Doc/checkout: checking out using @{-N} can lead to detached state
[git/git.git] / Documentation / git-checkout.txt
1 git-checkout(1)
2 ===============
3
4 NAME
5 ----
6 git-checkout - Switch branches or restore working tree files
7
8 SYNOPSIS
9 --------
10 [verse]
11 'git checkout' [-q] [-f] [-m] [<branch>]
12 'git checkout' [-q] [-f] [-m] --detach [<branch>]
13 'git checkout' [-q] [-f] [-m] [--detach] <commit>
14 'git checkout' [-q] [-f] [-m] [[-b|-B|--orphan] <new_branch>] [<start_point>]
15 'git checkout' [-f|--ours|--theirs|-m|--conflict=<style>] [<tree-ish>] [--] <paths>...
16 'git checkout' [-p|--patch] [<tree-ish>] [--] [<paths>...]
17
18 DESCRIPTION
19 -----------
20 Updates files in the working tree to match the version in the index
21 or the specified tree. If no paths are given, 'git checkout' will
22 also update `HEAD` to set the specified branch as the current
23 branch.
24
25 'git checkout' <branch>::
26 To prepare for working on <branch>, switch to it by updating
27 the index and the files in the working tree, and by pointing
28 HEAD at the branch. Local modifications to the files in the
29 working tree are kept, so that they can be committed to the
30 <branch>.
31 +
32 If <branch> is not found but there does exist a tracking branch in
33 exactly one remote (call it <remote>) with a matching name, treat as
34 equivalent to
35 +
36 ------------
37 $ git checkout -b <branch> --track <remote>/<branch>
38 ------------
39 +
40 You could omit <branch>, in which case the command degenerates to
41 "check out the current branch", which is a glorified no-op with a
42 rather expensive side-effects to show only the tracking information,
43 if exists, for the current branch.
44
45 'git checkout' -b|-B <new_branch> [<start point>]::
46
47 Specifying `-b` causes a new branch to be created as if
48 linkgit:git-branch[1] were called and then checked out. In
49 this case you can use the `--track` or `--no-track` options,
50 which will be passed to 'git branch'. As a convenience,
51 `--track` without `-b` implies branch creation; see the
52 description of `--track` below.
53 +
54 If `-B` is given, <new_branch> is created if it doesn't exist; otherwise, it
55 is reset. This is the transactional equivalent of
56 +
57 ------------
58 $ git branch -f <branch> [<start point>]
59 $ git checkout <branch>
60 ------------
61 +
62 that is to say, the branch is not reset/created unless "git checkout" is
63 successful.
64
65 'git checkout' --detach [<branch>]::
66 'git checkout' [--detach] <commit>::
67
68 Prepare to work on top of <commit>, by detaching HEAD at it
69 (see "DETACHED HEAD" section), and updating the index and the
70 files in the working tree. Local modifications to the files
71 in the working tree are kept, so that the resulting working
72 tree will be the state recorded in the commit plus the local
73 modifications.
74 +
75 When the <commit> argument is a branch name, the `--detach` option can
76 be used to detach HEAD at the tip of the branch (`git checkout
77 <branch>` would check out that branch without detaching HEAD).
78 +
79 Omitting <branch> detaches HEAD at the tip of the current branch.
80
81 'git checkout' [-p|--patch] [<tree-ish>] [--] <pathspec>...::
82
83 When <paths> or `--patch` are given, 'git checkout' does *not*
84 switch branches. It updates the named paths in the working tree
85 from the index file or from a named <tree-ish> (most often a
86 commit). In this case, the `-b` and `--track` options are
87 meaningless and giving either of them results in an error. The
88 <tree-ish> argument can be used to specify a specific tree-ish
89 (i.e. commit, tag or tree) to update the index for the given
90 paths before updating the working tree.
91 +
92 'git checkout' with <paths> or `--patch` is used to restore modified or
93 deleted paths to their original contents from the index or replace paths
94 with the contents from a named <tree-ish> (most often a commit-ish).
95 +
96 The index may contain unmerged entries because of a previous failed merge.
97 By default, if you try to check out such an entry from the index, the
98 checkout operation will fail and nothing will be checked out.
99 Using `-f` will ignore these unmerged entries. The contents from a
100 specific side of the merge can be checked out of the index by
101 using `--ours` or `--theirs`. With `-m`, changes made to the working tree
102 file can be discarded to re-create the original conflicted merge result.
103
104 OPTIONS
105 -------
106 -q::
107 --quiet::
108 Quiet, suppress feedback messages.
109
110 --[no-]progress::
111 Progress status is reported on the standard error stream
112 by default when it is attached to a terminal, unless `--quiet`
113 is specified. This flag enables progress reporting even if not
114 attached to a terminal, regardless of `--quiet`.
