contrib/subtree: ensure only one rev is provided
[git/git.git] / Documentation / git-reset.txt
1 git-reset(1)
2 ============
3
4 NAME
5 ----
6 git-reset - Reset current HEAD to the specified state
7
8 SYNOPSIS
9 --------
10 [verse]
11 'git reset' [-q] [<tree-ish>] [--] <paths>...
12 'git reset' (--patch | -p) [<tree-ish>] [--] [<paths>...]
13 'git reset' [--soft | --mixed [-N] | --hard | --merge | --keep] [-q] [<commit>]
14
15 DESCRIPTION
16 -----------
17 In the first and second form, copy entries from `<tree-ish>` to the index.
18 In the third form, set the current branch head (`HEAD`) to `<commit>`,
19 optionally modifying index and working tree to match.
20 The `<tree-ish>`/`<commit>` defaults to `HEAD` in all forms.
21
22 'git reset' [-q] [<tree-ish>] [--] <paths>...::
23 This form resets the index entries for all `<paths>` to their
24 state at `<tree-ish>`. (It does not affect the working tree or
25 the current branch.)
26 +
27 This means that `git reset <paths>` is the opposite of `git add
28 <paths>`.
29 +
30 After running `git reset <paths>` to update the index entry, you can
31 use linkgit:git-checkout[1] to check the contents out of the index to
32 the working tree.
33 Alternatively, using linkgit:git-checkout[1] and specifying a commit, you
34 can copy the contents of a path out of a commit to the index and to the
35 working tree in one go.
36
37 'git reset' (--patch | -p) [<tree-ish>] [--] [<paths>...]::
38 Interactively select hunks in the difference between the index
39 and `<tree-ish>` (defaults to `HEAD`). The chosen hunks are applied
40 in reverse to the index.
41 +
42 This means that `git reset -p` is the opposite of `git add -p`, i.e.
43 you can use it to selectively reset hunks. See the ``Interactive Mode''
44 section of linkgit:git-add[1] to learn how to operate the `--patch` mode.
45
46 'git reset' [<mode>] [<commit>]::
47 This form resets the current branch head to `<commit>` and
48 possibly updates the index (resetting it to the tree of `<commit>`) and
49 the working tree depending on `<mode>`. If `<mode>` is omitted,
50 defaults to `--mixed`. The `<mode>` must be one of the following:
51 +
52 --
53 --soft::
54 Does not touch the index file or the working tree at all (but
55 resets the head to `<commit>`, just like all modes do). This leaves
56 all your changed files "Changes to be committed", as `git status`
57 would put it.
58
59 --mixed::
60 Resets the index but not the working tree (i.e., the changed files
61 are preserved but not marked for commit) and reports what has not
62 been updated. This is the default action.
63 +
64 If `-N` is specified, removed paths are marked as intent-to-add (see
65 linkgit:git-add[1]).
66
67 --hard::
68 Resets the index and working tree. Any changes to tracked files in the
69 working tree since `<commit>` are discarded.
70
71 --merge::
72 Resets the index and updates the files in the working tree that are
73 different between `<commit>` and `HEAD`, but keeps those which are
74 different between the index and working tree (i.e. which have changes
75 which have not been added).
76 If a file that is different between `<commit>` and the index has
77 unstaged changes, reset is aborted.
78 +
79 In other words, `--merge` does something like a `git read-tree -u -m <commit>`,
80 but carries forward unmerged index entries.
81
82 --keep::
83 Resets index entries and updates files in the working tree that are
84 different between `<commit>` and `HEAD`.
85 If a file that is different between `<commit>` and `HEAD` has local
86 changes, reset is aborted.
87 --
88
89 If you want to undo a commit other than the latest on a branch,
90 linkgit:git-revert[1] is your friend.
91
92
93 OPTIONS
94 -------
95
96 -q::
97 --quiet::
98 --no-quiet::
99 Be quiet, only report errors. The default behavior is set by the
100 `reset.quiet` config option. `--quiet` and `--no-quiet` will
101 override the default behavior.
102
103
104 EXAMPLES
105 --------
106
107 Undo add::
108 +
109 ------------
110 $ edit <1>
111 $ git add frotz.c filfre.c
112 $ mailx <2>
113 $ git reset <3>
114 $ git pull git://info.example.com/ nitfol <4>
115 ------------
116 +
117 <1> You are happily working on something, and find the changes
118 in these files are in good order. You do not want to see them
119 when you run `git diff`, because you plan to work on other files
120 and changes with these files are distracting.
