Merge branch 'qq/maint'
[git/git.git] / Documentation / user-manual.txt
CommitLineData
0eb4f7cd 1Git User's Manual (for version 1.5.3 or newer)
71f4b183 2______________________________________________
d19fbc3c 3
99eaefdd
BF
4
5Git is a fast distributed revision control system.
6
02783075 7This manual is designed to be readable by someone with basic UNIX
79c96c57 8command-line skills, but no previous knowledge of git.
d19fbc3c 9
2624d9a5
BF
10<<repositories-and-branches>> and <<exploring-git-history>> explain how
11to fetch and study a project using git--read these chapters to learn how
12to build and test a particular version of a software project, search for
13regressions, and so on.
ef89f701 14
2624d9a5
BF
15People needing to do actual development will also want to read
16<<Developing-with-git>> and <<sharing-development>>.
6bd9b682
BF
17
18Further chapters cover more specialized topics.
19
d19fbc3c 20Comprehensive reference documentation is available through the man
3861cd55 21pages. For a command such as "git clone <repo>", just use
d19fbc3c
BF
22
23------------------------------------------------
24$ man git-clone
25------------------------------------------------
26
2624d9a5
BF
27See also <<git-quick-start>> for a brief overview of git commands,
28without any explanation.
b181d57f 29
99f171bb 30Finally, see <<todo>> for ways that you can help make this manual more
2624d9a5 31complete.
b181d57f 32
b181d57f 33
e34caace 34[[repositories-and-branches]]
d19fbc3c
BF
35Repositories and Branches
36=========================
37
e34caace 38[[how-to-get-a-git-repository]]
d19fbc3c
BF
39How to get a git repository
40---------------------------
41
42It will be useful to have a git repository to experiment with as you
43read this manual.
44
5162e697 45The best way to get one is by using the linkgit:git-clone[1] command to
a5f90f31
BF
46download a copy of an existing repository. If you don't already have a
47project in mind, here are some interesting examples:
d19fbc3c
BF
48
49------------------------------------------------
50 # git itself (approx. 10MB download):
51$ git clone git://git.kernel.org/pub/scm/git/git.git
52 # the linux kernel (approx. 150MB download):
53$ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
54------------------------------------------------
55
56The initial clone may be time-consuming for a large project, but you
57will only need to clone once.
58
0c4a33b5
BF
59The clone command creates a new directory named after the project ("git"
60or "linux-2.6" in the examples above). After you cd into this
d19fbc3c 61directory, you will see that it contains a copy of the project files,
0c4a33b5
BF
62called the <<def_working_tree,working tree>>, together with a special
63top-level directory named ".git", which contains all the information
64about the history of the project.
d19fbc3c 65
e34caace 66[[how-to-check-out]]
d19fbc3c
BF
67How to check out a different version of a project
68-------------------------------------------------
69
a2ef9d63
BF
70Git is best thought of as a tool for storing the history of a collection
71of files. It stores the history as a compressed collection of
72interrelated snapshots of the project's contents. In git each such
73version is called a <<def_commit,commit>>.
d19fbc3c 74
0c4a33b5
BF
75Those snapshots aren't necessarily all arranged in a single line from
76oldest to newest; instead, work may simultaneously proceed along
57283291 77parallel lines of development, called <<def_branch,branches>>, which may
0c4a33b5
BF
78merge and diverge.
79
80A single git repository can track development on multiple branches. It
81does this by keeping a list of <<def_head,heads>> which reference the
5162e697 82latest commit on each branch; the linkgit:git-branch[1] command shows
81b6c950 83you the list of branch heads:
d19fbc3c
BF
84
85------------------------------------------------
86$ git branch
87* master
88------------------------------------------------
89
4f752407
BF
90A freshly cloned repository contains a single branch head, by default
91named "master", with the working directory initialized to the state of
92the project referred to by that branch head.
d19fbc3c 93
81b6c950
BF
94Most projects also use <<def_tag,tags>>. Tags, like heads, are
95references into the project's history, and can be listed using the
5162e697 96linkgit:git-tag[1] command:
d19fbc3c
BF
97
98------------------------------------------------
99$ git tag -l
100v2.6.11
101v2.6.11-tree
102v2.6.12
103v2.6.12-rc2
104v2.6.12-rc3
105v2.6.12-rc4
106v2.6.12-rc5
107v2.6.12-rc6
108v2.6.13
109...
110------------------------------------------------
111
fe4b3e59 112Tags are expected to always point at the same version of a project,
81b6c950 113while heads are expected to advance as development progresses.
fe4b3e59 114
81b6c950 115Create a new branch head pointing to one of these versions and check it
5162e697 116out using linkgit:git-checkout[1]:
d19fbc3c
BF
117
118------------------------------------------------
119$ git checkout -b new v2.6.13
120------------------------------------------------
121
122The working directory then reflects the contents that the project had
5162e697 123when it was tagged v2.6.13, and linkgit:git-branch[1] shows two
d19fbc3c
BF
124branches, with an asterisk marking the currently checked-out branch:
125
126------------------------------------------------
127$ git branch
128 master
129* new
130------------------------------------------------
131
132If you decide that you'd rather see version 2.6.17, you can modify
133the current branch to point at v2.6.17 instead, with
134
135------------------------------------------------
136$ git reset --hard v2.6.17
137------------------------------------------------
138
81b6c950 139Note that if the current branch head was your only reference to a
d19fbc3c 140particular point in history, then resetting that branch may leave you
81b6c950
BF
141with no way to find the history it used to point to; so use this command
142carefully.
d19fbc3c 143
e34caace 144[[understanding-commits]]
d19fbc3c
BF
145Understanding History: Commits
146------------------------------
147
148Every change in the history of a project is represented by a commit.
5162e697 149The linkgit:git-show[1] command shows the most recent commit on the
d19fbc3c
BF
150current branch:
151
152------------------------------------------------
153$ git show
e2618ff4
BF
154commit 17cf781661e6d38f737f15f53ab552f1e95960d7
155Author: Linus Torvalds <torvalds@ppc970.osdl.org.(none)>
156Date: Tue Apr 19 14:11:06 2005 -0700
157
158 Remove duplicate getenv(DB_ENVIRONMENT) call
159
160 Noted by Tony Luck.
161
162diff --git a/init-db.c b/init-db.c
163index 65898fa..b002dc6 100644
164--- a/init-db.c
165+++ b/init-db.c
166@@ -7,7 +7,7 @@
d19fbc3c 167
e2618ff4
BF
168 int main(int argc, char **argv)
169 {
170- char *sha1_dir = getenv(DB_ENVIRONMENT), *path;
171+ char *sha1_dir, *path;
172 int len, i;
173
174 if (mkdir(".git", 0755) < 0) {
d19fbc3c
BF
175------------------------------------------------
176
177As you can see, a commit shows who made the latest change, what they
178did, and why.
179
35121930 180Every commit has a 40-hexdigit id, sometimes called the "object name" or the
b1889c36 181"SHA1 id", shown on the first line of the "git-show" output. You can usually
35121930
BF
182refer to a commit by a shorter name, such as a tag or a branch name, but this
183longer name can also be useful. Most importantly, it is a globally unique
184name for this commit: so if you tell somebody else the object name (for
185example in email), then you are guaranteed that name will refer to the same
186commit in their repository that it does in yours (assuming their repository
187has that commit at all). Since the object name is computed as a hash over the
188contents of the commit, you are guaranteed that the commit can never change
189without its name also changing.
190
036f8199 191In fact, in <<git-concepts>> we shall see that everything stored in git
35121930
BF
192history, including file data and directory contents, is stored in an object
193with a name that is a hash of its contents.
d19fbc3c 194
e34caace 195[[understanding-reachability]]
d19fbc3c
BF
196Understanding history: commits, parents, and reachability
197~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
198
199Every commit (except the very first commit in a project) also has a
200parent commit which shows what happened before this commit.
201Following the chain of parents will eventually take you back to the
202beginning of the project.
203
204However, the commits do not form a simple list; git allows lines of
205development to diverge and then reconverge, and the point where two
206lines of development reconverge is called a "merge". The commit
207representing a merge can therefore have more than one parent, with
208each parent representing the most recent commit on one of the lines
209of development leading to that point.
210
5162e697 211The best way to see how this works is using the linkgit:gitk[1]
d19fbc3c
BF
212command; running gitk now on a git repository and looking for merge
213commits will help understand how the git organizes history.
214
215In the following, we say that commit X is "reachable" from commit Y
216if commit X is an ancestor of commit Y. Equivalently, you could say
02783075 217that Y is a descendant of X, or that there is a chain of parents
d19fbc3c
BF
218leading from commit Y to commit X.
219
e34caace 220[[history-diagrams]]
3dff5379
PR
221Understanding history: History diagrams
222~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c
BF
223
224We will sometimes represent git history using diagrams like the one
225below. Commits are shown as "o", and the links between them with
226lines drawn with - / and \. Time goes left to right:
227
1dc71a91
BF
228
229................................................
d19fbc3c
BF
230 o--o--o <-- Branch A
231 /
232 o--o--o <-- master
233 \
234 o--o--o <-- Branch B
1dc71a91 235................................................
d19fbc3c
BF
236
237If we need to talk about a particular commit, the character "o" may
238be replaced with another letter or number.
239
e34caace 240[[what-is-a-branch]]
d19fbc3c
BF
241Understanding history: What is a branch?
242~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
243
81b6c950
BF
244When we need to be precise, we will use the word "branch" to mean a line
245of development, and "branch head" (or just "head") to mean a reference
246to the most recent commit on a branch. In the example above, the branch
247head named "A" is a pointer to one particular commit, but we refer to
248the line of three commits leading up to that point as all being part of
d19fbc3c
BF
249"branch A".
250
81b6c950
BF
251However, when no confusion will result, we often just use the term
252"branch" both for branches and for branch heads.
d19fbc3c 253
e34caace 254[[manipulating-branches]]
d19fbc3c
BF
255Manipulating branches
256---------------------
257
258Creating, deleting, and modifying branches is quick and easy; here's
259a summary of the commands:
260
261git branch::
262 list all branches
263git branch <branch>::
264 create a new branch named <branch>, referencing the same
265 point in history as the current branch
266git branch <branch> <start-point>::
267 create a new branch named <branch>, referencing
268 <start-point>, which may be specified any way you like,
269 including using a branch name or a tag name
270git branch -d <branch>::
271 delete the branch <branch>; if the branch you are deleting
c64415e2
BF
272 points to a commit which is not reachable from the current
273 branch, this command will fail with a warning.
d19fbc3c
BF
274git branch -D <branch>::
275 even if the branch points to a commit not reachable
276 from the current branch, you may know that that commit
277 is still reachable from some other branch or tag. In that
278 case it is safe to use this command to force git to delete
279 the branch.
280git checkout <branch>::
281 make the current branch <branch>, updating the working
282 directory to reflect the version referenced by <branch>
283git checkout -b <new> <start-point>::
284 create a new branch <new> referencing <start-point>, and
285 check it out.
286
72a76c95
BF
287The special symbol "HEAD" can always be used to refer to the current
288branch. In fact, git uses a file named "HEAD" in the .git directory to
289remember which branch is current:
290
291------------------------------------------------
292$ cat .git/HEAD
293ref: refs/heads/master
294------------------------------------------------
295
25d9f3fa 296[[detached-head]]
72a76c95
BF
297Examining an old version without creating a new branch
298------------------------------------------------------
299
300The git-checkout command normally expects a branch head, but will also
301accept an arbitrary commit; for example, you can check out the commit
302referenced by a tag:
303
304------------------------------------------------
305$ git checkout v2.6.17
306Note: moving to "v2.6.17" which isn't a local branch
307If you want to create a new branch from this checkout, you may do so
308(now or later) by using -b with the checkout command again. Example:
309 git checkout -b <new_branch_name>
310HEAD is now at 427abfa... Linux v2.6.17
311------------------------------------------------
312
313The HEAD then refers to the SHA1 of the commit instead of to a branch,
314and git branch shows that you are no longer on a branch:
315
316------------------------------------------------
317$ cat .git/HEAD
318427abfa28afedffadfca9dd8b067eb6d36bac53f
953f3d6f 319$ git branch
72a76c95
BF
320* (no branch)
321 master
322------------------------------------------------
323
324In this case we say that the HEAD is "detached".
325
953f3d6f
BF
326This is an easy way to check out a particular version without having to
327make up a name for the new branch. You can still create a new branch
328(or tag) for this version later if you decide to.
d19fbc3c 329
e34caace 330[[examining-remote-branches]]
d19fbc3c
BF
331Examining branches from a remote repository
332-------------------------------------------
333
334The "master" branch that was created at the time you cloned is a copy
335of the HEAD in the repository that you cloned from. That repository
336may also have had other branches, though, and your local repository
337keeps branches which track each of those remote branches, which you
5162e697 338can view using the "-r" option to linkgit:git-branch[1]:
d19fbc3c
BF
339
340------------------------------------------------
341$ git branch -r
342 origin/HEAD
343 origin/html
344 origin/maint
345 origin/man
346 origin/master
347 origin/next
348 origin/pu
349 origin/todo
350------------------------------------------------
351
352You cannot check out these remote-tracking branches, but you can
353examine them on a branch of your own, just as you would a tag:
354
355------------------------------------------------
356$ git checkout -b my-todo-copy origin/todo
357------------------------------------------------
358
359Note that the name "origin" is just the name that git uses by default
360to refer to the repository that you cloned from.
