies --merge.
365 Because 'git rebase' replays each commit from the working branch
366 on top of the <upstream> branch using the given strategy, using
367 the 'ours' strategy simply empties all patches from the <branch>,
368 which makes little sense.
370 See also INCOMPATIBLE OPTIONS below.
372 -X <strategy-option>::
373 --strategy-option=<strategy-option>::
374 Pass the <strategy-option> through to the merge strategy.
375 This implies `--merge` and, if no strategy has been
376 specified, `-s recursive`. Note the reversal of 'ours' and
377 'theirs' as noted above for the `-m` option.
379 See also INCOMPATIBLE OPTIONS below.
381 --rerere-autoupdate::
382 --no-rerere-autoupdate::
383 Allow the rerere mechanism to update the index with the
384 result of auto-conflict resolution if possible.
387 --gpg-sign[=<keyid>]::
389 GPG-sign commits. The `keyid` argument is optional and
390 defaults to the committer identity; if specified, it must be
391 stuck to the option without a space. `--no-gpg-sign` is useful to
392 countermand both `commit.gpgSign` configuration variable, and
393 earlier `--gpg-sign`.
397 Be quiet. Implies --no-stat.
401 Be verbose. Implies --stat.
404 Show a diffstat of what changed upstream since the last rebase. The
405 diffstat is also controlled by the configuration option rebase.stat.
409 Do not show a diffstat as part of the rebase process.
412 This option bypasses the pre-rebase hook. See also linkgit:githooks[5].
415 Allows the pre-rebase hook to run, which is the default. This option can
416 be used to override --no-verify. See also linkgit:githooks[5].
419 Ensure at least <n> lines of surrounding context match before
420 and after each change. When fewer lines of surrounding
421 context exist they all must match. By default no context is
422 ever ignored. Implies --apply.
424 See also INCOMPATIBLE OPTIONS below.
429 Individually replay all rebased commits instead of fast-forwarding
430 over the unchanged ones. This ensures that the entire history of
431 the rebased branch is composed of new commits.
433 You may find this helpful after reverting a topic branch merge, as this option
434 recreates the topic branch with fresh commits so it can be remerged
435 successfully without needing to "revert the reversion" (see the
436 link:howto/revert-a-faulty-merge.html[revert-a-faulty-merge How-To] for
441 Use reflog to find a better common ancestor between <upstream>
442 and <branch> when calculating which commits have been
443 introduced by <branch>.
445 When --fork-point is active, 'fork_point' will be used instead of
446 <upstream> to calculate the set of commits to rebase, where
447 'fork_point' is the result of `git merge-base --fork-point <upstream>
448 <branch>` command (see linkgit:git-merge-base[1]). If 'fork_point'
449 ends up being empty, the <upstream> will be used as a fallback.
451 If either <upstream> or --root is given on the command line, then the
452 default is `--no-fork-point`, otherwise the default is `--fork-point`.
454 If your branch was based on <upstream> but <upstream> was rewound and
455 your branch contains commits which were dropped, this option can be used
456 with `--keep-base` in order to drop those commits from your branch.
458 --ignore-whitespace::
459 --whitespace=<option>::
460 These flags are passed to the 'git apply' program
461 (see linkgit:git-apply[1]) that applies the patch.
464 See also INCOMPATIBLE OPTIONS below.
466 --committer-date-is-author-date::
468 These flags are passed to 'git am' to easily change the dates
469 of the rebased commits (see linkgit:git-am[1]).
471 See also INCOMPATIBLE OPTIONS below.
474 Add a Signed-off-by: trailer to all the rebased commits. Note
475 that if `--interactive` is given then only commits marked to be
476 picked, edited or reworded will have the trailer added.
478 See also INCOMPATIBLE OPTIONS below.
482 Make a list of the commits which are about to be rebased. Let the
483 user edit that list before rebasing. This mode can also be used to
484 split commits (see SPLITTING COMMITS below).
486 The commit list format can be changed by setting the configuration option
487 rebase.instructionFormat. A customized instruction format will automatically
488 have the long commit hash prepended to the format.