121 <2> Somebody asks you to pull, and the changes sound worthy of merging.
122 <3> However, you already dirtied the index (i.e. your index does
123 not match the `HEAD` commit). But you know the pull you are going
124 to make does not affect `frotz.c` or `filfre.c`, so you revert the
125 index changes for these two files. Your changes in working tree
126 remain there.
127 <4> Then you can pull and merge, leaving `frotz.c` and `filfre.c`
128 changes still in the working tree.
129
130 Undo a commit and redo::
131 +
132 ------------
133 $ git commit ...
134 $ git reset --soft HEAD^ <1>
135 $ edit <2>
136 $ git commit -a -c ORIG_HEAD <3>
137 ------------
138 +
139 <1> This is most often done when you remembered what you
140 just committed is incomplete, or you misspelled your commit
141 message, or both. Leaves working tree as it was before "reset".
142 <2> Make corrections to working tree files.
143 <3> "reset" copies the old head to `.git/ORIG_HEAD`; redo the
144 commit by starting with its log message. If you do not need to
145 edit the message further, you can give `-C` option instead.
146 +
147 See also the `--amend` option to linkgit:git-commit[1].
148
149 Undo a commit, making it a topic branch::
150 +
151 ------------
152 $ git branch topic/wip <1>
153 $ git reset --hard HEAD~3 <2>
154 $ git checkout topic/wip <3>
155 ------------
156 +
157 <1> You have made some commits, but realize they were premature
158 to be in the `master` branch. You want to continue polishing
159 them in a topic branch, so create `topic/wip` branch off of the
160 current `HEAD`.
161 <2> Rewind the master branch to get rid of those three commits.
162 <3> Switch to `topic/wip` branch and keep working.
163
164 Undo commits permanently::
165 +
166 ------------
167 $ git commit ...
168 $ git reset --hard HEAD~3 <1>
169 ------------
170 +
171 <1> The last three commits (`HEAD`, `HEAD^`, and `HEAD~2`) were bad
172 and you do not want to ever see them again. Do *not* do this if
173 you have already given these commits to somebody else. (See the
174 "RECOVERING FROM UPSTREAM REBASE" section in linkgit:git-rebase[1]
175 for the implications of doing so.)
176
177 Undo a merge or pull::
178 +
179 ------------
180 $ git pull <1>
181 Auto-merging nitfol
182 CONFLICT (content): Merge conflict in nitfol
183 Automatic merge failed; fix conflicts and then commit the result.
184 $ git reset --hard <2>
185 $ git pull . topic/branch <3>
186 Updating from 41223... to 13134...
187 Fast-forward
188 $ git reset --hard ORIG_HEAD <4>
189 ------------
190 +
191 <1> Try to update from the upstream resulted in a lot of
192 conflicts; you were not ready to spend a lot of time merging
193 right now, so you decide to do that later.
194 <2> "pull" has not made merge commit, so `git reset --hard`
195 which is a synonym for `git reset --hard HEAD` clears the mess
196 from the index file and the working tree.
197 <3> Merge a topic branch into the current branch, which resulted
198 in a fast-forward.
199 <4> But you decided that the topic branch is not ready for public
200 consumption yet. "pull" or "merge" always leaves the original
201 tip of the current branch in `ORIG_HEAD`, so resetting hard to it
202 brings your index file and the working tree back to that state,
203 and resets the tip of the branch to that commit.
204
205 Undo a merge or pull inside a dirty working tree::
206 +
207 ------------
208 $ git pull <1>
209 Auto-merging nitfol
210 Merge made by recursive.
211 nitfol | 20 +++++----
212 ...
213 $ git reset --merge ORIG_HEAD <2>
214 ------------
215 +
216 <1> Even if you may have local modifications in your
217 working tree, you can safely say `git pull` when you know
218 that the change in the other branch does not overlap with
219 them.
220 <2> After inspecting the result of the merge, you may find
221 that the change in the other branch is unsatisfactory. Running
222 `git reset --hard ORIG_HEAD` will let you go back to where you
223 were, but it will discard your local changes, which you do not
224 want. `git reset --merge` keeps your local changes.