361
362[[how-git-stores-references]]
f60b9642
BF
363Naming branches, tags, and other references
364-------------------------------------------
d19fbc3c
BF
365
366Branches, remote-tracking branches, and tags are all references to
f60b9642
BF
367commits. All references are named with a slash-separated path name
368starting with "refs"; the names we've been using so far are actually
369shorthand:
d19fbc3c 370
f60b9642
BF
371 - The branch "test" is short for "refs/heads/test".
372 - The tag "v2.6.18" is short for "refs/tags/v2.6.18".
373 - "origin/master" is short for "refs/remotes/origin/master".
d19fbc3c 374
f60b9642
BF
375The full name is occasionally useful if, for example, there ever
376exists a tag and a branch with the same name.
d19fbc3c 377
fc74ecc1
BF
378(Newly created refs are actually stored in the .git/refs directory,
379under the path given by their name. However, for efficiency reasons
380they may also be packed together in a single file; see
5162e697 381linkgit:git-pack-refs[1]).
fc74ecc1 382
c64415e2
BF
383As another useful shortcut, the "HEAD" of a repository can be referred
384to just using the name of that repository. So, for example, "origin"
385is usually a shortcut for the HEAD branch in the repository "origin".
d19fbc3c
BF
386
387For the complete list of paths which git checks for references, and
f60b9642
BF
388the order it uses to decide which to choose when there are multiple
389references with the same shorthand name, see the "SPECIFYING
5162e697 390REVISIONS" section of linkgit:git-rev-parse[1].
d19fbc3c
BF
391
392[[Updating-a-repository-with-git-fetch]]
b1889c36 393Updating a repository with git-fetch
d19fbc3c
BF
394------------------------------------
395
396Eventually the developer cloned from will do additional work in her
397repository, creating new commits and advancing the branches to point
398at the new commits.
399
400The command "git fetch", with no arguments, will update all of the
401remote-tracking branches to the latest version found in her
402repository. It will not touch any of your own branches--not even the
403"master" branch that was created for you on clone.
404
e34caace 405[[fetching-branches]]
d5cd5de4
BF
406Fetching branches from other repositories
407-----------------------------------------
408
409You can also track branches from repositories other than the one you
5162e697 410cloned from, using linkgit:git-remote[1]:
d5cd5de4
BF
411
412-------------------------------------------------
413$ git remote add linux-nfs git://linux-nfs.org/pub/nfs-2.6.git
04483524 414$ git fetch linux-nfs
d5cd5de4
BF
415* refs/remotes/linux-nfs/master: storing branch 'master' ...
416 commit: bf81b46
417-------------------------------------------------
418
419New remote-tracking branches will be stored under the shorthand name
b1889c36 420that you gave "git-remote add", in this case linux-nfs:
d5cd5de4
BF
421
422-------------------------------------------------
423$ git branch -r
424linux-nfs/master
425origin/master
426-------------------------------------------------
427
428If you run "git fetch <remote>" later, the tracking branches for the
429named <remote> will be updated.
430
431If you examine the file .git/config, you will see that git has added
432a new stanza:
433
434-------------------------------------------------
435$ cat .git/config
436...
437[remote "linux-nfs"]
923642fe
BF
438 url = git://linux-nfs.org/pub/nfs-2.6.git
439 fetch = +refs/heads/*:refs/remotes/linux-nfs/*
d5cd5de4
BF
440...
441-------------------------------------------------
442
fc90c536
BF
443This is what causes git to track the remote's branches; you may modify
444or delete these configuration options by editing .git/config with a
445text editor. (See the "CONFIGURATION FILE" section of
5162e697 446linkgit:git-config[1] for details.)
d5cd5de4 447
e34caace 448[[exploring-git-history]]
d19fbc3c
BF
449Exploring git history
450=====================
451
452Git is best thought of as a tool for storing the history of a
453collection of files. It does this by storing compressed snapshots of
1130845b 454the contents of a file hierarchy, together with "commits" which show
d19fbc3c
BF
455the relationships between these snapshots.
456
457Git provides extremely flexible and fast tools for exploring the
458history of a project.
459
aacd404e 460We start with one specialized tool that is useful for finding the
d19fbc3c
BF
461commit that introduced a bug into a project.
462
e34caace 463[[using-bisect]]
d19fbc3c
BF
464How to use bisect to find a regression
465--------------------------------------
466
467Suppose version 2.6.18 of your project worked, but the version at
468"master" crashes. Sometimes the best way to find the cause of such a
469regression is to perform a brute-force search through the project's
470history to find the particular commit that caused the problem. The
5162e697 471linkgit:git-bisect[1] command can help you do this:
d19fbc3c
BF
472
473-------------------------------------------------
474$ git bisect start
475$ git bisect good v2.6.18
476$ git bisect bad master
477Bisecting: 3537 revisions left to test after this
478[65934a9a028b88e83e2b0f8b36618fe503349f8e] BLOCK: Make USB storage depend on SCSI rather than selecting it [try #6]
479-------------------------------------------------
480
481If you run "git branch" at this point, you'll see that git has
482temporarily moved you to a new branch named "bisect". This branch
483points to a commit (with commit id 65934...) that is reachable from
2f29dac5 484"master" but not from v2.6.18. Compile and test it, and see whether
d19fbc3c
BF
485it crashes. Assume it does crash. Then:
486
487-------------------------------------------------
488$ git bisect bad
489Bisecting: 1769 revisions left to test after this
490[7eff82c8b1511017ae605f0c99ac275a7e21b867] i2c-core: Drop useless bitmaskings
491-------------------------------------------------
492
493checks out an older version. Continue like this, telling git at each
494stage whether the version it gives you is good or bad, and notice
495that the number of revisions left to test is cut approximately in
496half each time.
497
498After about 13 tests (in this case), it will output the commit id of
499the guilty commit. You can then examine the commit with
5162e697 500linkgit:git-show[1], find out who wrote it, and mail them your bug
d19fbc3c
BF
501report with the commit id. Finally, run
502
503-------------------------------------------------
504$ git bisect reset
505-------------------------------------------------
506
507to return you to the branch you were on before and delete the
508temporary "bisect" branch.
509
510Note that the version which git-bisect checks out for you at each
511point is just a suggestion, and you're free to try a different
512version if you think it would be a good idea. For example,
513occasionally you may land on a commit that broke something unrelated;
514run
515
516-------------------------------------------------
04483524 517$ git bisect visualize
d19fbc3c
BF
518-------------------------------------------------
519
520which will run gitk and label the commit it chose with a marker that
521says "bisect". Chose a safe-looking commit nearby, note its commit
522id, and check it out with:
523
524-------------------------------------------------
525$ git reset --hard fb47ddb2db...
526-------------------------------------------------
527
528then test, run "bisect good" or "bisect bad" as appropriate, and
529continue.
530
e34caace 531[[naming-commits]]
d19fbc3c
BF
532Naming commits
533--------------
534
535We have seen several ways of naming commits already:
536
d55ae921 537 - 40-hexdigit object name
d19fbc3c
BF
538 - branch name: refers to the commit at the head of the given
539 branch
540 - tag name: refers to the commit pointed to by the given tag
541 (we've seen branches and tags are special cases of
542 <<how-git-stores-references,references>>).
543 - HEAD: refers to the head of the current branch
544
eb6ae7f4 545There are many more; see the "SPECIFYING REVISIONS" section of the
5162e697 546linkgit:git-rev-parse[1] man page for the complete list of ways to
d19fbc3c
BF
547name revisions. Some examples:
548
549-------------------------------------------------
d55ae921 550$ git show fb47ddb2 # the first few characters of the object name
d19fbc3c
BF
551 # are usually enough to specify it uniquely
552$ git show HEAD^ # the parent of the HEAD commit
553$ git show HEAD^^ # the grandparent
554$ git show HEAD~4 # the great-great-grandparent
555-------------------------------------------------
556
557Recall that merge commits may have more than one parent; by default,
558^ and ~ follow the first parent listed in the commit, but you can
559also choose:
560
561-------------------------------------------------
562$ git show HEAD^1 # show the first parent of HEAD
563$ git show HEAD^2 # show the second parent of HEAD
564-------------------------------------------------
565
566In addition to HEAD, there are several other special names for
567commits:
568
569Merges (to be discussed later), as well as operations such as
570git-reset, which change the currently checked-out commit, generally
571set ORIG_HEAD to the value HEAD had before the current operation.
572
573The git-fetch operation always stores the head of the last fetched
574branch in FETCH_HEAD. For example, if you run git fetch without
575specifying a local branch as the target of the operation
576
577-------------------------------------------------
578$ git fetch git://example.com/proj.git theirbranch
579-------------------------------------------------
580
581the fetched commits will still be available from FETCH_HEAD.
582
583When we discuss merges we'll also see the special name MERGE_HEAD,
584which refers to the other branch that we're merging in to the current
585branch.
586
5162e697 587The linkgit:git-rev-parse[1] command is a low-level command that is
d55ae921
BF
588occasionally useful for translating some name for a commit to the object
589name for that commit:
aec053bb
BF
590
591-------------------------------------------------
592$ git rev-parse origin
593e05db0fd4f31dde7005f075a84f96b360d05984b
594-------------------------------------------------
595
e34caace 596[[creating-tags]]
d19fbc3c
BF
597Creating tags
598-------------
599
600We can also create a tag to refer to a particular commit; after
601running
602
603-------------------------------------------------
04483524 604$ git tag stable-1 1b2e1d63ff
d19fbc3c
BF
605-------------------------------------------------
606
607You can use stable-1 to refer to the commit 1b2e1d63ff.
608
c64415e2
BF
609This creates a "lightweight" tag. If you would also like to include a
610comment with the tag, and possibly sign it cryptographically, then you
5162e697 611should create a tag object instead; see the linkgit:git-tag[1] man page
c64415e2 612for details.
d19fbc3c 613
e34caace 614[[browsing-revisions]]
d19fbc3c
BF
615Browsing revisions
616------------------
617
5162e697 618The linkgit:git-log[1] command can show lists of commits. On its
d19fbc3c
BF
619own, it shows all commits reachable from the parent commit; but you
620can also make more specific requests:
621
622-------------------------------------------------
623$ git log v2.5.. # commits since (not reachable from) v2.5
624$ git log test..master # commits reachable from master but not test
625$ git log master..test # ...reachable from test but not master
626$ git log master...test # ...reachable from either test or master,
627 # but not both
628$ git log --since="2 weeks ago" # commits from the last 2 weeks
629$ git log Makefile # commits which modify Makefile
630$ git log fs/ # ... which modify any file under fs/
631$ git log -S'foo()' # commits which add or remove any file data
632 # matching the string 'foo()'
633-------------------------------------------------
634
635And of course you can combine all of these; the following finds
636commits since v2.5 which touch the Makefile or any file under fs:
637
638-------------------------------------------------
639$ git log v2.5.. Makefile fs/
640-------------------------------------------------
641
642You can also ask git log to show patches:
643
644-------------------------------------------------
645$ git log -p
646-------------------------------------------------
647
5162e697 648See the "--pretty" option in the linkgit:git-log[1] man page for more
d19fbc3c
BF
649display options.
650
651Note that git log starts with the most recent commit and works
652backwards through the parents; however, since git history can contain
3dff5379 653multiple independent lines of development, the particular order that
d19fbc3c
BF
654commits are listed in may be somewhat arbitrary.
655
e34caace 656[[generating-diffs]]
d19fbc3c
BF
657Generating diffs
658----------------
659
660You can generate diffs between any two versions using
5162e697 661linkgit:git-diff[1]:
d19fbc3c
BF
662
663-------------------------------------------------
664$ git diff master..test
665-------------------------------------------------
666
5b98d9bc
BF
667That will produce the diff between the tips of the two branches. If
668you'd prefer to find the diff from their common ancestor to test, you
669can use three dots instead of two:
670
671-------------------------------------------------
672$ git diff master...test
673-------------------------------------------------
674
675Sometimes what you want instead is a set of patches; for this you can
5162e697 676use linkgit:git-format-patch[1]:
d19fbc3c
BF
677
678-------------------------------------------------
679$ git format-patch master..test
680-------------------------------------------------
681
682will generate a file with a patch for each commit reachable from test
5b98d9bc 683but not from master.
d19fbc3c 684
e34caace 685[[viewing-old-file-versions]]
d19fbc3c
BF
686Viewing old file versions
687-------------------------
688
689You can always view an old version of a file by just checking out the
690correct revision first. But sometimes it is more convenient to be
691able to view an old version of a single file without checking
692anything out; this command does that:
693
694-------------------------------------------------
695$ git show v2.5:fs/locks.c
696-------------------------------------------------
697
698Before the colon may be anything that names a commit, and after it
699may be any path to a file tracked by git.
700
e34caace 701[[history-examples]]
aec053bb
BF
702Examples
703--------
704
46acd3fa
BF
705[[counting-commits-on-a-branch]]
706Counting the number of commits on a branch
707~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
708
709Suppose you want to know how many commits you've made on "mybranch"
710since it diverged from "origin":
711
712-------------------------------------------------
713$ git log --pretty=oneline origin..mybranch | wc -l
714-------------------------------------------------
715
716Alternatively, you may often see this sort of thing done with the
5162e697 717lower-level command linkgit:git-rev-list[1], which just lists the SHA1's
46acd3fa
BF
718of all the given commits:
719
720-------------------------------------------------
721$ git rev-list origin..mybranch | wc -l
722-------------------------------------------------
723
e34caace 724[[checking-for-equal-branches]]
aec053bb 725Check whether two branches point at the same history
2f99710c 726~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
aec053bb
BF
727
728Suppose you want to check whether two branches point at the same point
729in history.