490 See also INCOMPATIBLE OPTIONS below.
493 --rebase-merges[=(rebase-cousins|no-rebase-cousins)]::
494 By default, a rebase will simply drop merge commits from the todo
495 list, and put the rebased commits into a single, linear branch.
496 With `--rebase-merges`, the rebase will instead try to preserve
497 the branching structure within the commits that are to be rebased,
498 by recreating the merge commits. Any resolved merge conflicts or
499 manual amendments in these merge commits will have to be
500 resolved/re-applied manually.
502 By default, or when `no-rebase-cousins` was specified, commits which do not
503 have `<upstream>` as direct ancestor will keep their original branch point,
504 i.e. commits that would be excluded by linkgit:git-log[1]'s
505 `--ancestry-path` option will keep their original ancestry by default. If
506 the `rebase-cousins` mode is turned on, such commits are instead rebased
507 onto `<upstream>` (or `<onto>`, if specified).
509 The `--rebase-merges` mode is similar in spirit to the deprecated
510 `--preserve-merges` but works with interactive rebases,
511 where commits can be reordered, inserted and dropped at will.
513 It is currently only possible to recreate the merge commits using the
514 `recursive` merge strategy; Different merge strategies can be used only via
515 explicit `exec git merge -s <strategy> [...]` commands.
517 See also REBASING MERGES and INCOMPATIBLE OPTIONS below.
521 [DEPRECATED: use `--rebase-merges` instead] Recreate merge commits
522 instead of flattening the history by replaying commits a merge commit
523 introduces. Merge conflict resolutions or manual amendments to merge
524 commits are not preserved.
526 This uses the `--interactive` machinery internally, but combining it
527 with the `--interactive` option explicitly is generally not a good
528 idea unless you know what you are doing (see BUGS below).
530 See also INCOMPATIBLE OPTIONS below.
534 Append "exec <cmd>" after each line creating a commit in the
535 final history. <cmd> will be interpreted as one or more shell
536 commands. Any command that fails will interrupt the rebase,
539 You may execute several commands by either using one instance of `--exec`
540 with several commands:
542 git rebase -i --exec "cmd1 && cmd2 && ..."
544 or by giving more than one `--exec`:
546 git rebase -i --exec "cmd1" --exec "cmd2" --exec ...
548 If `--autosquash` is used, "exec" lines will not be appended for
549 the intermediate commits, and will only appear at the end of each
552 This uses the `--interactive` machinery internally, but it can be run
553 without an explicit `--interactive`.
555 See also INCOMPATIBLE OPTIONS below.
558 Rebase all commits reachable from <branch>, instead of
559 limiting them with an <upstream>. This allows you to rebase
560 the root commit(s) on a branch. When used with --onto, it
561 will skip changes already contained in <newbase> (instead of
562 <upstream>) whereas without --onto it will operate on every change.
563 When used together with both --onto and --preserve-merges,
564 'all' root commits will be rewritten to have <newbase> as parent
567 See also INCOMPATIBLE OPTIONS below.
571 When the commit log message begins with "squash! ..." (or
572 "fixup! ..."), and there is already a commit in the todo list that
573 matches the same `...`, automatically modify the todo list of rebase
574 -i so that the commit marked for squashing comes right after the
575 commit to be modified, and change the action of the moved commit
576 from `pick` to `squash` (or `fixup`). A commit matches the `...` if
577 the commit subject matches, or if the `...` refers to the commit's
578 hash. As a fall-back, partial matches of the commit subject work,
579 too. The recommended way to create fixup/squash commits is by using
580 the `--fixup`/`--squash` options of linkgit:git-commit[1].
582 If the `--autosquash` option is enabled by default using the
583 configuration variable `rebase.autoSquash`, this option can be
584 used to override and disable this setting.
586 See also INCOMPATIBLE OPTIONS below.
590 Automatically create a temporary stash entry before the operation
591 begins, and apply it after the operation ends. This means
592 that you can run rebase on a dirty worktree. However, use
593 with care: the final stash application after a successful
594 rebase might result in non-trivial conflicts.