115
116 -f::
117 --force::
118 When switching branches, proceed even if the index or the
119 working tree differs from HEAD. This is used to throw away
120 local changes.
121 +
122 When checking out paths from the index, do not fail upon unmerged
123 entries; instead, unmerged entries are ignored.
124
125 --ours::
126 --theirs::
127 When checking out paths from the index, check out stage #2
128 ('ours') or #3 ('theirs') for unmerged paths.
129 +
130 Note that during `git rebase` and `git pull --rebase`, 'ours' and
131 'theirs' may appear swapped; `--ours` gives the version from the
132 branch the changes are rebased onto, while `--theirs` gives the
133 version from the branch that holds your work that is being rebased.
134 +
135 This is because `rebase` is used in a workflow that treats the
136 history at the remote as the shared canonical one, and treats the
137 work done on the branch you are rebasing as the third-party work to
138 be integrated, and you are temporarily assuming the role of the
139 keeper of the canonical history during the rebase. As the keeper of
140 the canonical history, you need to view the history from the remote
141 as `ours` (i.e. "our shared canonical history"), while what you did
142 on your side branch as `theirs` (i.e. "one contributor's work on top
143 of it").
144
145 -b <new_branch>::
146 Create a new branch named <new_branch> and start it at
147 <start_point>; see linkgit:git-branch[1] for details.
148
149 -B <new_branch>::
150 Creates the branch <new_branch> and start it at <start_point>;
151 if it already exists, then reset it to <start_point>. This is
152 equivalent to running "git branch" with "-f"; see
153 linkgit:git-branch[1] for details.
154
155 -t::
156 --track::
157 When creating a new branch, set up "upstream" configuration. See
158 "--track" in linkgit:git-branch[1] for details.
159 +
160 If no `-b` option is given, the name of the new branch will be
161 derived from the remote-tracking branch, by looking at the local part of
162 the refspec configured for the corresponding remote, and then stripping
163 the initial part up to the "*".
164 This would tell us to use "hack" as the local branch when branching
165 off of "origin/hack" (or "remotes/origin/hack", or even
166 "refs/remotes/origin/hack"). If the given name has no slash, or the above
167 guessing results in an empty name, the guessing is aborted. You can
168 explicitly give a name with `-b` in such a case.
169
170 --no-track::
171 Do not set up "upstream" configuration, even if the
172 branch.autoSetupMerge configuration variable is true.
173
174 -l::
175 Create the new branch's reflog; see linkgit:git-branch[1] for
176 details.
177
178 --detach::
179 Rather than checking out a branch to work on it, check out a
180 commit for inspection and discardable experiments.
181 This is the default behavior of "git checkout <commit>" when
182 <commit> is not a branch name. See the "DETACHED HEAD" section
183 below for details.
184
185 --orphan <new_branch>::
186 Create a new 'orphan' branch, named <new_branch>, started from
187 <start_point> and switch to it. The first commit made on this
188 new branch will have no parents and it will be the root of a new
189 history totally disconnected from all the other branches and
190 commits.
191 +
192 The index and the working tree are adjusted as if you had previously run
193 "git checkout <start_point>". This allows you to start a new history
194 that records a set of paths similar to <start_point> by easily running
195 "git commit -a" to make the root commit.
196 +
197 This can be useful when you want to publish the tree from a commit
198 without exposing its full history. You might want to do this to publish
199 an open source branch of a project whose current tree is "clean", but
200 whose full history contains proprietary or otherwise encumbered bits of
201 code.
202 +
203 If you want to start a disconnected history that records a set of paths
204 that is totally different from the one of <start_point>, then you should
205 clear the index and the working tree right after creating the orphan
206 branch by running "git rm -rf ." from the top level of the working tree.
207 Afterwards you will be ready to prepare your new files, repopulating the
208 working tree, by copying them from elsewhere, extracting a tarball, etc.
209
210 --ignore-skip-worktree-bits::
211 In sparse checkout mode, `git checkout -- <paths>` would
212 update only entries matched by <paths> and sparse patterns
213 in $GIT_DIR/info/sparse-checkout. This option ignores
214 the sparse patterns and adds back any files in <paths>.
215
216 -m::
217 --merge::
218 When switching branches,
219 if you have local modifications to one or more files that
220 are different between the current branch and the branch to
221 which you are switching, the command refuses to switch
222 branches in order to preserve your modifications in context.
223 However, with this option, a three-way merge between the current
224 branch, your working tree contents, and the new branch
225 is done, and you will be on the new branch.
226 +
227 When a merge conflict happens, the index entries for conflicting
228 paths are left unmerged, and you need to resolve the conflicts
229 and mark the resolved paths with `git add` (or `git rm` if the merge
230 should result in deletion of the path).