730
731-------------------------------------------------
732$ git diff origin..master
733-------------------------------------------------
734
69f7ad73
BF
735will tell you whether the contents of the project are the same at the
736two branches; in theory, however, it's possible that the same project
737contents could have been arrived at by two different historical
d55ae921 738routes. You could compare the object names:
aec053bb
BF
739
740-------------------------------------------------
741$ git rev-list origin
742e05db0fd4f31dde7005f075a84f96b360d05984b
743$ git rev-list master
744e05db0fd4f31dde7005f075a84f96b360d05984b
745-------------------------------------------------
746
69f7ad73
BF
747Or you could recall that the ... operator selects all commits
748contained reachable from either one reference or the other but not
749both: so
aec053bb
BF
750
751-------------------------------------------------
752$ git log origin...master
753-------------------------------------------------
754
755will return no commits when the two branches are equal.
756
e34caace 757[[finding-tagged-descendants]]
b181d57f
BF
758Find first tagged version including a given fix
759~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
aec053bb 760
69f7ad73
BF
761Suppose you know that the commit e05db0fd fixed a certain problem.
762You'd like to find the earliest tagged release that contains that
763fix.
764
765Of course, there may be more than one answer--if the history branched
766after commit e05db0fd, then there could be multiple "earliest" tagged
767releases.
768
769You could just visually inspect the commits since e05db0fd:
770
771-------------------------------------------------
772$ gitk e05db0fd..
773-------------------------------------------------
774
5162e697 775Or you can use linkgit:git-name-rev[1], which will give the commit a
b181d57f
BF
776name based on any tag it finds pointing to one of the commit's
777descendants:
778
779-------------------------------------------------
04483524 780$ git name-rev --tags e05db0fd
b181d57f
BF
781e05db0fd tags/v1.5.0-rc1^0~23
782-------------------------------------------------
783
5162e697 784The linkgit:git-describe[1] command does the opposite, naming the
b181d57f
BF
785revision using a tag on which the given commit is based:
786
787-------------------------------------------------
788$ git describe e05db0fd
04483524 789v1.5.0-rc0-260-ge05db0f
b181d57f
BF
790-------------------------------------------------
791
792but that may sometimes help you guess which tags might come after the
793given commit.
794
795If you just want to verify whether a given tagged version contains a
5162e697 796given commit, you could use linkgit:git-merge-base[1]:
b181d57f
BF
797
798-------------------------------------------------
799$ git merge-base e05db0fd v1.5.0-rc1
800e05db0fd4f31dde7005f075a84f96b360d05984b
801-------------------------------------------------
802
803The merge-base command finds a common ancestor of the given commits,
804and always returns one or the other in the case where one is a
805descendant of the other; so the above output shows that e05db0fd
806actually is an ancestor of v1.5.0-rc1.
807
808Alternatively, note that
809
810-------------------------------------------------
4a7979ca 811$ git log v1.5.0-rc1..e05db0fd
b181d57f
BF
812-------------------------------------------------
813
4a7979ca 814will produce empty output if and only if v1.5.0-rc1 includes e05db0fd,
b181d57f 815because it outputs only commits that are not reachable from v1.5.0-rc1.
aec053bb 816
5162e697 817As yet another alternative, the linkgit:git-show-branch[1] command lists
4a7979ca
BF
818the commits reachable from its arguments with a display on the left-hand
819side that indicates which arguments that commit is reachable from. So,
820you can run something like
821
822-------------------------------------------------
823$ git show-branch e05db0fd v1.5.0-rc0 v1.5.0-rc1 v1.5.0-rc2
824! [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
825available
826 ! [v1.5.0-rc0] GIT v1.5.0 preview
827 ! [v1.5.0-rc1] GIT v1.5.0-rc1
828 ! [v1.5.0-rc2] GIT v1.5.0-rc2
829...
830-------------------------------------------------
831
832then search for a line that looks like
833
834-------------------------------------------------
835+ ++ [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
836available
837-------------------------------------------------
838
839Which shows that e05db0fd is reachable from itself, from v1.5.0-rc1, and
840from v1.5.0-rc2, but not from v1.5.0-rc0.
841
629d9f78
BF
842[[showing-commits-unique-to-a-branch]]
843Showing commits unique to a given branch
844~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4a7979ca 845
629d9f78
BF
846Suppose you would like to see all the commits reachable from the branch
847head named "master" but not from any other head in your repository.
d19fbc3c 848
629d9f78 849We can list all the heads in this repository with
5162e697 850linkgit:git-show-ref[1]:
d19fbc3c 851
629d9f78
BF
852-------------------------------------------------
853$ git show-ref --heads
854bf62196b5e363d73353a9dcf094c59595f3153b7 refs/heads/core-tutorial
855db768d5504c1bb46f63ee9d6e1772bd047e05bf9 refs/heads/maint
856a07157ac624b2524a059a3414e99f6f44bebc1e7 refs/heads/master
85724dbc180ea14dc1aebe09f14c8ecf32010690627 refs/heads/tutorial-2
8581e87486ae06626c2f31eaa63d26fc0fd646c8af2 refs/heads/tutorial-fixes
859-------------------------------------------------
d19fbc3c 860
629d9f78
BF
861We can get just the branch-head names, and remove "master", with
862the help of the standard utilities cut and grep:
863
864-------------------------------------------------
865$ git show-ref --heads | cut -d' ' -f2 | grep -v '^refs/heads/master'
866refs/heads/core-tutorial
867refs/heads/maint
868refs/heads/tutorial-2
869refs/heads/tutorial-fixes
870-------------------------------------------------
871
872And then we can ask to see all the commits reachable from master
873but not from these other heads:
874
875-------------------------------------------------
876$ gitk master --not $( git show-ref --heads | cut -d' ' -f2 |
877 grep -v '^refs/heads/master' )
878-------------------------------------------------
879
880Obviously, endless variations are possible; for example, to see all
881commits reachable from some head but not from any tag in the repository:
882
883-------------------------------------------------
c78974f7 884$ gitk $( git show-ref --heads ) --not $( git show-ref --tags )
629d9f78
BF
885-------------------------------------------------
886
5162e697 887(See linkgit:git-rev-parse[1] for explanations of commit-selecting
629d9f78
BF
888syntax such as `--not`.)
889
82c8bf28
BF
890[[making-a-release]]
891Creating a changelog and tarball for a software release
892~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
893
5162e697 894The linkgit:git-archive[1] command can create a tar or zip archive from
82c8bf28
BF
895any version of a project; for example:
896
897-------------------------------------------------
898$ git archive --format=tar --prefix=project/ HEAD | gzip >latest.tar.gz
899-------------------------------------------------
900
901will use HEAD to produce a tar archive in which each filename is
ccd71866 902preceded by "project/".
82c8bf28
BF
903
904If you're releasing a new version of a software project, you may want
905to simultaneously make a changelog to include in the release
906announcement.
907
908Linus Torvalds, for example, makes new kernel releases by tagging them,
909then running:
910
911-------------------------------------------------
912$ release-script 2.6.12 2.6.13-rc6 2.6.13-rc7
913-------------------------------------------------
914
915where release-script is a shell script that looks like:
916
917-------------------------------------------------
918#!/bin/sh
919stable="$1"
920last="$2"
921new="$3"
922echo "# git tag v$new"
923echo "git archive --prefix=linux-$new/ v$new | gzip -9 > ../linux-$new.tar.gz"
924echo "git diff v$stable v$new | gzip -9 > ../patch-$new.gz"
925echo "git log --no-merges v$new ^v$last > ../ChangeLog-$new"
926echo "git shortlog --no-merges v$new ^v$last > ../ShortLog"
927echo "git diff --stat --summary -M v$last v$new > ../diffstat-$new"
928-------------------------------------------------
929
930and then he just cut-and-pastes the output commands after verifying that
931they look OK.
4a7979ca 932
8ceca74a 933[[Finding-comments-with-given-content]]
187b0d80 934Finding commits referencing a file with given content
d5821de2 935~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
187b0d80
BF
936
937Somebody hands you a copy of a file, and asks which commits modified a
938file such that it contained the given content either before or after the
939commit. You can find out with this:
940
941-------------------------------------------------
477ff5b7 942$ git log --raw --abbrev=40 --pretty=oneline |
187b0d80
BF
943 grep -B 1 `git hash-object filename`
944-------------------------------------------------
945
946Figuring out why this works is left as an exercise to the (advanced)
5162e697
DM
947student. The linkgit:git-log[1], linkgit:git-diff-tree[1], and
948linkgit:git-hash-object[1] man pages may prove helpful.
187b0d80 949
e34caace 950[[Developing-with-git]]
d19fbc3c
BF
951Developing with git
952===================
953
e34caace 954[[telling-git-your-name]]
d19fbc3c
BF
955Telling git your name
956---------------------
957
958Before creating any commits, you should introduce yourself to git. The
58c19d1f
BF
959easiest way to do so is to make sure the following lines appear in a
960file named .gitconfig in your home directory:
d19fbc3c
BF
961
962------------------------------------------------
d19fbc3c
BF
963[user]
964 name = Your Name Comes Here
965 email = you@yourdomain.example.com
d19fbc3c
BF
966------------------------------------------------
967
5162e697 968(See the "CONFIGURATION FILE" section of linkgit:git-config[1] for
fc90c536
BF
969details on the configuration file.)
970
d19fbc3c 971
e34caace 972[[creating-a-new-repository]]
d19fbc3c
BF
973Creating a new repository
974-------------------------
975
976Creating a new repository from scratch is very easy:
977
978-------------------------------------------------
979$ mkdir project
980$ cd project
f1d2b477 981$ git init
d19fbc3c
BF
982-------------------------------------------------
983
984If you have some initial content (say, a tarball):
985
986-------------------------------------------------
987$ tar -xzvf project.tar.gz
988$ cd project
f1d2b477 989$ git init
d19fbc3c
BF
990$ git add . # include everything below ./ in the first commit:
991$ git commit
992-------------------------------------------------
993
994[[how-to-make-a-commit]]
ae25c67a 995How to make a commit
d19fbc3c
BF
996--------------------
997
998Creating a new commit takes three steps:
999
1000 1. Making some changes to the working directory using your
1001 favorite editor.
1002 2. Telling git about your changes.
1003 3. Creating the commit using the content you told git about
1004 in step 2.
1005
1006In practice, you can interleave and repeat steps 1 and 2 as many
1007times as you want: in order to keep track of what you want committed
1008at step 3, git maintains a snapshot of the tree's contents in a
1009special staging area called "the index."
1010
01997b4a
BF
1011At the beginning, the content of the index will be identical to
1012that of the HEAD. The command "git diff --cached", which shows
1013the difference between the HEAD and the index, should therefore
1014produce no output at that point.
eb6ae7f4 1015
d19fbc3c
BF
1016Modifying the index is easy:
1017
1018To update the index with the new contents of a modified file, use
1019
1020-------------------------------------------------
1021$ git add path/to/file
1022-------------------------------------------------
1023
1024To add the contents of a new file to the index, use
1025
1026-------------------------------------------------
1027$ git add path/to/file
1028-------------------------------------------------
1029
eb6ae7f4 1030To remove a file from the index and from the working tree,
d19fbc3c
BF
1031
1032-------------------------------------------------
1033$ git rm path/to/file
1034-------------------------------------------------
1035
1036After each step you can verify that
1037
1038-------------------------------------------------
1039$ git diff --cached
1040-------------------------------------------------
1041
1042always shows the difference between the HEAD and the index file--this
1043is what you'd commit if you created the commit now--and that
1044
1045-------------------------------------------------
1046$ git diff
1047-------------------------------------------------
1048
1049shows the difference between the working tree and the index file.
1050
b1889c36 1051Note that "git-add" always adds just the current contents of a file
d19fbc3c
BF
1052to the index; further changes to the same file will be ignored unless
1053you run git-add on the file again.
1054
1055When you're ready, just run
1056
1057-------------------------------------------------
1058$ git commit
1059-------------------------------------------------
1060
1061and git will prompt you for a commit message and then create the new
3dff5379 1062commit. Check to make sure it looks like what you expected with
d19fbc3c
BF
1063
1064-------------------------------------------------
1065$ git show
1066-------------------------------------------------
1067
1068As a special shortcut,
a6080a0a 1069
d19fbc3c
BF
1070-------------------------------------------------
1071$ git commit -a
1072-------------------------------------------------
1073
1074will update the index with any files that you've modified or removed
1075and create a commit, all in one step.
1076
1077A number of commands are useful for keeping track of what you're
1078about to commit:
1079
1080-------------------------------------------------
1081$ git diff --cached # difference between HEAD and the index; what
1130845b 1082 # would be committed if you ran "commit" now.
d19fbc3c
BF
1083$ git diff # difference between the index file and your
1084 # working directory; changes that would not
1085 # be included if you ran "commit" now.
c64415e2
BF
1086$ git diff HEAD # difference between HEAD and working tree; what
1087 # would be committed if you ran "commit -a" now.
d19fbc3c
BF
1088$ git status # a brief per-file summary of the above.
1089-------------------------------------------------
1090
5162e697 1091You can also use linkgit:git-gui[1] to create commits, view changes in
407c0c87
BF
1092the index and the working tree files, and individually select diff hunks
1093for inclusion in the index (by right-clicking on the diff hunk and
1094choosing "Stage Hunk For Commit").