596 --reschedule-failed-exec::
597 --no-reschedule-failed-exec::
598 Automatically reschedule `exec` commands that failed. This only makes
599 sense in interactive mode (or when an `--exec` option was provided).
604 The following options:
607 * --committer-date-is-author-date
609 * --ignore-whitespace
613 are incompatible with the following options:
618 * --allow-empty-message
626 * --reapply-cherry-picks
628 * --root when used in combination with --onto
630 In addition, the following pairs of options are incompatible:
632 * --preserve-merges and --interactive
633 * --preserve-merges and --signoff
634 * --preserve-merges and --rebase-merges
635 * --preserve-merges and --empty=
636 * --keep-base and --onto
637 * --keep-base and --root
639 BEHAVIORAL DIFFERENCES
640 -----------------------
642 git rebase has two primary backends: apply and merge. (The apply
643 backend used to be known as the 'am' backend, but the name led to
644 confusion as it looks like a verb instead of a noun. Also, the merge
645 backend used to be known as the interactive backend, but it is now
646 used for non-interactive cases as well. Both were renamed based on
647 lower-level functionality that underpinned each.) There are some
648 subtle differences in how these two backends behave:
653 The apply backend unfortunately drops intentionally empty commits, i.e.
654 commits that started empty, though these are rare in practice. It
655 also drops commits that become empty and has no option for controlling
658 The merge backend keeps intentionally empty commits by default (though
659 with -i they are marked as empty in the todo list editor, or they can
660 be dropped automatically with --no-keep-empty).
662 Similar to the apply backend, by default the merge backend drops
663 commits that become empty unless -i/--interactive is specified (in
664 which case it stops and asks the user what to do). The merge backend
665 also has an --empty={drop,keep,ask} option for changing the behavior
666 of handling commits that become empty.
668 Directory rename detection
669 ~~~~~~~~~~~~~~~~~~~~~~~~~~
671 Due to the lack of accurate tree information (arising from
672 constructing fake ancestors with the limited information available in
673 patches), directory rename detection is disabled in the apply backend.
674 Disabled directory rename detection means that if one side of history
675 renames a directory and the other adds new files to the old directory,
676 then the new files will be left behind in the old directory without
677 any warning at the time of rebasing that you may want to move these
678 files into the new directory.
680 Directory rename detection works with the merge backend to provide you
681 warnings in such cases.
686 The apply backend works by creating a sequence of patches (by calling
687 `format-patch` internally), and then applying the patches in sequence
688 (calling `am` internally). Patches are composed of multiple hunks,
689 each with line numbers, a context region, and the actual changes. The
690 line numbers have to be taken with some fuzz, since the other side
691 will likely have inserted or deleted lines earlier in the file. The
692 context region is meant to help find how to adjust the line numbers in
693 order to apply the changes to the right lines. However, if multiple
694 areas of the code have the same surrounding lines of context, the
695 wrong one can be picked. There are real-world cases where this has
696 caused commits to be reapplied incorrectly with no conflicts reported.
697 Setting diff.context to a larger value may prevent such types of
698 problems, but increases the chance of spurious conflicts (since it
699 will require more lines of matching context to apply).
701 The merge backend works with a full copy of each relevant file,
702 insulating it from these types of problems.
704 Labelling of conflicts markers
705 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
707 When there are content conflicts, the merge machinery tries to
708 annotate each side's conflict markers with the commits where the
709 content came from. Since the apply backend drops the original
710 information about the rebased commits and their parents (and instead
711 generates new fake commits based off limited information in the
712 generated patches), those commits cannot be identified; instead it has
713 to fall back to a commit summary. Also, when merge.conflictStyle is
714 set to diff3, the apply backend will use "constructed merge base" to
715 label the content from the merge base, and thus provide no information
716 about the merge base commit whatsoever.
718 The merge backend works with the full commits on both sides of history
719 and thus has no such limitations.
724 The apply backend has not traditionally called the post-commit hook,
725 while the merge backend has. However, this was by accident of
726 implementation rather than by design. Both backends should have the
727 same behavior, though it is not clear which one is correct.