225
226
227 Interrupted workflow::
228 +
229 Suppose you are interrupted by an urgent fix request while you
230 are in the middle of a large change. The files in your
231 working tree are not in any shape to be committed yet, but you
232 need to get to the other branch for a quick bugfix.
233 +
234 ------------
235 $ git checkout feature ;# you were working in "feature" branch and
236 $ work work work ;# got interrupted
237 $ git commit -a -m "snapshot WIP" <1>
238 $ git checkout master
239 $ fix fix fix
240 $ git commit ;# commit with real log
241 $ git checkout feature
242 $ git reset --soft HEAD^ ;# go back to WIP state <2>
243 $ git reset <3>
244 ------------
245 +
246 <1> This commit will get blown away so a throw-away log message is OK.
247 <2> This removes the 'WIP' commit from the commit history, and sets
248 your working tree to the state just before you made that snapshot.
249 <3> At this point the index file still has all the WIP changes you
250 committed as 'snapshot WIP'. This updates the index to show your
251 WIP files as uncommitted.
252 +
253 See also linkgit:git-stash[1].
254
255 Reset a single file in the index::
256 +
257 Suppose you have added a file to your index, but later decide you do not
258 want to add it to your commit. You can remove the file from the index
259 while keeping your changes with git reset.
260 +
261 ------------
262 $ git reset -- frotz.c <1>
263 $ git commit -m "Commit files in index" <2>
264 $ git add frotz.c <3>
265 ------------
266 +
267 <1> This removes the file from the index while keeping it in the working
268 directory.
269 <2> This commits all other changes in the index.
270 <3> Adds the file to the index again.
271
272 Keep changes in working tree while discarding some previous commits::
273 +
274 Suppose you are working on something and you commit it, and then you
275 continue working a bit more, but now you think that what you have in
276 your working tree should be in another branch that has nothing to do
277 with what you committed previously. You can start a new branch and
278 reset it while keeping the changes in your working tree.
279 +
280 ------------
281 $ git tag start
282 $ git checkout -b branch1
283 $ edit
284 $ git commit ... <1>
285 $ edit
286 $ git checkout -b branch2 <2>
287 $ git reset --keep start <3>
288 ------------
289 +
290 <1> This commits your first edits in `branch1`.
291 <2> In the ideal world, you could have realized that the earlier
292 commit did not belong to the new topic when you created and switched
293 to `branch2` (i.e. `git checkout -b branch2 start`), but nobody is
294 perfect.
295 <3> But you can use `reset --keep` to remove the unwanted commit after
296 you switched to `branch2`.
297
298 Split a commit apart into a sequence of commits::
299 +
300 Suppose that you have created lots of logically separate changes and committed
301 them together. Then, later you decide that it might be better to have each
302 logical chunk associated with its own commit. You can use git reset to rewind
303 history without changing the contents of your local files, and then successively
304 use `git add -p` to interactively select which hunks to include into each commit,
305 using `git commit -c` to pre-populate the commit message.
306 +
307 ------------
308 $ git reset -N HEAD^ <1>
309 $ git add -p <2>
310 $ git diff --cached <3>
311 $ git commit -c HEAD@{1} <4>
312 ... <5>
313 $ git add ... <6>
314 $ git diff --cached <7>
315 $ git commit ... <8>
316 ------------
317 +
318 <1> First, reset the history back one commit so that we remove the original
319 commit, but leave the working tree with all the changes. The -N ensures
320 that any new files added with `HEAD` are still marked so that `git add -p`
321 will find them.
322 <2> Next, we interactively select diff hunks to add using the `git add -p`
323 facility. This will ask you about each diff hunk in sequence and you can
324 use simple commands such as "yes, include this", "No don't include this"
325 or even the very powerful "edit" facility.
326 <3> Once satisfied with the hunks you want to include, you should verify what
327 has been prepared for the first commit by using `git diff --cached`. This
328 shows all the changes that have been moved into the index and are about
329 to be committed.
330 <4> Next, commit the changes stored in the index. The `-c` option specifies to
331 pre-populate the commit message from the original message that you started
332 with in the first commit. This is helpful to avoid retyping it. The
333 `HEAD@{1}` is a special notation for the commit that `HEAD` used to be at
334 prior to the original reset commit (1 change ago).