1095
e34caace 1096[[creating-good-commit-messages]]
ae25c67a 1097Creating good commit messages
d19fbc3c
BF
1098-----------------------------
1099
1100Though not required, it's a good idea to begin the commit message
1101with a single short (less than 50 character) line summarizing the
1102change, followed by a blank line and then a more thorough
1103description. Tools that turn commits into email, for example, use
1104the first line on the Subject line and the rest of the commit in the
1105body.
1106
2dc53617
JH
1107[[ignoring-files]]
1108Ignoring files
1109--------------
1110
1111A project will often generate files that you do 'not' want to track with git.
1112This typically includes files generated by a build process or temporary
1113backup files made by your editor. Of course, 'not' tracking files with git
b1889c36 1114is just a matter of 'not' calling "`git-add`" on them. But it quickly becomes
2dc53617
JH
1115annoying to have these untracked files lying around; e.g. they make
1116"`git add .`" and "`git commit -a`" practically useless, and they keep
464a8a7a 1117showing up in the output of "`git status`".
2dc53617 1118
464a8a7a
BF
1119You can tell git to ignore certain files by creating a file called .gitignore
1120in the top level of your working directory, with contents such as:
2dc53617
JH
1121
1122-------------------------------------------------
1123# Lines starting with '#' are considered comments.
464a8a7a 1124# Ignore any file named foo.txt.
2dc53617
JH
1125foo.txt
1126# Ignore (generated) html files,
1127*.html
1128# except foo.html which is maintained by hand.
1129!foo.html
1130# Ignore objects and archives.
1131*.[oa]
1132-------------------------------------------------
1133
5162e697 1134See linkgit:gitignore[5] for a detailed explanation of the syntax. You can
464a8a7a
BF
1135also place .gitignore files in other directories in your working tree, and they
1136will apply to those directories and their subdirectories. The `.gitignore`
1137files can be added to your repository like any other files (just run `git add
1138.gitignore` and `git commit`, as usual), which is convenient when the exclude
1139patterns (such as patterns matching build output files) would also make sense
1140for other users who clone your repository.
1141
1142If you wish the exclude patterns to affect only certain repositories
1143(instead of every repository for a given project), you may instead put
1144them in a file in your repository named .git/info/exclude, or in any file
1145specified by the `core.excludesfile` configuration variable. Some git
1146commands can also take exclude patterns directly on the command line.
5162e697 1147See linkgit:gitignore[5] for the details.
2dc53617 1148
e34caace 1149[[how-to-merge]]
ae25c67a 1150How to merge
d19fbc3c
BF
1151------------
1152
1153You can rejoin two diverging branches of development using
5162e697 1154linkgit:git-merge[1]:
d19fbc3c
BF
1155
1156-------------------------------------------------
1157$ git merge branchname
1158-------------------------------------------------
1159
1160merges the development in the branch "branchname" into the current
1161branch. If there are conflicts--for example, if the same file is
1162modified in two different ways in the remote branch and the local
1163branch--then you are warned; the output may look something like this:
1164
1165-------------------------------------------------
fabbd8f6
BF
1166$ git merge next
1167 100% (4/4) done
1168Auto-merged file.txt
d19fbc3c
BF
1169CONFLICT (content): Merge conflict in file.txt
1170Automatic merge failed; fix conflicts and then commit the result.
1171-------------------------------------------------
1172
1173Conflict markers are left in the problematic files, and after
1174you resolve the conflicts manually, you can update the index
1175with the contents and run git commit, as you normally would when
1176creating a new file.
1177
1178If you examine the resulting commit using gitk, you will see that it
1179has two parents, one pointing to the top of the current branch, and
1180one to the top of the other branch.
1181
d19fbc3c
BF
1182[[resolving-a-merge]]
1183Resolving a merge
1184-----------------
1185
1186When a merge isn't resolved automatically, git leaves the index and
1187the working tree in a special state that gives you all the
1188information you need to help resolve the merge.
1189
1190Files with conflicts are marked specially in the index, so until you
5162e697 1191resolve the problem and update the index, linkgit:git-commit[1] will
ef561ac7 1192fail:
d19fbc3c
BF
1193
1194-------------------------------------------------
1195$ git commit
1196file.txt: needs merge
1197-------------------------------------------------
1198
5162e697 1199Also, linkgit:git-status[1] will list those files as "unmerged", and the
ef561ac7
BF
1200files with conflicts will have conflict markers added, like this:
1201
1202-------------------------------------------------
1203<<<<<<< HEAD:file.txt
1204Hello world
1205=======
1206Goodbye
1207>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1208-------------------------------------------------
1209
1210All you need to do is edit the files to resolve the conflicts, and then
1211
1212-------------------------------------------------
1213$ git add file.txt
1214$ git commit
1215-------------------------------------------------
1216
1217Note that the commit message will already be filled in for you with
1218some information about the merge. Normally you can just use this
1219default message unchanged, but you may add additional commentary of
1220your own if desired.
1221
1222The above is all you need to know to resolve a simple merge. But git
1223also provides more information to help resolve conflicts:
1224
e34caace 1225[[conflict-resolution]]
ef561ac7
BF
1226Getting conflict-resolution help during a merge
1227~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c
BF
1228
1229All of the changes that git was able to merge automatically are
5162e697 1230already added to the index file, so linkgit:git-diff[1] shows only
ef561ac7 1231the conflicts. It uses an unusual syntax:
d19fbc3c
BF
1232
1233-------------------------------------------------
1234$ git diff
1235diff --cc file.txt
1236index 802992c,2b60207..0000000
1237--- a/file.txt
1238+++ b/file.txt
1239@@@ -1,1 -1,1 +1,5 @@@
1240++<<<<<<< HEAD:file.txt
1241 +Hello world
1242++=======
1243+ Goodbye
1244++>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1245-------------------------------------------------
1246
1130845b 1247Recall that the commit which will be committed after we resolve this
d19fbc3c
BF
1248conflict will have two parents instead of the usual one: one parent
1249will be HEAD, the tip of the current branch; the other will be the
1250tip of the other branch, which is stored temporarily in MERGE_HEAD.
1251
ef561ac7
BF
1252During the merge, the index holds three versions of each file. Each of
1253these three "file stages" represents a different version of the file:
1254
1255-------------------------------------------------
1256$ git show :1:file.txt # the file in a common ancestor of both branches
4209752d
JH
1257$ git show :2:file.txt # the version from HEAD.
1258$ git show :3:file.txt # the version from MERGE_HEAD.
ef561ac7
BF
1259-------------------------------------------------
1260
4209752d
JH
1261When you ask linkgit:git-diff[1] to show the conflicts, it runs a
1262three-way diff between the conflicted merge results in the work tree with
1263stages 2 and 3 to show only hunks whose contents come from both sides,
1264mixed (in other words, when a hunk's merge results come only from stage 2,
1265that part is not conflicting and is not shown. Same for stage 3).
ef561ac7
BF
1266
1267The diff above shows the differences between the working-tree version of
1268file.txt and the stage 2 and stage 3 versions. So instead of preceding
1269each line by a single "+" or "-", it now uses two columns: the first
1270column is used for differences between the first parent and the working
1271directory copy, and the second for differences between the second parent
1272and the working directory copy. (See the "COMBINED DIFF FORMAT" section
5162e697 1273of linkgit:git-diff-files[1] for a details of the format.)
ef561ac7
BF
1274
1275After resolving the conflict in the obvious way (but before updating the
1276index), the diff will look like:
d19fbc3c
BF
1277
1278-------------------------------------------------
1279$ git diff
1280diff --cc file.txt
1281index 802992c,2b60207..0000000
1282--- a/file.txt
1283+++ b/file.txt
1284@@@ -1,1 -1,1 +1,1 @@@
1285- Hello world
1286 -Goodbye
1287++Goodbye world
1288-------------------------------------------------
1289
1290This shows that our resolved version deleted "Hello world" from the
1291first parent, deleted "Goodbye" from the second parent, and added
1292"Goodbye world", which was previously absent from both.
1293
ef561ac7
BF
1294Some special diff options allow diffing the working directory against
1295any of these stages:
1296
1297-------------------------------------------------
1298$ git diff -1 file.txt # diff against stage 1
1299$ git diff --base file.txt # same as the above
1300$ git diff -2 file.txt # diff against stage 2
1301$ git diff --ours file.txt # same as the above
1302$ git diff -3 file.txt # diff against stage 3
1303$ git diff --theirs file.txt # same as the above.
1304-------------------------------------------------
1305
0cafe944 1306The linkgit:git-log[1] and linkgit:gitk[1] commands also provide special help
ef561ac7 1307for merges:
d19fbc3c
BF
1308
1309-------------------------------------------------
1310$ git log --merge
ef561ac7 1311$ gitk --merge
d19fbc3c
BF
1312-------------------------------------------------
1313
ef561ac7
BF
1314These will display all commits which exist only on HEAD or on
1315MERGE_HEAD, and which touch an unmerged file.
d19fbc3c 1316
5162e697 1317You may also use linkgit:git-mergetool[1], which lets you merge the
c64415e2
BF
1318unmerged files using external tools such as emacs or kdiff3.
1319
ef561ac7 1320Each time you resolve the conflicts in a file and update the index:
d19fbc3c
BF
1321
1322-------------------------------------------------
1323$ git add file.txt
d19fbc3c
BF
1324-------------------------------------------------
1325
ef561ac7
BF
1326the different stages of that file will be "collapsed", after which
1327git-diff will (by default) no longer show diffs for that file.
d19fbc3c
BF
1328
1329[[undoing-a-merge]]
ae25c67a 1330Undoing a merge
d19fbc3c
BF
1331---------------
1332
1333If you get stuck and decide to just give up and throw the whole mess
1334away, you can always return to the pre-merge state with
1335
1336-------------------------------------------------
1337$ git reset --hard HEAD
1338-------------------------------------------------
1339
1130845b 1340Or, if you've already committed the merge that you want to throw away,
d19fbc3c
BF
1341
1342-------------------------------------------------
1c73bb0e 1343$ git reset --hard ORIG_HEAD
d19fbc3c
BF
1344-------------------------------------------------
1345
1346However, this last command can be dangerous in some cases--never
1347throw away a commit you have already committed if that commit may
1348itself have been merged into another branch, as doing so may confuse
1349further merges.
1350
e34caace 1351[[fast-forwards]]
d19fbc3c
BF
1352Fast-forward merges
1353-------------------
1354
1355There is one special case not mentioned above, which is treated
1356differently. Normally, a merge results in a merge commit, with two
1357parents, one pointing at each of the two lines of development that
1358were merged.
1359
59723040
BF
1360However, if the current branch is a descendant of the other--so every
1361commit present in the one is already contained in the other--then git
1362just performs a "fast forward"; the head of the current branch is moved
1363forward to point at the head of the merged-in branch, without any new
1364commits being created.
d19fbc3c 1365
e34caace 1366[[fixing-mistakes]]
b684f830
BF
1367Fixing mistakes
1368---------------
1369
1370If you've messed up the working tree, but haven't yet committed your
1371mistake, you can return the entire working tree to the last committed
1372state with
1373
1374-------------------------------------------------
1375$ git reset --hard HEAD
1376-------------------------------------------------
1377
1378If you make a commit that you later wish you hadn't, there are two
1379fundamentally different ways to fix the problem:
1380
1381 1. You can create a new commit that undoes whatever was done
93cbbd71 1382 by the old commit. This is the correct thing if your
b684f830
BF
1383 mistake has already been made public.
1384
1385 2. You can go back and modify the old commit. You should
1386 never do this if you have already made the history public;
1387 git does not normally expect the "history" of a project to
1388 change, and cannot correctly perform repeated merges from
1389 a branch that has had its history changed.
1390
e34caace 1391[[reverting-a-commit]]
b684f830
BF
1392Fixing a mistake with a new commit
1393~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1394
1395Creating a new commit that reverts an earlier change is very easy;
5162e697 1396just pass the linkgit:git-revert[1] command a reference to the bad
b684f830
BF
1397commit; for example, to revert the most recent commit:
1398
1399-------------------------------------------------
1400$ git revert HEAD
1401-------------------------------------------------
1402
1403This will create a new commit which undoes the change in HEAD. You
1404will be given a chance to edit the commit message for the new commit.
1405
1406You can also revert an earlier change, for example, the next-to-last:
1407
1408-------------------------------------------------
1409$ git revert HEAD^
1410-------------------------------------------------
1411
1412In this case git will attempt to undo the old change while leaving
1413intact any changes made since then. If more recent changes overlap
1414with the changes to be reverted, then you will be asked to fix
1415conflicts manually, just as in the case of <<resolving-a-merge,
1416resolving a merge>>.
1417
7cb192ea
BF
1418[[fixing-a-mistake-by-rewriting-history]]
1419Fixing a mistake by rewriting history
b684f830
BF
1420~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1421
1422If the problematic commit is the most recent commit, and you have not
1423yet made that commit public, then you may just
1424<<undoing-a-merge,destroy it using git-reset>>.
1425
1426Alternatively, you
1427can edit the working directory and update the index to fix your
1428mistake, just as if you were going to <<how-to-make-a-commit,create a
1429new commit>>, then run
1430
1431-------------------------------------------------
1432$ git commit --amend
1433-------------------------------------------------
1434
1435which will replace the old commit by a new commit incorporating your
1436changes, giving you a chance to edit the old commit message first.