732 The apply backend has safety problems with an ill-timed interrupt; if
733 the user presses Ctrl-C at the wrong time to try to abort the rebase,
734 the rebase can enter a state where it cannot be aborted with a
735 subsequent `git rebase --abort`. The merge backend does not appear to
736 suffer from the same shortcoming. (See
737 https://lore.kernel.org/git/20200207132152.GC2868@szeder.dev/ for
743 When a conflict occurs while rebasing, rebase stops and asks the user
744 to resolve. Since the user may need to make notable changes while
745 resolving conflicts, after conflicts are resolved and the user has run
746 `git rebase --continue`, the rebase should open an editor and ask the
747 user to update the commit message. The merge backend does this, while
748 the apply backend blindly applies the original commit message.
750 Miscellaneous differences
751 ~~~~~~~~~~~~~~~~~~~~~~~~~
753 There are a few more behavioral differences that most folks would
754 probably consider inconsequential but which are mentioned for
757 * Reflog: The two backends will use different wording when describing
758 the changes made in the reflog, though both will make use of the
761 * Progress, informational, and error messages: The two backends
762 provide slightly different progress and informational messages.
763 Also, the apply backend writes error messages (such as "Your files
764 would be overwritten...") to stdout, while the merge backend writes
767 * State directories: The two backends keep their state in different
768 directories under .git/
770 include::merge-strategies.txt[]
775 You should understand the implications of using 'git rebase' on a
776 repository that you share. See also RECOVERING FROM UPSTREAM REBASE
779 When the git-rebase command is run, it will first execute a "pre-rebase"
780 hook if one exists. You can use this hook to do sanity checks and
781 reject the rebase if it isn't appropriate. Please see the template
782 pre-rebase hook script for an example.
784 Upon completion, <branch> will be the current branch.
789 Rebasing interactively means that you have a chance to edit the commits
790 which are rebased. You can reorder the commits, and you can
791 remove them (weeding out bad or otherwise unwanted patches).
793 The interactive mode is meant for this type of workflow:
795 1. have a wonderful idea
797 3. prepare a series for submission
800 where point 2. consists of several instances of
804 1. finish something worthy of a commit
809 1. realize that something does not work
813 Sometimes the thing fixed in b.2. cannot be amended to the not-quite
814 perfect commit it fixes, because that commit is buried deeply in a
815 patch series. That is exactly what interactive rebase is for: use it
816 after plenty of "a"s and "b"s, by rearranging and editing
817 commits, and squashing multiple commits into one.
819 Start it with the last commit you want to retain as-is:
821 git rebase -i <after-this-commit>
823 An editor will be fired up with all the commits in your current branch
824 (ignoring merge commits), which come after the given commit. You can
825 reorder the commits in this list to your heart's content, and you can
826 remove them. The list looks more or less like this:
828 -------------------------------------------
829 pick deadbee The oneline of this commit
830 pick fa1afe1 The oneline of the next commit
832 -------------------------------------------
834 The oneline descriptions are purely for your pleasure; 'git rebase' will
835 not look at them but at the commit names ("deadbee" and "fa1afe1" in this
836 example), so do not delete or edit the names.
838 By replacing the command "pick" with the command "edit", you can tell
839 'git rebase' to stop after applying that commit, so that you can edit
840 the files and/or the commit message, amend the commit, and continue
843 To interrupt the rebase (just like an "edit" command would do, but without
844 cherry-picking any commit first), use the "break" command.
846 If you just want to edit the commit message for a commit, replace the
847 command "pick" with the command "reword".
849 To drop a commit, replace the command "pick" with "drop", or just
850 delete the matching line.
852 If you want to fold two or more commits into one, replace the command
853 "pick" for the second and subsequent commits with "squash" or "fixup".
854 If the commits had different authors, the folded commit will be
855 attributed to the author of the first commit. The suggested commit
856 message for the folded commit is the concatenation of the commit
857 messages of the first commit and of those with the "squash" command,
858 but omits the commit messages of commits with the "fixup" command.
860 'git rebase' will stop when "pick" has been replaced with "edit" or
861 when a command fails due to merge errors. When you are done editing
862 and/or resolving conflicts you can continue with `git rebase --continue`.