231 +
232 When checking out paths from the index, this option lets you recreate
233 the conflicted merge in the specified paths.
234
235 --conflict=<style>::
236 The same as --merge option above, but changes the way the
237 conflicting hunks are presented, overriding the
238 merge.conflictStyle configuration variable. Possible values are
239 "merge" (default) and "diff3" (in addition to what is shown by
240 "merge" style, shows the original contents).
241
242 -p::
243 --patch::
244 Interactively select hunks in the difference between the
245 <tree-ish> (or the index, if unspecified) and the working
246 tree. The chosen hunks are then applied in reverse to the
247 working tree (and if a <tree-ish> was specified, the index).
248 +
249 This means that you can use `git checkout -p` to selectively discard
250 edits from your current working tree. See the ``Interactive Mode''
251 section of linkgit:git-add[1] to learn how to operate the `--patch` mode.
252
253 --ignore-other-worktrees::
254 `git checkout` refuses when the wanted ref is already checked
255 out by another worktree. This option makes it check the ref
256 out anyway. In other words, the ref can be held by more than one
257 worktree.
258
259 <branch>::
260 Branch to checkout; if it refers to a branch (i.e., a name that,
261 when prepended with "refs/heads/", is a valid ref), then that
262 branch is checked out. Otherwise, if it refers to a valid
263 commit, your HEAD becomes "detached" and you are no longer on
264 any branch (see below for details).
265 +
266 You can use the `"@{-N}"` syntax to refer to the N-th last
267 branch/commit checked out using "git checkout" operation. You may
268 also specify `-` which is synonymous to `"@{-1}`.
269 +
270 As a special case, you may use `"A...B"` as a shortcut for the
271 merge base of `A` and `B` if there is exactly one merge base. You can
272 leave out at most one of `A` and `B`, in which case it defaults to `HEAD`.
273
274 <new_branch>::
275 Name for the new branch.
276
277 <start_point>::
278 The name of a commit at which to start the new branch; see
279 linkgit:git-branch[1] for details. Defaults to HEAD.
280
281 <tree-ish>::
282 Tree to checkout from (when paths are given). If not specified,
283 the index will be used.
284
285
286
287 DETACHED HEAD
288 -------------
289 HEAD normally refers to a named branch (e.g. 'master'). Meanwhile, each
290 branch refers to a specific commit. Let's look at a repo with three
291 commits, one of them tagged, and with branch 'master' checked out:
292
293 ------------
294 HEAD (refers to branch 'master')
295 |
296 v
297 a---b---c branch 'master' (refers to commit 'c')
298 ^
299 |
300 tag 'v2.0' (refers to commit 'b')
301 ------------
302
303 When a commit is created in this state, the branch is updated to refer to
304 the new commit. Specifically, 'git commit' creates a new commit 'd', whose
305 parent is commit 'c', and then updates branch 'master' to refer to new
306 commit 'd'. HEAD still refers to branch 'master' and so indirectly now refers
307 to commit 'd':
308
309 ------------
310 $ edit; git add; git commit
311
312 HEAD (refers to branch 'master')
313 |
314 v
315 a---b---c---d branch 'master' (refers to commit 'd')
316 ^
317 |
318 tag 'v2.0' (refers to commit 'b')
319 ------------
320
321 It is sometimes useful to be able to checkout a commit that is not at
322 the tip of any named branch, or even to create a new commit that is not
323 referenced by a named branch. Let's look at what happens when we
324 checkout commit 'b' (here we show two ways this may be done):
325
326 ------------
327 $ git checkout v2.0 # or
328 $ git checkout master^^
329
330 HEAD (refers to commit 'b')
331 |
332 v
333 a---b---c---d branch 'master' (refers to commit 'd')
334 ^
335 |
336 tag 'v2.0' (refers to commit 'b')
337 ------------
338
339 Notice that regardless of which checkout command we use, HEAD now refers
340 directly to commit 'b'. This is known as being in detached HEAD state.