1437
1438Again, you should never do this to a commit that may already have
5162e697 1439been merged into another branch; use linkgit:git-revert[1] instead in
b684f830
BF
1440that case.
1441
7cb192ea 1442It is also possible to replace commits further back in the history, but
b684f830
BF
1443this is an advanced topic to be left for
1444<<cleaning-up-history,another chapter>>.
1445
e34caace 1446[[checkout-of-path]]
b684f830
BF
1447Checking out an old version of a file
1448~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1449
1450In the process of undoing a previous bad change, you may find it
1451useful to check out an older version of a particular file using
b1889c36 1452linkgit:git-checkout[1]. We've used git-checkout before to switch
b684f830
BF
1453branches, but it has quite different behavior if it is given a path
1454name: the command
1455
1456-------------------------------------------------
1457$ git checkout HEAD^ path/to/file
1458-------------------------------------------------
1459
1460replaces path/to/file by the contents it had in the commit HEAD^, and
1461also updates the index to match. It does not change branches.
1462
1463If you just want to look at an old version of the file, without
1464modifying the working directory, you can do that with
5162e697 1465linkgit:git-show[1]:
b684f830
BF
1466
1467-------------------------------------------------
ed4eb0d8 1468$ git show HEAD^:path/to/file
b684f830
BF
1469-------------------------------------------------
1470
1471which will display the given version of the file.
1472
7a7cc594
JH
1473[[interrupted-work]]
1474Temporarily setting aside work in progress
1475~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1476
1477While you are in the middle of working on something complicated, you
1478find an unrelated but obvious and trivial bug. You would like to fix it
5162e697 1479before continuing. You can use linkgit:git-stash[1] to save the current
7a7cc594
JH
1480state of your work, and after fixing the bug (or, optionally after doing
1481so on a different branch and then coming back), unstash the
1482work-in-progress changes.
1483
1484------------------------------------------------
1485$ git stash "work in progress for foo feature"
1486------------------------------------------------
1487
1488This command will save your changes away to the `stash`, and
1489reset your working tree and the index to match the tip of your
1490current branch. Then you can make your fix as usual.
1491
1492------------------------------------------------
1493... edit and test ...
1494$ git commit -a -m "blorpl: typofix"
1495------------------------------------------------
1496
1497After that, you can go back to what you were working on with
1498`git stash apply`:
1499
1500------------------------------------------------
1501$ git stash apply
1502------------------------------------------------
1503
1504
e34caace 1505[[ensuring-good-performance]]
d19fbc3c
BF
1506Ensuring good performance
1507-------------------------
1508
1509On large repositories, git depends on compression to keep the history
06ada152 1510information from taking up too much space on disk or in memory.
d19fbc3c
BF
1511
1512This compression is not performed automatically. Therefore you
5162e697 1513should occasionally run linkgit:git-gc[1]:
d19fbc3c
BF
1514
1515-------------------------------------------------
1516$ git gc
1517-------------------------------------------------
1518
17217090
BF
1519to recompress the archive. This can be very time-consuming, so
1520you may prefer to run git-gc when you are not doing other work.
d19fbc3c 1521
e34caace
BF
1522
1523[[ensuring-reliability]]
11e016a3
BF
1524Ensuring reliability
1525--------------------
1526
e34caace 1527[[checking-for-corruption]]
11e016a3
BF
1528Checking the repository for corruption
1529~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1530
5162e697 1531The linkgit:git-fsck[1] command runs a number of self-consistency checks
1191ee18 1532on the repository, and reports on any problems. This may take some
21dcb3b7
BF
1533time. The most common warning by far is about "dangling" objects:
1534
1535-------------------------------------------------
04e50e94 1536$ git fsck
21dcb3b7
BF
1537dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1538dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1539dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1540dangling blob 218761f9d90712d37a9c5e36f406f92202db07eb
1541dangling commit bf093535a34a4d35731aa2bd90fe6b176302f14f
1542dangling commit 8e4bec7f2ddaa268bef999853c25755452100f8e
1543dangling tree d50bb86186bf27b681d25af89d3b5b68382e4085
1544dangling tree b24c2473f1fd3d91352a624795be026d64c8841f
1545...
1546-------------------------------------------------
1547
59723040 1548Dangling objects are not a problem. At worst they may take up a little
54782859 1549extra disk space. They can sometimes provide a last-resort method for
208641cf 1550recovering lost work--see <<dangling-objects>> for details.
1cdade2c 1551
e34caace 1552[[recovering-lost-changes]]
11e016a3
BF
1553Recovering lost changes
1554~~~~~~~~~~~~~~~~~~~~~~~
1555
e34caace 1556[[reflogs]]
559e4d7a
BF
1557Reflogs
1558^^^^^^^
1559
5162e697 1560Say you modify a branch with `linkgit:git-reset[1] --hard`, and then
559e4d7a
BF
1561realize that the branch was the only reference you had to that point in
1562history.
1563
1564Fortunately, git also keeps a log, called a "reflog", of all the
1565previous values of each branch. So in this case you can still find the
a6080a0a 1566old history using, for example,
559e4d7a
BF
1567
1568-------------------------------------------------
1569$ git log master@{1}
1570-------------------------------------------------
1571
e502c2c3
SO
1572This lists the commits reachable from the previous version of the
1573"master" branch head. This syntax can be used with any git command
1574that accepts a commit, not just with git log. Some other examples:
559e4d7a
BF
1575
1576-------------------------------------------------
1577$ git show master@{2} # See where the branch pointed 2,
1578$ git show master@{3} # 3, ... changes ago.
1579$ gitk master@{yesterday} # See where it pointed yesterday,
1580$ gitk master@{"1 week ago"} # ... or last week
953f3d6f
BF
1581$ git log --walk-reflogs master # show reflog entries for master
1582-------------------------------------------------
1583
1584A separate reflog is kept for the HEAD, so
1585
1586-------------------------------------------------
1587$ git show HEAD@{"1 week ago"}
559e4d7a
BF
1588-------------------------------------------------
1589
953f3d6f
BF
1590will show what HEAD pointed to one week ago, not what the current branch
1591pointed to one week ago. This allows you to see the history of what
1592you've checked out.
1593
559e4d7a 1594The reflogs are kept by default for 30 days, after which they may be
5162e697 1595pruned. See linkgit:git-reflog[1] and linkgit:git-gc[1] to learn
559e4d7a 1596how to control this pruning, and see the "SPECIFYING REVISIONS"
5162e697 1597section of linkgit:git-rev-parse[1] for details.
559e4d7a
BF
1598
1599Note that the reflog history is very different from normal git history.
1600While normal history is shared by every repository that works on the
1601same project, the reflog history is not shared: it tells you only about
1602how the branches in your local repository have changed over time.
1603
59723040 1604[[dangling-object-recovery]]
559e4d7a
BF
1605Examining dangling objects
1606^^^^^^^^^^^^^^^^^^^^^^^^^^
1607
59723040
BF
1608In some situations the reflog may not be able to save you. For example,
1609suppose you delete a branch, then realize you need the history it
1610contained. The reflog is also deleted; however, if you have not yet
1611pruned the repository, then you may still be able to find the lost
1612commits in the dangling objects that git-fsck reports. See
1613<<dangling-objects>> for the details.
559e4d7a
BF
1614
1615-------------------------------------------------
1616$ git fsck
1617dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1618dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1619dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1620...
1621-------------------------------------------------
1622
aacd404e 1623You can examine
559e4d7a
BF
1624one of those dangling commits with, for example,
1625
1626------------------------------------------------
1627$ gitk 7281251ddd --not --all
1628------------------------------------------------
1629
1630which does what it sounds like: it says that you want to see the commit
1631history that is described by the dangling commit(s), but not the
1632history that is described by all your existing branches and tags. Thus
1633you get exactly the history reachable from that commit that is lost.
1634(And notice that it might not be just one commit: we only report the
1635"tip of the line" as being dangling, but there might be a whole deep
79c96c57 1636and complex commit history that was dropped.)
559e4d7a
BF
1637
1638If you decide you want the history back, you can always create a new
1639reference pointing to it, for example, a new branch:
1640
1641------------------------------------------------
a6080a0a 1642$ git branch recovered-branch 7281251ddd
559e4d7a
BF
1643------------------------------------------------
1644
59723040
BF
1645Other types of dangling objects (blobs and trees) are also possible, and
1646dangling objects can arise in other situations.
1647
11e016a3 1648
e34caace 1649[[sharing-development]]
d19fbc3c 1650Sharing development with others
b684f830 1651===============================
d19fbc3c
BF
1652
1653[[getting-updates-with-git-pull]]
b1889c36 1654Getting updates with git-pull
b684f830 1655-----------------------------
d19fbc3c
BF
1656
1657After you clone a repository and make a few changes of your own, you
1658may wish to check the original repository for updates and merge them
1659into your own work.
1660
1661We have already seen <<Updating-a-repository-with-git-fetch,how to
5162e697 1662keep remote tracking branches up to date>> with linkgit:git-fetch[1],
d19fbc3c
BF
1663and how to merge two branches. So you can merge in changes from the
1664original repository's master branch with:
1665
1666-------------------------------------------------
1667$ git fetch
1668$ git merge origin/master
1669-------------------------------------------------
1670
5162e697 1671However, the linkgit:git-pull[1] command provides a way to do this in
d19fbc3c
BF
1672one step:
1673
1674-------------------------------------------------
1675$ git pull origin master
1676-------------------------------------------------
1677
0eb4f7cd
BF
1678In fact, if you have "master" checked out, then by default "git pull"
1679merges from the HEAD branch of the origin repository. So often you can
1680accomplish the above with just a simple
d19fbc3c
BF
1681
1682-------------------------------------------------
1683$ git pull
1684-------------------------------------------------
1685
0eb4f7cd
BF
1686More generally, a branch that is created from a remote branch will pull
1687by default from that branch. See the descriptions of the
1688branch.<name>.remote and branch.<name>.merge options in
5162e697
DM
1689linkgit:git-config[1], and the discussion of the `--track` option in
1690linkgit:git-checkout[1], to learn how to control these defaults.
d19fbc3c
BF
1691
1692In addition to saving you keystrokes, "git pull" also helps you by
1693producing a default commit message documenting the branch and
1694repository that you pulled from.
1695
1696(But note that no such commit will be created in the case of a
1697<<fast-forwards,fast forward>>; instead, your branch will just be
79c96c57 1698updated to point to the latest commit from the upstream branch.)
d19fbc3c 1699
1191ee18
BF
1700The git-pull command can also be given "." as the "remote" repository,
1701in which case it just merges in a branch from the current repository; so
4c63ff45
BF
1702the commands
1703
1704-------------------------------------------------
1705$ git pull . branch
1706$ git merge branch
1707-------------------------------------------------
1708
1709are roughly equivalent. The former is actually very commonly used.
1710
e34caace 1711[[submitting-patches]]
d19fbc3c 1712Submitting patches to a project
b684f830 1713-------------------------------
d19fbc3c
BF
1714
1715If you just have a few changes, the simplest way to submit them may
1716just be to send them as patches in email:
1717
5162e697 1718First, use linkgit:git-format-patch[1]; for example:
d19fbc3c
BF
1719
1720-------------------------------------------------
eb6ae7f4 1721$ git format-patch origin
d19fbc3c
BF
1722-------------------------------------------------
1723
1724will produce a numbered series of files in the current directory, one
1725for each patch in the current branch but not in origin/HEAD.
1726
1727You can then import these into your mail client and send them by
1728hand. However, if you have a lot to send at once, you may prefer to
5162e697 1729use the linkgit:git-send-email[1] script to automate the process.
d19fbc3c
BF
1730Consult the mailing list for your project first to determine how they
1731prefer such patches be handled.
1732
e34caace 1733[[importing-patches]]
d19fbc3c 1734Importing patches to a project
b684f830 1735------------------------------
d19fbc3c 1736
5162e697 1737Git also provides a tool called linkgit:git-am[1] (am stands for
d19fbc3c
BF
1738"apply mailbox"), for importing such an emailed series of patches.
1739Just save all of the patch-containing messages, in order, into a
1740single mailbox file, say "patches.mbox", then run
1741
1742-------------------------------------------------
eb6ae7f4 1743$ git am -3 patches.mbox
d19fbc3c
BF
1744-------------------------------------------------
1745
1746Git will apply each patch in order; if any conflicts are found, it
1747will stop, and you can fix the conflicts as described in
01997b4a
BF
1748"<<resolving-a-merge,Resolving a merge>>". (The "-3" option tells
1749git to perform a merge; if you would prefer it just to abort and
1750leave your tree and index untouched, you may omit that option.)
1751
1752Once the index is updated with the results of the conflict
1753resolution, instead of creating a new commit, just run
d19fbc3c
BF
1754
1755-------------------------------------------------
1756$ git am --resolved
1757-------------------------------------------------
1758
1759and git will create the commit for you and continue applying the
1760remaining patches from the mailbox.
1761
1762The final result will be a series of commits, one for each patch in
1763the original mailbox, with authorship and commit log message each
1764taken from the message containing each patch.