335 See linkgit:git-reflog[1] for more details. You may also use any other
336 valid commit reference.
337 <5> You can repeat steps 2-4 multiple times to break the original code into
338 any number of commits.
339 <6> Now you've split out many of the changes into their own commits, and might
340 no longer use the patch mode of `git add`, in order to select all remaining
341 uncommitted changes.
342 <7> Once again, check to verify that you've included what you want to. You may
343 also wish to verify that git diff doesn't show any remaining changes to be
344 committed later.
345 <8> And finally create the final commit.
346
347
348 DISCUSSION
349 ----------
350
351 The tables below show what happens when running:
352
353 ----------
354 git reset --option target
355 ----------
356
357 to reset the `HEAD` to another commit (`target`) with the different
358 reset options depending on the state of the files.
359
360 In these tables, `A`, `B`, `C` and `D` are some different states of a
361 file. For example, the first line of the first table means that if a
362 file is in state `A` in the working tree, in state `B` in the index, in
363 state `C` in `HEAD` and in state `D` in the target, then `git reset --soft
364 target` will leave the file in the working tree in state `A` and in the
365 index in state `B`. It resets (i.e. moves) the `HEAD` (i.e. the tip of
366 the current branch, if you are on one) to `target` (which has the file
367 in state `D`).
368
369 ....
370 working index HEAD target working index HEAD
371 ----------------------------------------------------
372 A B C D --soft A B D
373 --mixed A D D
374 --hard D D D
375 --merge (disallowed)
376 --keep (disallowed)
377 ....
378
379 ....
380 working index HEAD target working index HEAD
381 ----------------------------------------------------
382 A B C C --soft A B C
383 --mixed A C C
384 --hard C C C
385 --merge (disallowed)
386 --keep A C C
387 ....
388
389 ....
390 working index HEAD target working index HEAD
391 ----------------------------------------------------
392 B B C D --soft B B D
393 --mixed B D D
394 --hard D D D
395 --merge D D D
396 --keep (disallowed)
397 ....
398
399 ....
400 working index HEAD target working index HEAD
401 ----------------------------------------------------
402 B B C C --soft B B C
403 --mixed B C C
404 --hard C C C
405 --merge C C C
406 --keep B C C
407 ....
408
409 ....
410 working index HEAD target working index HEAD
411 ----------------------------------------------------
412 B C C D --soft B C D
413 --mixed B D D
414 --hard D D D
415 --merge (disallowed)
416 --keep (disallowed)
417 ....
418
419 ....
420 working index HEAD target working index HEAD
421 ----------------------------------------------------
422 B C C C --soft B C C
423 --mixed B C C
424 --hard C C C
425 --merge B C C
426 --keep B C C
427 ....
428
429 `reset --merge` is meant to be used when resetting out of a conflicted
430 merge. Any mergy operation guarantees that the working tree file that is
431 involved in the merge does not have local change wrt the index before
432 it starts, and that it writes the result out to the working tree. So if
433 we see some difference between the index and the target and also
434 between the index and the working tree, then it means that we are not
435 resetting out from a state that a mergy operation left after failing
436 with a conflict. That is why we disallow `--merge` option in this case.
437
438 `reset --keep` is meant to be used when removing some of the last
439 commits in the current branch while keeping changes in the working
440 tree. If there could be conflicts between the changes in the commit we
441 want to remove and the changes in the working tree we want to keep,
442 the reset is disallowed. That's why it is disallowed if there are both
443 changes between the working tree and `HEAD`, and between `HEAD` and the
444 target. To be safe, it is also disallowed when there are unmerged
445 entries.
446
447 The following tables show what happens when there are unmerged
448 entries:
449
450 ....
451 working index HEAD target working index HEAD
452 ----------------------------------------------------
453 X U A B --soft (disallowed)
454 --mixed X B B
455 --hard B B B
456 --merge B B B
457 --keep (disallowed)
458 ....
459
460 ....
461 working index HEAD target working index HEAD
462 ----------------------------------------------------
463 X U A A --soft (disallowed)
464 --mixed X A A
465 --hard A A A
466 --merge A A A
467 --keep (disallowed)
468 ....
469
470 `X` means any state and `U` means an unmerged index.
471
472 GIT
473 ---
474 Part of the linkgit:git[1] suite