341 It means simply that HEAD refers to a specific commit, as opposed to
342 referring to a named branch. Let's see what happens when we create a commit:
343
344 ------------
345 $ edit; git add; git commit
346
347 HEAD (refers to commit 'e')
348 |
349 v
350 e
351 /
352 a---b---c---d branch 'master' (refers to commit 'd')
353 ^
354 |
355 tag 'v2.0' (refers to commit 'b')
356 ------------
357
358 There is now a new commit 'e', but it is referenced only by HEAD. We can
359 of course add yet another commit in this state:
360
361 ------------
362 $ edit; git add; git commit
363
364 HEAD (refers to commit 'f')
365 |
366 v
367 e---f
368 /
369 a---b---c---d branch 'master' (refers to commit 'd')
370 ^
371 |
372 tag 'v2.0' (refers to commit 'b')
373 ------------
374
375 In fact, we can perform all the normal Git operations. But, let's look
376 at what happens when we then checkout master:
377
378 ------------
379 $ git checkout master
380
381 HEAD (refers to branch 'master')
382 e---f |
383 / v
384 a---b---c---d branch 'master' (refers to commit 'd')
385 ^
386 |
387 tag 'v2.0' (refers to commit 'b')
388 ------------
389
390 It is important to realize that at this point nothing refers to commit
391 'f'. Eventually commit 'f' (and by extension commit 'e') will be deleted
392 by the routine Git garbage collection process, unless we create a reference
393 before that happens. If we have not yet moved away from commit 'f',
394 any of these will create a reference to it:
395
396 ------------
397 $ git checkout -b foo <1>
398 $ git branch foo <2>
399 $ git tag foo <3>
400 ------------
401
402 <1> creates a new branch 'foo', which refers to commit 'f', and then
403 updates HEAD to refer to branch 'foo'. In other words, we'll no longer
404 be in detached HEAD state after this command.
405
406 <2> similarly creates a new branch 'foo', which refers to commit 'f',
407 but leaves HEAD detached.
408
409 <3> creates a new tag 'foo', which refers to commit 'f',
410 leaving HEAD detached.
411
412 If we have moved away from commit 'f', then we must first recover its object
413 name (typically by using git reflog), and then we can create a reference to
414 it. For example, to see the last two commits to which HEAD referred, we
415 can use either of these commands:
416
417 ------------
418 $ git reflog -2 HEAD # or
419 $ git log -g -2 HEAD
420 ------------
421
422 ARGUMENT DISAMBIGUATION
423 -----------------------
424
425 When there is only one argument given and it is not `--` (e.g. "git
426 checkout abc"), and when the argument is both a valid `<tree-ish>`
427 (e.g. a branch "abc" exists) and a valid `<pathspec>` (e.g. a file
428 or a directory whose name is "abc" exists), Git would usually ask
429 you to disambiguate. Because checking out a branch is so common an
430 operation, however, "git checkout abc" takes "abc" as a `<tree-ish>`
431 in such a situation. Use `git checkout -- <pathspec>` if you want
432 to checkout these paths out of the index.
433
434 EXAMPLES
435 --------
436
437 . The following sequence checks out the `master` branch, reverts
438 the `Makefile` to two revisions back, deletes hello.c by
439 mistake, and gets it back from the index.
440 +
441 ------------
442 $ git checkout master <1>
443 $ git checkout master~2 Makefile <2>
444 $ rm -f hello.c
445 $ git checkout hello.c <3>
446 ------------
447 +
448 <1> switch branch
449 <2> take a file out of another commit
450 <3> restore hello.c from the index
451 +
452 If you want to check out _all_ C source files out of the index,
453 you can say
454 +
455 ------------
456 $ git checkout -- '*.c'
457 ------------
458 +
459 Note the quotes around `*.c`. The file `hello.c` will also be
460 checked out, even though it is no longer in the working tree,
461 because the file globbing is used to match entries in the index
462 (not in the working tree by the shell).
463 +
464 If you have an unfortunate branch that is named `hello.c`, this
465 step would be confused as an instruction to switch to that branch.
466 You should instead write:
467 +
468 ------------
469 $ git checkout -- hello.c
470 ------------
471
472 . After working in the wrong branch, switching to the correct
473 branch would be done using:
474 +
475 ------------
476 $ git checkout mytopic
477 ------------
478 +
479 However, your "wrong" branch and correct "mytopic" branch may
480 differ in files that you have modified locally, in which case
481 the above checkout would fail like this:
482 +
483 ------------
484 $ git checkout mytopic
485 error: You have local changes to 'frotz'; not switching branches.
486 ------------
487 +
488 You can give the `-m` flag to the command, which would try a
489 three-way merge:
490 +
491 ------------
492 $ git checkout -m mytopic
493 Auto-merging frotz
494 ------------
495 +
496 After this three-way merge, the local modifications are _not_
497 registered in your index file, so `git diff` would show you what
498 changes you made since the tip of the new branch.
499
500 . When a merge conflict happens during switching branches with
501 the `-m` option, you would see something like this:
502 +
503 ------------
504 $ git checkout -m mytopic
505 Auto-merging frotz
506 ERROR: Merge conflict in frotz
507 fatal: merge program failed
508 ------------
509 +
510 At this point, `git diff` shows the changes cleanly merged as in
511 the previous example, as well as the changes in the conflicted
512 files. Edit and resolve the conflict and mark it resolved with
513 `git add` as usual:
514 +
515 ------------
516 $ edit frotz
517 $ git add frotz
518 ------------
519
520 GIT
521 ---
522 Part of the linkgit:git[1] suite