1765
eda69449
BF
1766[[public-repositories]]
1767Public git repositories
1768-----------------------
d19fbc3c 1769
6e30fb0c
DK
1770Another way to submit changes to a project is to tell the maintainer
1771of that project to pull the changes from your repository using
5162e697 1772linkgit:git-pull[1]. In the section "<<getting-updates-with-git-pull,
b1889c36 1773Getting updates with git-pull>>" we described this as a way to get
6e30fb0c
DK
1774updates from the "main" repository, but it works just as well in the
1775other direction.
d19fbc3c 1776
eda69449
BF
1777If you and the maintainer both have accounts on the same machine, then
1778you can just pull changes from each other's repositories directly;
11d51533 1779commands that accept repository URLs as arguments will also accept a
eda69449 1780local directory name:
d19fbc3c
BF
1781
1782-------------------------------------------------
1783$ git clone /path/to/repository
1784$ git pull /path/to/other/repository
1785-------------------------------------------------
1786
c9016158 1787or an ssh URL:
11d51533
BF
1788
1789-------------------------------------------------
1790$ git clone ssh://yourhost/~you/repository
1791-------------------------------------------------
1792
1793For projects with few developers, or for synchronizing a few private
1794repositories, this may be all you need.
1795
eda69449
BF
1796However, the more common way to do this is to maintain a separate public
1797repository (usually on a different host) for others to pull changes
1798from. This is usually more convenient, and allows you to cleanly
1799separate private work in progress from publicly visible work.
d19fbc3c
BF
1800
1801You will continue to do your day-to-day work in your personal
1802repository, but periodically "push" changes from your personal
1803repository into your public repository, allowing other developers to
1804pull from that repository. So the flow of changes, in a situation
1805where there is one other developer with a public repository, looks
1806like this:
1807
1808 you push
1809 your personal repo ------------------> your public repo
a6080a0a 1810 ^ |
d19fbc3c
BF
1811 | |
1812 | you pull | they pull
1813 | |
1814 | |
1815 | they push V
1816 their public repo <------------------- their repo
1817
11d51533
BF
1818We explain how to do this in the following sections.
1819
eda69449
BF
1820[[setting-up-a-public-repository]]
1821Setting up a public repository
1822~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1823
1824Assume your personal repository is in the directory ~/proj. We
1825first create a new clone of the repository and tell git-daemon that it
1826is meant to be public:
d19fbc3c
BF
1827
1828-------------------------------------------------
52c80037 1829$ git clone --bare ~/proj proj.git
eda69449 1830$ touch proj.git/git-daemon-export-ok
d19fbc3c
BF
1831-------------------------------------------------
1832
52c80037 1833The resulting directory proj.git contains a "bare" git repository--it is
eda69449
BF
1834just the contents of the ".git" directory, without any files checked out
1835around it.
d19fbc3c 1836
c64415e2 1837Next, copy proj.git to the server where you plan to host the
d19fbc3c
BF
1838public repository. You can use scp, rsync, or whatever is most
1839convenient.
1840
eda69449
BF
1841[[exporting-via-git]]
1842Exporting a git repository via the git protocol
1843~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1844
1845This is the preferred method.
1846
1847If someone else administers the server, they should tell you what
c9016158 1848directory to put the repository in, and what git:// URL it will appear
eda69449 1849at. You can then skip to the section
d19fbc3c
BF
1850"<<pushing-changes-to-a-public-repository,Pushing changes to a public
1851repository>>", below.
1852
5162e697 1853Otherwise, all you need to do is start linkgit:git-daemon[1]; it will
eda69449
BF
1854listen on port 9418. By default, it will allow access to any directory
1855that looks like a git directory and contains the magic file
1856git-daemon-export-ok. Passing some directory paths as git-daemon
1857arguments will further restrict the exports to those paths.
1858
1859You can also run git-daemon as an inetd service; see the
5162e697 1860linkgit:git-daemon[1] man page for details. (See especially the
eda69449 1861examples section.)
d19fbc3c
BF
1862
1863[[exporting-via-http]]
1864Exporting a git repository via http
eda69449 1865~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c
BF
1866
1867The git protocol gives better performance and reliability, but on a
1868host with a web server set up, http exports may be simpler to set up.
1869
1870All you need to do is place the newly created bare git repository in
1871a directory that is exported by the web server, and make some
1872adjustments to give web clients some extra information they need:
1873
1874-------------------------------------------------
1875$ mv proj.git /home/you/public_html/proj.git
1876$ cd proj.git
c64415e2 1877$ git --bare update-server-info
d19fbc3c
BF
1878$ chmod a+x hooks/post-update
1879-------------------------------------------------
1880
1881(For an explanation of the last two lines, see
6998e4db 1882linkgit:git-update-server-info[1] and linkgit:githooks[5].)
d19fbc3c 1883
c9016158
RW
1884Advertise the URL of proj.git. Anybody else should then be able to
1885clone or pull from that URL, for example with a command line like:
d19fbc3c
BF
1886
1887-------------------------------------------------
1888$ git clone http://yourserver.com/~you/proj.git
1889-------------------------------------------------
1890
1891(See also
1892link:howto/setup-git-server-over-http.txt[setup-git-server-over-http]
1893for a slightly more sophisticated setup using WebDAV which also
1894allows pushing over http.)
1895
d19fbc3c
BF
1896[[pushing-changes-to-a-public-repository]]
1897Pushing changes to a public repository
eda69449 1898~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c 1899
eda69449 1900Note that the two techniques outlined above (exporting via
d19fbc3c
BF
1901<<exporting-via-http,http>> or <<exporting-via-git,git>>) allow other
1902maintainers to fetch your latest changes, but they do not allow write
1903access, which you will need to update the public repository with the
1904latest changes created in your private repository.
1905
5162e697 1906The simplest way to do this is using linkgit:git-push[1] and ssh; to
d19fbc3c
BF
1907update the remote branch named "master" with the latest state of your
1908branch named "master", run
1909
1910-------------------------------------------------
1911$ git push ssh://yourserver.com/~you/proj.git master:master
1912-------------------------------------------------
1913
1914or just
1915
1916-------------------------------------------------
1917$ git push ssh://yourserver.com/~you/proj.git master
1918-------------------------------------------------
1919
81eb417a
BF
1920As with git-fetch, git-push will complain if this does not result in a
1921<<fast-forwards,fast forward>>; see the following section for details on
1922handling this case.
d19fbc3c 1923
11d51533
BF
1924Note that the target of a "push" is normally a
1925<<def_bare_repository,bare>> repository. You can also push to a
1926repository that has a checked-out working tree, but the working tree
1927will not be updated by the push. This may lead to unexpected results if
1928the branch you push to is the currently checked-out branch!
1929
d19fbc3c
BF
1930As with git-fetch, you may also set up configuration options to
1931save typing; so, for example, after
1932
1933-------------------------------------------------
c64415e2 1934$ cat >>.git/config <<EOF
d19fbc3c
BF
1935[remote "public-repo"]
1936 url = ssh://yourserver.com/~you/proj.git
1937EOF
1938-------------------------------------------------
1939
1940you should be able to perform the above push with just
1941
1942-------------------------------------------------
1943$ git push public-repo master
1944-------------------------------------------------
1945
1946See the explanations of the remote.<name>.url, branch.<name>.remote,
5162e697 1947and remote.<name>.push options in linkgit:git-config[1] for
d19fbc3c
BF
1948details.
1949
81eb417a
BF
1950[[forcing-push]]
1951What to do when a push fails
1952~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1953
1954If a push would not result in a <<fast-forwards,fast forward>> of the
1955remote branch, then it will fail with an error like:
1956
1957-------------------------------------------------
1958error: remote 'refs/heads/master' is not an ancestor of
1959 local 'refs/heads/master'.
1960 Maybe you are not up-to-date and need to pull first?
1961error: failed to push to 'ssh://yourserver.com/~you/proj.git'
1962-------------------------------------------------
1963
1964This can happen, for example, if you:
1965
467c0197
JN
1966 - use `git-reset --hard` to remove already-published commits, or
1967 - use `git-commit --amend` to replace already-published commits
7cb192ea 1968 (as in <<fixing-a-mistake-by-rewriting-history>>), or
467c0197 1969 - use `git-rebase` to rebase any already-published commits (as
81eb417a
BF
1970 in <<using-git-rebase>>).
1971
1972You may force git-push to perform the update anyway by preceding the
1973branch name with a plus sign:
1974
1975-------------------------------------------------
1976$ git push ssh://yourserver.com/~you/proj.git +master
1977-------------------------------------------------
1978
1979Normally whenever a branch head in a public repository is modified, it
9e5d87d4 1980is modified to point to a descendant of the commit that it pointed to
81eb417a
BF
1981before. By forcing a push in this situation, you break that convention.
1982(See <<problems-with-rewriting-history>>.)
1983
1984Nevertheless, this is a common practice for people that need a simple
1985way to publish a work-in-progress patch series, and it is an acceptable
1986compromise as long as you warn other developers that this is how you
1987intend to manage the branch.
1988
1989It's also possible for a push to fail in this way when other people have
1990the right to push to the same repository. In that case, the correct
1991solution is to retry the push after first updating your work by either a
1992pull or a fetch followed by a rebase; see the
1993<<setting-up-a-shared-repository,next section>> and
6998e4db 1994linkgit:gitcvs-migration[7] for more.
81eb417a 1995
e34caace 1996[[setting-up-a-shared-repository]]
d19fbc3c 1997Setting up a shared repository
eda69449 1998~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c
BF
1999
2000Another way to collaborate is by using a model similar to that
2001commonly used in CVS, where several developers with special rights
2002all push to and pull from a single shared repository. See
6998e4db 2003linkgit:gitcvs-migration[7] for instructions on how to
d19fbc3c
BF
2004set this up.
2005
8fae2225
BF
2006However, while there is nothing wrong with git's support for shared
2007repositories, this mode of operation is not generally recommended,
2008simply because the mode of collaboration that git supports--by
2009exchanging patches and pulling from public repositories--has so many
2010advantages over the central shared repository:
2011
2012 - Git's ability to quickly import and merge patches allows a
2013 single maintainer to process incoming changes even at very
2014 high rates. And when that becomes too much, git-pull provides
2015 an easy way for that maintainer to delegate this job to other
2016 maintainers while still allowing optional review of incoming
2017 changes.
2018 - Since every developer's repository has the same complete copy
2019 of the project history, no repository is special, and it is
2020 trivial for another developer to take over maintenance of a
2021 project, either by mutual agreement, or because a maintainer
2022 becomes unresponsive or difficult to work with.
2023 - The lack of a central group of "committers" means there is
2024 less need for formal decisions about who is "in" and who is
2025 "out".
2026
e34caace 2027[[setting-up-gitweb]]
eda69449
BF
2028Allowing web browsing of a repository
2029~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c 2030
a8cd1402
BF
2031The gitweb cgi script provides users an easy way to browse your
2032project's files and history without having to install git; see the file
04483524 2033gitweb/INSTALL in the git source tree for instructions on setting it up.
d19fbc3c 2034
e34caace 2035[[sharing-development-examples]]
b684f830
BF
2036Examples
2037--------
d19fbc3c 2038
9e2163ea
BF
2039[[maintaining-topic-branches]]
2040Maintaining topic branches for a Linux subsystem maintainer
2041~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2042
2043This describes how Tony Luck uses git in his role as maintainer of the
2044IA64 architecture for the Linux kernel.
2045
2046He uses two public branches:
2047
2048 - A "test" tree into which patches are initially placed so that they
2049 can get some exposure when integrated with other ongoing development.
2050 This tree is available to Andrew for pulling into -mm whenever he
2051 wants.
2052
2053 - A "release" tree into which tested patches are moved for final sanity
2054 checking, and as a vehicle to send them upstream to Linus (by sending
2055 him a "please pull" request.)
2056
2057He also uses a set of temporary branches ("topic branches"), each
2058containing a logical grouping of patches.
2059
2060To set this up, first create your work tree by cloning Linus's public
2061tree:
2062
2063-------------------------------------------------
2064$ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git work
2065$ cd work
2066-------------------------------------------------
2067
2068Linus's tree will be stored in the remote branch named origin/master,
5162e697
DM
2069and can be updated using linkgit:git-fetch[1]; you can track other
2070public trees using linkgit:git-remote[1] to set up a "remote" and
2071linkgit:git-fetch[1] to keep them up-to-date; see
6e30fb0c 2072<<repositories-and-branches>>.
9e2163ea
BF
2073
2074Now create the branches in which you are going to work; these start out
2075at the current tip of origin/master branch, and should be set up (using
5162e697 2076the --track option to linkgit:git-branch[1]) to merge changes in from
9e2163ea
BF
2077Linus by default.
2078
2079-------------------------------------------------
2080$ git branch --track test origin/master
2081$ git branch --track release origin/master
2082-------------------------------------------------
2083
5162e697 2084These can be easily kept up to date using linkgit:git-pull[1].
9e2163ea
BF
2085
2086-------------------------------------------------
2087$ git checkout test && git pull
2088$ git checkout release && git pull
2089-------------------------------------------------
2090
2091Important note! If you have any local changes in these branches, then
2092this merge will create a commit object in the history (with no local
2093changes git will simply do a "Fast forward" merge). Many people dislike
2094the "noise" that this creates in the Linux history, so you should avoid
2095doing this capriciously in the "release" branch, as these noisy commits
2096will become part of the permanent history when you ask Linus to pull
2097from the release branch.
2098
5162e697 2099A few configuration variables (see linkgit:git-config[1]) can
9e2163ea
BF
2100make it easy to push both branches to your public tree. (See
2101<<setting-up-a-public-repository>>.)