864 For example, if you want to reorder the last 5 commits, such that what
865 was HEAD~4 becomes the new HEAD. To achieve that, you would call
866 'git rebase' like this:
868 ----------------------
869 $ git rebase -i HEAD~5
870 ----------------------
872 And move the first patch to the end of the list.
874 You might want to recreate merge commits, e.g. if you have a history
885 Suppose you want to rebase the side branch starting at "A" to "Q". Make
886 sure that the current HEAD is "B", and call
888 -----------------------------
889 $ git rebase -i -r --onto Q O
890 -----------------------------
892 Reordering and editing commits usually creates untested intermediate
893 steps. You may want to check that your history editing did not break
894 anything by running a test, or at least recompiling at intermediate
895 points in history by using the "exec" command (shortcut "x"). You may
896 do so by creating a todo list like this one:
898 -------------------------------------------
899 pick deadbee Implement feature XXX
900 fixup f1a5c00 Fix to feature XXX
902 pick c0ffeee The oneline of the next commit
903 edit deadbab The oneline of the commit after
904 exec cd subdir; make test
906 -------------------------------------------
908 The interactive rebase will stop when a command fails (i.e. exits with
909 non-0 status) to give you an opportunity to fix the problem. You can
910 continue with `git rebase --continue`.
912 The "exec" command launches the command in a shell (the one specified
913 in `$SHELL`, or the default shell if `$SHELL` is not set), so you can
914 use shell features (like "cd", ">", ";" ...). The command is run from
915 the root of the working tree.
917 ----------------------------------
918 $ git rebase -i --exec "make test"
919 ----------------------------------
921 This command lets you check that intermediate commits are compilable.
922 The todo list becomes like that:
938 In interactive mode, you can mark commits with the action "edit". However,
939 this does not necessarily mean that 'git rebase' expects the result of this
940 edit to be exactly one commit. Indeed, you can undo the commit, or you can
941 add other commits. This can be used to split a commit into two:
943 - Start an interactive rebase with `git rebase -i <commit>^`, where
944 <commit> is the commit you want to split. In fact, any commit range
945 will do, as long as it contains that commit.
947 - Mark the commit you want to split with the action "edit".
949 - When it comes to editing that commit, execute `git reset HEAD^`. The
950 effect is that the HEAD is rewound by one, and the index follows suit.
951 However, the working tree stays the same.
953 - Now add the changes to the index that you want to have in the first
954 commit. You can use `git add` (possibly interactively) or
955 'git gui' (or both) to do that.
957 - Commit the now-current index with whatever commit message is appropriate
960 - Repeat the last two steps until your working tree is clean.
962 - Continue the rebase with `git rebase --continue`.
964 If you are not absolutely sure that the intermediate revisions are
965 consistent (they compile, pass the testsuite, etc.) you should use
966 'git stash' to stash away the not-yet-committed changes
967 after each commit, test, and amend the commit if fixes are necessary.
970 RECOVERING FROM UPSTREAM REBASE
971 -------------------------------
973 Rebasing (or any other form of rewriting) a branch that others have
974 based work on is a bad idea: anyone downstream of it is forced to
975 manually fix their history. This section explains how to do the fix
976 from the downstream's point of view. The real fix, however, would be
977 to avoid rebasing the upstream in the first place.
979 To illustrate, suppose you are in a situation where someone develops a
980 'subsystem' branch, and you are working on a 'topic' that is dependent
981 on this 'subsystem'. You might end up with a history like the
985 o---o---o---o---o---o---o---o master
987 o---o---o---o---o subsystem
992 If 'subsystem' is rebased against 'master', the following happens:
995 o---o---o---o---o---o---o---o master
997 o---o---o---o---o o'--o'--o'--o'--o' subsystem
1002 If you now continue development as usual, and eventually merge 'topic'
1003 to 'subsystem', the commits from 'subsystem' will remain duplicated forever:
1006 o---o---o---o---o---o---o---o master
1008 o---o---o---o---o o'--o'--o'--o'--o'--M subsystem
1010 *---*---*-..........-*--* topic
1013 Such duplicates are generally frowned upon because they clutter up
1014 history, making it harder to follow. To clean things up, you need to
1015 transplant the commits on 'topic' to the new 'subsystem' tip, i.e.,
1016 rebase 'topic'. This becomes a ripple effect: anyone downstream from
1017 'topic' is forced to rebase too, and so on!