2102
2103-------------------------------------------------
2104$ cat >> .git/config <<EOF
2105[remote "mytree"]
2106 url = master.kernel.org:/pub/scm/linux/kernel/git/aegl/linux-2.6.git
2107 push = release
2108 push = test
2109EOF
2110-------------------------------------------------
2111
2112Then you can push both the test and release trees using
5162e697 2113linkgit:git-push[1]:
9e2163ea
BF
2114
2115-------------------------------------------------
2116$ git push mytree
2117-------------------------------------------------
2118
2119or push just one of the test and release branches using:
2120
2121-------------------------------------------------
2122$ git push mytree test
2123-------------------------------------------------
2124
2125or
2126
2127-------------------------------------------------
2128$ git push mytree release
2129-------------------------------------------------
2130
2131Now to apply some patches from the community. Think of a short
2132snappy name for a branch to hold this patch (or related group of
2133patches), and create a new branch from the current tip of Linus's
2134branch:
2135
2136-------------------------------------------------
2137$ git checkout -b speed-up-spinlocks origin
2138-------------------------------------------------
2139
2140Now you apply the patch(es), run some tests, and commit the change(s). If
2141the patch is a multi-part series, then you should apply each as a separate
2142commit to this branch.
2143
2144-------------------------------------------------
2145$ ... patch ... test ... commit [ ... patch ... test ... commit ]*
2146-------------------------------------------------
2147
2148When you are happy with the state of this change, you can pull it into the
2149"test" branch in preparation to make it public:
2150
2151-------------------------------------------------
2152$ git checkout test && git pull . speed-up-spinlocks
2153-------------------------------------------------
2154
2155It is unlikely that you would have any conflicts here ... but you might if you
2156spent a while on this step and had also pulled new versions from upstream.
2157
2158Some time later when enough time has passed and testing done, you can pull the
2159same branch into the "release" tree ready to go upstream. This is where you
2160see the value of keeping each patch (or patch series) in its own branch. It
2161means that the patches can be moved into the "release" tree in any order.
2162
2163-------------------------------------------------
2164$ git checkout release && git pull . speed-up-spinlocks
2165-------------------------------------------------
2166
2167After a while, you will have a number of branches, and despite the
2168well chosen names you picked for each of them, you may forget what
2169they are for, or what status they are in. To get a reminder of what
2170changes are in a specific branch, use:
2171
2172-------------------------------------------------
467c0197 2173$ git log linux..branchname | git shortlog
9e2163ea
BF
2174-------------------------------------------------
2175
06ada152 2176To see whether it has already been merged into the test or release branches,
9e2163ea
BF
2177use:
2178
2179-------------------------------------------------
2180$ git log test..branchname
2181-------------------------------------------------
2182
2183or
2184
2185-------------------------------------------------
2186$ git log release..branchname
2187-------------------------------------------------
2188
06ada152 2189(If this branch has not yet been merged, you will see some log entries.
9e2163ea
BF
2190If it has been merged, then there will be no output.)
2191
2192Once a patch completes the great cycle (moving from test to release,
2193then pulled by Linus, and finally coming back into your local
06ada152 2194"origin/master" branch), the branch for this change is no longer needed.
9e2163ea
BF
2195You detect this when the output from:
2196
2197-------------------------------------------------
2198$ git log origin..branchname
2199-------------------------------------------------
2200
2201is empty. At this point the branch can be deleted:
2202
2203-------------------------------------------------
2204$ git branch -d branchname
2205-------------------------------------------------
2206
2207Some changes are so trivial that it is not necessary to create a separate
2208branch and then merge into each of the test and release branches. For
2209these changes, just apply directly to the "release" branch, and then
2210merge that into the "test" branch.
2211
2212To create diffstat and shortlog summaries of changes to include in a "please
2213pull" request to Linus you can use:
2214
2215-------------------------------------------------
2216$ git diff --stat origin..release
2217-------------------------------------------------
2218
2219and
2220
2221-------------------------------------------------
2222$ git log -p origin..release | git shortlog
2223-------------------------------------------------
2224
2225Here are some of the scripts that simplify all this even further.
2226
2227-------------------------------------------------
2228==== update script ====
2229# Update a branch in my GIT tree. If the branch to be updated
2230# is origin, then pull from kernel.org. Otherwise merge
2231# origin/master branch into test|release branch
2232
2233case "$1" in
2234test|release)
2235 git checkout $1 && git pull . origin
2236 ;;
2237origin)
fc74ecc1 2238 before=$(git rev-parse refs/remotes/origin/master)
9e2163ea 2239 git fetch origin
fc74ecc1 2240 after=$(git rev-parse refs/remotes/origin/master)
9e2163ea
BF
2241 if [ $before != $after ]
2242 then
2243 git log $before..$after | git shortlog
2244 fi
2245 ;;
2246*)
2247 echo "Usage: $0 origin|test|release" 1>&2
2248 exit 1
2249 ;;
2250esac
2251-------------------------------------------------
2252
2253-------------------------------------------------
2254==== merge script ====
2255# Merge a branch into either the test or release branch
2256
2257pname=$0
2258
2259usage()
2260{
2261 echo "Usage: $pname branch test|release" 1>&2
2262 exit 1
2263}
2264
fc74ecc1 2265git show-ref -q --verify -- refs/heads/"$1" || {
9e2163ea
BF
2266 echo "Can't see branch <$1>" 1>&2
2267 usage
fc74ecc1 2268}
9e2163ea
BF
2269
2270case "$2" in
2271test|release)
2272 if [ $(git log $2..$1 | wc -c) -eq 0 ]
2273 then
2274 echo $1 already merged into $2 1>&2
2275 exit 1
2276 fi
2277 git checkout $2 && git pull . $1
2278 ;;
2279*)
2280 usage
2281 ;;
2282esac
2283-------------------------------------------------
2284
2285-------------------------------------------------
2286==== status script ====
2287# report on status of my ia64 GIT tree
2288
2289gb=$(tput setab 2)
2290rb=$(tput setab 1)
2291restore=$(tput setab 9)
2292
2293if [ `git rev-list test..release | wc -c` -gt 0 ]
2294then
2295 echo $rb Warning: commits in release that are not in test $restore
2296 git log test..release
2297fi
2298
fc74ecc1 2299for branch in `git show-ref --heads | sed 's|^.*/||'`
9e2163ea
BF
2300do
2301 if [ $branch = test -o $branch = release ]
2302 then
2303 continue
2304 fi
2305
2306 echo -n $gb ======= $branch ====== $restore " "
2307 status=
2308 for ref in test release origin/master
2309 do
2310 if [ `git rev-list $ref..$branch | wc -c` -gt 0 ]
2311 then
2312 status=$status${ref:0:1}
2313 fi
2314 done
2315 case $status in
2316 trl)
2317 echo $rb Need to pull into test $restore
2318 ;;
2319 rl)
2320 echo "In test"
2321 ;;
2322 l)
2323 echo "Waiting for linus"
2324 ;;
2325 "")
2326 echo $rb All done $restore
2327 ;;
2328 *)
2329 echo $rb "<$status>" $restore
2330 ;;
2331 esac
2332 git log origin/master..$branch | git shortlog
2333done
2334-------------------------------------------------
d19fbc3c 2335
d19fbc3c 2336
d19fbc3c 2337[[cleaning-up-history]]
4c63ff45
BF
2338Rewriting history and maintaining patch series
2339==============================================
2340
2341Normally commits are only added to a project, never taken away or
2342replaced. Git is designed with this assumption, and violating it will
2343cause git's merge machinery (for example) to do the wrong thing.
2344
2345However, there is a situation in which it can be useful to violate this
2346assumption.
2347
e34caace 2348[[patch-series]]
4c63ff45
BF
2349Creating the perfect patch series
2350---------------------------------
2351
2352Suppose you are a contributor to a large project, and you want to add a
2353complicated feature, and to present it to the other developers in a way
2354that makes it easy for them to read your changes, verify that they are
2355correct, and understand why you made each change.
2356
b181d57f 2357If you present all of your changes as a single patch (or commit), they
79c96c57 2358may find that it is too much to digest all at once.
4c63ff45
BF
2359
2360If you present them with the entire history of your work, complete with
2361mistakes, corrections, and dead ends, they may be overwhelmed.
2362
2363So the ideal is usually to produce a series of patches such that:
2364
2365 1. Each patch can be applied in order.
2366
2367 2. Each patch includes a single logical change, together with a
2368 message explaining the change.
2369
2370 3. No patch introduces a regression: after applying any initial
2371 part of the series, the resulting project still compiles and
2372 works, and has no bugs that it didn't have before.
2373
2374 4. The complete series produces the same end result as your own
2375 (probably much messier!) development process did.
2376
b181d57f
BF
2377We will introduce some tools that can help you do this, explain how to
2378use them, and then explain some of the problems that can arise because
2379you are rewriting history.
4c63ff45 2380
e34caace 2381[[using-git-rebase]]
4c63ff45
BF
2382Keeping a patch series up to date using git-rebase
2383--------------------------------------------------
2384
79c96c57
MC
2385Suppose that you create a branch "mywork" on a remote-tracking branch
2386"origin", and create some commits on top of it:
4c63ff45
BF
2387
2388-------------------------------------------------
2389$ git checkout -b mywork origin
2390$ vi file.txt
2391$ git commit
2392$ vi otherfile.txt
2393$ git commit
2394...
2395-------------------------------------------------
2396
2397You have performed no merges into mywork, so it is just a simple linear
2398sequence of patches on top of "origin":
2399
1dc71a91 2400................................................
4c63ff45
BF
2401 o--o--o <-- origin
2402 \
2403 o--o--o <-- mywork
1dc71a91 2404................................................
4c63ff45
BF
2405
2406Some more interesting work has been done in the upstream project, and
2407"origin" has advanced:
2408
1dc71a91 2409................................................
4c63ff45
BF
2410 o--o--O--o--o--o <-- origin
2411 \
2412 a--b--c <-- mywork
1dc71a91 2413................................................
4c63ff45
BF
2414
2415At this point, you could use "pull" to merge your changes back in;
2416the result would create a new merge commit, like this:
2417
1dc71a91 2418................................................
4c63ff45
BF
2419 o--o--O--o--o--o <-- origin
2420 \ \
2421 a--b--c--m <-- mywork
1dc71a91 2422................................................
a6080a0a 2423
4c63ff45
BF
2424However, if you prefer to keep the history in mywork a simple series of
2425commits without any merges, you may instead choose to use
5162e697 2426linkgit:git-rebase[1]:
4c63ff45
BF
2427
2428-------------------------------------------------
2429$ git checkout mywork
2430$ git rebase origin
2431-------------------------------------------------
2432
b181d57f
BF
2433This will remove each of your commits from mywork, temporarily saving
2434them as patches (in a directory named ".dotest"), update mywork to
2435point at the latest version of origin, then apply each of the saved
2436patches to the new mywork. The result will look like:
4c63ff45
BF
2437
2438
1dc71a91 2439................................................
4c63ff45
BF
2440 o--o--O--o--o--o <-- origin
2441 \
2442 a'--b'--c' <-- mywork
1dc71a91 2443................................................
4c63ff45 2444
b181d57f 2445In the process, it may discover conflicts. In that case it will stop
467c0197 2446and allow you to fix the conflicts; after fixing conflicts, use "git-add"
7a7d4ef6 2447to update the index with those contents, and then, instead of
b181d57f 2448running git-commit, just run
4c63ff45
BF
2449
2450-------------------------------------------------
2451$ git rebase --continue
2452-------------------------------------------------
2453
2454and git will continue applying the rest of the patches.
2455
b6cbca38 2456At any point you may use the `--abort` option to abort this process and
4c63ff45
BF
2457return mywork to the state it had before you started the rebase:
2458
2459-------------------------------------------------
2460$ git rebase --abort
2461-------------------------------------------------
2462
7cb192ea
BF
2463[[rewriting-one-commit]]
2464Rewriting a single commit
365aa199
BF
2465-------------------------
2466
7cb192ea 2467We saw in <<fixing-a-mistake-by-rewriting-history>> that you can replace the
365aa199
BF
2468most recent commit using
2469
2470-------------------------------------------------
2471$ git commit --amend
2472-------------------------------------------------
2473
2474which will replace the old commit by a new commit incorporating your
2475changes, giving you a chance to edit the old commit message first.
2476
5162e697 2477You can also use a combination of this and linkgit:git-rebase[1] to
7cb192ea
BF
2478replace a commit further back in your history and recreate the
2479intervening changes on top of it. First, tag the problematic commit
2480with
365aa199
BF
2481
2482-------------------------------------------------
2483$ git tag bad mywork~5
2484-------------------------------------------------
2485
2486(Either gitk or git-log may be useful for finding the commit.)
2487
25d9f3fa
BF
2488Then check out that commit, edit it, and rebase the rest of the series
2489on top of it (note that we could check out the commit on a temporary
2490branch, but instead we're using a <<detached-head,detached head>>):
365aa199
BF
2491
2492-------------------------------------------------
25d9f3fa 2493$ git checkout bad
365aa199
BF
2494$ # make changes here and update the index
2495$ git commit --amend
25d9f3fa 2496$ git rebase --onto HEAD bad mywork
365aa199
BF
2497-------------------------------------------------
2498
25d9f3fa
BF
2499When you're done, you'll be left with mywork checked out, with the top
2500patches on mywork reapplied on top of your modified commit. You can
365aa199
BF
2501then clean up with
2502
2503-------------------------------------------------
365aa199
BF
2504$ git tag -d bad
2505-------------------------------------------------
2506
2507Note that the immutable nature of git history means that you haven't really
2508"modified" existing commits; instead, you have replaced the old commits with
2509new commits having new object names.