1019 There are two kinds of fixes, discussed in the following subsections:
1021 Easy case: The changes are literally the same.::
1023 This happens if the 'subsystem' rebase was a simple rebase and
1026 Hard case: The changes are not the same.::
1028 This happens if the 'subsystem' rebase had conflicts, or used
1029 `--interactive` to omit, edit, squash, or fixup commits; or
1030 if the upstream used one of `commit --amend`, `reset`, or
1031 a full history rewriting command like
1032 https://github.com/newren/git-filter-repo[`filter-repo`].
1038 Only works if the changes (patch IDs based on the diff contents) on
1039 'subsystem' are literally the same before and after the rebase
1042 In that case, the fix is easy because 'git rebase' knows to skip
1043 changes that are already present in the new upstream (unless
1044 `--reapply-cherry-picks` is given). So if you say
1045 (assuming you're on 'topic')
1047 $ git rebase subsystem
1049 you will end up with the fixed history
1051 o---o---o---o---o---o---o---o master
1053 o'--o'--o'--o'--o' subsystem
1062 Things get more complicated if the 'subsystem' changes do not exactly
1063 correspond to the ones before the rebase.
1065 NOTE: While an "easy case recovery" sometimes appears to be successful
1066 even in the hard case, it may have unintended consequences. For
1067 example, a commit that was removed via `git rebase
1068 --interactive` will be **resurrected**!
1070 The idea is to manually tell 'git rebase' "where the old 'subsystem'
1071 ended and your 'topic' began", that is, what the old merge base
1072 between them was. You will have to find a way to name the last commit
1073 of the old 'subsystem', for example:
1075 * With the 'subsystem' reflog: after 'git fetch', the old tip of
1076 'subsystem' is at `subsystem@{1}`. Subsequent fetches will
1077 increase the number. (See linkgit:git-reflog[1].)
1079 * Relative to the tip of 'topic': knowing that your 'topic' has three
1080 commits, the old tip of 'subsystem' must be `topic~3`.
1082 You can then transplant the old `subsystem..topic` to the new tip by
1083 saying (for the reflog case, and assuming you are on 'topic' already):
1085 $ git rebase --onto subsystem subsystem@{1}
1088 The ripple effect of a "hard case" recovery is especially bad:
1089 'everyone' downstream from 'topic' will now have to perform a "hard
1095 The interactive rebase command was originally designed to handle
1096 individual patch series. As such, it makes sense to exclude merge
1097 commits from the todo list, as the developer may have merged the
1098 then-current `master` while working on the branch, only to rebase
1099 all the commits onto `master` eventually (skipping the merge
1102 However, there are legitimate reasons why a developer may want to
1103 recreate merge commits: to keep the branch structure (or "commit
1104 topology") when working on multiple, inter-related branches.
1106 In the following example, the developer works on a topic branch that
1107 refactors the way buttons are defined, and on another topic branch
1108 that uses that refactoring to implement a "Report a bug" button. The
1109 output of `git log --graph --format=%s -5` may look like this:
1112 * Merge branch 'report-a-bug'
1114 | * Add the feedback button
1115 * | Merge branch 'refactor-button'
1118 | * Use the Button class for all buttons
1119 | * Extract a generic Button class from the DownloadButton one
1122 The developer might want to rebase those commits to a newer `master`
1123 while keeping the branch topology, for example when the first topic
1124 branch is expected to be integrated into `master` much earlier than the
1125 second one, say, to resolve merge conflicts with changes to the
1126 DownloadButton class that made it into `master`.
1128 This rebase can be performed using the `--rebase-merges` option.
1129 It will generate a todo list looking like this:
1134 # Branch: refactor-button
1136 pick 123456 Extract a generic Button class from the DownloadButton one
1137 pick 654321 Use the Button class for all buttons
1138 label refactor-button
1140 # Branch: report-a-bug
1141 reset refactor-button # Use the Button class for all buttons
1142 pick abcdef Add the feedback button
1146 merge -C a1b2c3 refactor-button # Merge 'refactor-button'
1147 merge -C 6f5e4d report-a-bug # Merge 'report-a-bug'
1150 In contrast to a regular interactive rebase, there are `label`, `reset`
1151 and `merge` commands in addition to `pick` ones.
1153 The `label` command associates a label with the current HEAD when that
1154 command is executed. These labels are created as worktree-local refs
1155 (`refs/rewritten/<label>`) that will be deleted when the rebase
1156 finishes. That way, rebase operations in multiple worktrees linked to
1157 the same repository do not interfere with one another. If the `label`
1158 command fails, it is rescheduled immediately, with a helpful message how
1161 The `reset` command resets the HEAD, index and worktree to the specified
1162 revision. It is similar to an `exec git reset --hard <label>`, but
1163 refuses to overwrite untracked files. If the `reset` command fails, it is
1164 rescheduled immediately, with a helpful message how to edit the todo list
1165 (this typically happens when a `reset` command was inserted into the todo
1166 list manually and contains a typo).
1168 The `merge` command will merge the specified revision(s) into whatever
1169 is HEAD at that time. With `-C <original-commit>`, the commit message of
1170 the specified merge commit will be used. When the `-C` is changed to
1171 a lower-case `-c`, the message will be opened in an editor after a
1172 successful merge so that the user can edit the message.
1174 If a `merge` command fails for any reason other than merge conflicts (i.e.
1175 when the merge operation did not even start), it is rescheduled immediately.
1177 At this time, the `merge` command will *always* use the `recursive`
1178 merge strategy for regular merges, and `octopus` for octopus merges,
1179 with no way to choose a different one. To work around
1180 this, an `exec` command can be used to call `git merge` explicitly,
1181 using the fact that the labels are worktree-local refs (the ref
1182 `refs/rewritten/onto` would correspond to the label `onto`, for example).
1184 Note: the first command (`label onto`) labels the revision onto which
1185 the commits are rebased; The name `onto` is just a convention, as a nod
1186 to the `--onto` option.
1188 It is also possible to introduce completely new merge commits from scratch
1189 by adding a command of the form `merge <merge-head>`. This form will
1190 generate a tentative commit message and always open an editor to let the
1191 user edit it. This can be useful e.g. when a topic branch turns out to
1192 address more than a single concern and wants to be split into two or
1193 even more topic branches. Consider this todo list:
1196 pick 192837 Switch from GNU Makefiles to CMake
1197 pick 5a6c7e Document the switch to CMake
1198 pick 918273 Fix detection of OpenSSL in CMake
1199 pick afbecd http: add support for TLS v1.3
1200 pick fdbaec Fix detection of cURL in CMake on Windows
1203 The one commit in this list that is not related to CMake may very well
1204 have been motivated by working on fixing all those bugs introduced by
1205 switching to CMake, but it addresses a different concern. To split this
1206 branch into two topic branches, the todo list could be edited like this:
1211 pick afbecd http: add support for TLS v1.3
1215 pick 192837 Switch from GNU Makefiles to CMake
1216 pick 918273 Fix detection of OpenSSL in CMake
1217 pick fdbaec Fix detection of cURL in CMake on Windows
1218 pick 5a6c7e Document the switch to CMake
1228 The todo list presented by the deprecated `--preserve-merges --interactive`
1229 does not represent the topology of the revision graph (use `--rebase-merges`
1230 instead). Editing commits and rewording their commit messages should work
1231 fine, but attempts to reorder commits tend to produce counterintuitive results.
1232 Use `--rebase-merges` in such scenarios instead.
1234 For example, an attempt to rearrange
1236 1 --- 2 --- 3 --- 4 --- 5
1240 1 --- 2 --- 4 --- 3 --- 5
1242 by moving the "pick 4" line will result in the following history:
1251 Part of the linkgit:git[1] suite