2510
e34caace 2511[[reordering-patch-series]]
4c63ff45
BF
2512Reordering or selecting from a patch series
2513-------------------------------------------
2514
5162e697 2515Given one existing commit, the linkgit:git-cherry-pick[1] command
b181d57f
BF
2516allows you to apply the change introduced by that commit and create a
2517new commit that records it. So, for example, if "mywork" points to a
2518series of patches on top of "origin", you might do something like:
2519
2520-------------------------------------------------
2521$ git checkout -b mywork-new origin
2522$ gitk origin..mywork &
2523-------------------------------------------------
2524
06ada152 2525and browse through the list of patches in the mywork branch using gitk,
b181d57f 2526applying them (possibly in a different order) to mywork-new using
b6cbca38 2527cherry-pick, and possibly modifying them as you go using `commit --amend`.
5162e697 2528The linkgit:git-gui[1] command may also help as it allows you to
6e30fb0c
DK
2529individually select diff hunks for inclusion in the index (by
2530right-clicking on the diff hunk and choosing "Stage Hunk for Commit").
b181d57f
BF
2531
2532Another technique is to use git-format-patch to create a series of
2533patches, then reset the state to before the patches:
4c63ff45 2534
b181d57f
BF
2535-------------------------------------------------
2536$ git format-patch origin
2537$ git reset --hard origin
2538-------------------------------------------------
4c63ff45 2539
b181d57f 2540Then modify, reorder, or eliminate patches as preferred before applying
5162e697 2541them again with linkgit:git-am[1].
4c63ff45 2542
e34caace 2543[[patch-series-tools]]
4c63ff45
BF
2544Other tools
2545-----------
2546
02783075 2547There are numerous other tools, such as StGIT, which exist for the
79c96c57 2548purpose of maintaining a patch series. These are outside of the scope of
b181d57f 2549this manual.
4c63ff45 2550
e34caace 2551[[problems-with-rewriting-history]]
4c63ff45
BF
2552Problems with rewriting history
2553-------------------------------
2554
b181d57f
BF
2555The primary problem with rewriting the history of a branch has to do
2556with merging. Suppose somebody fetches your branch and merges it into
2557their branch, with a result something like this:
2558
1dc71a91 2559................................................
b181d57f
BF
2560 o--o--O--o--o--o <-- origin
2561 \ \
2562 t--t--t--m <-- their branch:
1dc71a91 2563................................................
b181d57f
BF
2564
2565Then suppose you modify the last three commits:
2566
1dc71a91 2567................................................
b181d57f
BF
2568 o--o--o <-- new head of origin
2569 /
2570 o--o--O--o--o--o <-- old head of origin
1dc71a91 2571................................................
b181d57f
BF
2572
2573If we examined all this history together in one repository, it will
2574look like:
2575
1dc71a91 2576................................................
b181d57f
BF
2577 o--o--o <-- new head of origin
2578 /
2579 o--o--O--o--o--o <-- old head of origin
2580 \ \
2581 t--t--t--m <-- their branch:
1dc71a91 2582................................................
b181d57f
BF
2583
2584Git has no way of knowing that the new head is an updated version of
2585the old head; it treats this situation exactly the same as it would if
2586two developers had independently done the work on the old and new heads
2587in parallel. At this point, if someone attempts to merge the new head
2588in to their branch, git will attempt to merge together the two (old and
2589new) lines of development, instead of trying to replace the old by the
2590new. The results are likely to be unexpected.
2591
2592You may still choose to publish branches whose history is rewritten,
2593and it may be useful for others to be able to fetch those branches in
2594order to examine or test them, but they should not attempt to pull such
2595branches into their own work.
2596
2597For true distributed development that supports proper merging,
2598published branches should never be rewritten.
2599
3fb00282
SP
2600[[bisect-merges]]
2601Why bisecting merge commits can be harder than bisecting linear history
2602-----------------------------------------------------------------------
2603
5162e697 2604The linkgit:git-bisect[1] command correctly handles history that
3fb00282
SP
2605includes merge commits. However, when the commit that it finds is a
2606merge commit, the user may need to work harder than usual to figure out
2607why that commit introduced a problem.
2608
2609Imagine this history:
2610
2611................................................
2612 ---Z---o---X---...---o---A---C---D
2613 \ /
2614 o---o---Y---...---o---B
2615................................................
2616
2617Suppose that on the upper line of development, the meaning of one
2618of the functions that exists at Z is changed at commit X. The
2619commits from Z leading to A change both the function's
2620implementation and all calling sites that exist at Z, as well
2621as new calling sites they add, to be consistent. There is no
2622bug at A.
2623
2624Suppose that in the meantime on the lower line of development somebody
2625adds a new calling site for that function at commit Y. The
2626commits from Z leading to B all assume the old semantics of that
2627function and the callers and the callee are consistent with each
2628other. There is no bug at B, either.
2629
2630Suppose further that the two development lines merge cleanly at C,
2631so no conflict resolution is required.
2632
2633Nevertheless, the code at C is broken, because the callers added
2634on the lower line of development have not been converted to the new
2635semantics introduced on the upper line of development. So if all
2636you know is that D is bad, that Z is good, and that
5162e697 2637linkgit:git-bisect[1] identifies C as the culprit, how will you
3fb00282
SP
2638figure out that the problem is due to this change in semantics?
2639
2640When the result of a git-bisect is a non-merge commit, you should
2641normally be able to discover the problem by examining just that commit.
2642Developers can make this easy by breaking their changes into small
2643self-contained commits. That won't help in the case above, however,
2644because the problem isn't obvious from examination of any single
2645commit; instead, a global view of the development is required. To
2646make matters worse, the change in semantics in the problematic
2647function may be just one small part of the changes in the upper
2648line of development.
2649
2650On the other hand, if instead of merging at C you had rebased the
2651history between Z to B on top of A, you would have gotten this
2652linear history:
2653
2654................................................................
2655 ---Z---o---X--...---o---A---o---o---Y*--...---o---B*--D*
2656................................................................
2657
2658Bisecting between Z and D* would hit a single culprit commit Y*,
2659and understanding why Y* was broken would probably be easier.
2660
2661Partly for this reason, many experienced git users, even when
2662working on an otherwise merge-heavy project, keep the history
2663linear by rebasing against the latest upstream version before
2664publishing.
2665
e34caace 2666[[advanced-branch-management]]
b181d57f
BF
2667Advanced branch management
2668==========================
4c63ff45 2669
e34caace 2670[[fetching-individual-branches]]
b181d57f
BF
2671Fetching individual branches
2672----------------------------
2673
5162e697 2674Instead of using linkgit:git-remote[1], you can also choose just
b181d57f
BF
2675to update one branch at a time, and to store it locally under an
2676arbitrary name:
2677
2678-------------------------------------------------
2679$ git fetch origin todo:my-todo-work
2680-------------------------------------------------
2681
2682The first argument, "origin", just tells git to fetch from the
2683repository you originally cloned from. The second argument tells git
2684to fetch the branch named "todo" from the remote repository, and to
2685store it locally under the name refs/heads/my-todo-work.
2686
2687You can also fetch branches from other repositories; so
2688
2689-------------------------------------------------
2690$ git fetch git://example.com/proj.git master:example-master
2691-------------------------------------------------
2692
2693will create a new branch named "example-master" and store in it the
2694branch named "master" from the repository at the given URL. If you
2695already have a branch named example-master, it will attempt to
59723040
BF
2696<<fast-forwards,fast-forward>> to the commit given by example.com's
2697master branch. In more detail:
b181d57f 2698
59723040
BF
2699[[fetch-fast-forwards]]
2700git fetch and fast-forwards
2701---------------------------
b181d57f 2702
467c0197 2703In the previous example, when updating an existing branch, "git-fetch"
7a7d4ef6 2704checks to make sure that the most recent commit on the remote
b181d57f
BF
2705branch is a descendant of the most recent commit on your copy of the
2706branch before updating your copy of the branch to point at the new
59723040 2707commit. Git calls this process a <<fast-forwards,fast forward>>.
b181d57f
BF
2708
2709A fast forward looks something like this:
2710
1dc71a91 2711................................................
b181d57f
BF
2712 o--o--o--o <-- old head of the branch
2713 \
2714 o--o--o <-- new head of the branch
1dc71a91 2715................................................
b181d57f
BF
2716
2717
2718In some cases it is possible that the new head will *not* actually be
2719a descendant of the old head. For example, the developer may have
2720realized she made a serious mistake, and decided to backtrack,
2721resulting in a situation like:
2722
1dc71a91 2723................................................
b181d57f
BF
2724 o--o--o--o--a--b <-- old head of the branch
2725 \
2726 o--o--o <-- new head of the branch
1dc71a91 2727................................................
b181d57f 2728
467c0197 2729In this case, "git-fetch" will fail, and print out a warning.
b181d57f
BF
2730
2731In that case, you can still force git to update to the new head, as
2732described in the following section. However, note that in the
2733situation above this may mean losing the commits labeled "a" and "b",
2734unless you've already created a reference of your own pointing to
2735them.
2736
e34caace 2737[[forcing-fetch]]
b1889c36 2738Forcing git-fetch to do non-fast-forward updates
b181d57f
BF
2739------------------------------------------------
2740
2741If git fetch fails because the new head of a branch is not a
2742descendant of the old head, you may force the update with:
2743
2744-------------------------------------------------
2745$ git fetch git://example.com/proj.git +master:refs/remotes/example/master
2746-------------------------------------------------
2747
c64415e2
BF
2748Note the addition of the "+" sign. Alternatively, you can use the "-f"
2749flag to force updates of all the fetched branches, as in:
2750
2751-------------------------------------------------
2752$ git fetch -f origin
2753-------------------------------------------------
2754
2755Be aware that commits that the old version of example/master pointed at
2756may be lost, as we saw in the previous section.
b181d57f 2757
e34caace 2758[[remote-branch-configuration]]
b181d57f
BF
2759Configuring remote branches
2760---------------------------
2761
2762We saw above that "origin" is just a shortcut to refer to the
79c96c57 2763repository that you originally cloned from. This information is
b181d57f 2764stored in git configuration variables, which you can see using
5162e697 2765linkgit:git-config[1]:
b181d57f
BF
2766
2767-------------------------------------------------
9d13bda3 2768$ git config -l
b181d57f
BF
2769core.repositoryformatversion=0
2770core.filemode=true
2771core.logallrefupdates=true
2772remote.origin.url=git://git.kernel.org/pub/scm/git/git.git
2773remote.origin.fetch=+refs/heads/*:refs/remotes/origin/*
2774branch.master.remote=origin
2775branch.master.merge=refs/heads/master
2776-------------------------------------------------
2777
2778If there are other repositories that you also use frequently, you can
2779create similar configuration options to save typing; for example,
2780after
2781
2782-------------------------------------------------
9d13bda3 2783$ git config remote.example.url git://example.com/proj.git
b181d57f
BF
2784-------------------------------------------------
2785
2786then the following two commands will do the same thing:
2787
2788-------------------------------------------------
2789$ git fetch git://example.com/proj.git master:refs/remotes/example/master
2790$ git fetch example master:refs/remotes/example/master
2791-------------------------------------------------
2792
2793Even better, if you add one more option:
2794
2795-------------------------------------------------
9d13bda3 2796$ git config remote.example.fetch master:refs/remotes/example/master
b181d57f
BF
2797-------------------------------------------------
2798
2799then the following commands will all do the same thing:
2800
2801-------------------------------------------------
52c80037
BF
2802$ git fetch git://example.com/proj.git master:refs/remotes/example/master
2803$ git fetch example master:refs/remotes/example/master
b181d57f
BF
2804$ git fetch example
2805-------------------------------------------------
2806
2807You can also add a "+" to force the update each time:
2808
2809-------------------------------------------------
9d13bda3 2810$ git config remote.example.fetch +master:ref/remotes/example/master
b181d57f
BF
2811-------------------------------------------------
2812
b1889c36 2813Don't do this unless you're sure you won't mind "git-fetch" possibly
b181d57f
BF
2814throwing away commits on mybranch.
2815
2816Also note that all of the above configuration can be performed by
2817directly editing the file .git/config instead of using
5162e697 2818linkgit:git-config[1].
b181d57f 2819
5162e697 2820See linkgit:git-config[1] for more details on the configuration
b181d57f 2821options mentioned above.
d19fbc3c 2822
d19fbc3c 2823
036f8199
BF
2824[[git-concepts]]
2825Git concepts
2826============
d19fbc3c 2827
036f8199
BF
2828Git is built on a small number of simple but powerful ideas. While it
2829is possible to get things done without understanding them, you will find
2830git much more intuitive if you do.
2831
2832We start with the most important, the <<def_object_database,object
2833database>> and the <<def_index,index>>.
b181d57f 2834
e34caace 2835[[the-object-database]]
b181d57f
BF
2836The Object Database
2837-------------------
2838
1bbf1c79
BF
2839
2840We already saw in <<understanding-commits>> that all commits are stored
2841under a 40-digit "object name". In fact, all the information needed to
2842represent the history of a project is stored in objects with such names.
2843In each case the name is calculated by taking the SHA1 hash of the
2844contents of the object. The SHA1 hash is a cryptographic hash function.
2845What that means to us is that it is impossible to find two different
2846objects with the same name. This has a number of advantages; among
2847others:
2848
2849- Git can quickly determine whether two objects are identical or not,
2850 just by comparing names.
06ada152 2851- Since object names are computed the same way in every repository, the
1bbf1c79
BF
2852 same content stored in two repositories will always be stored under
2853 the same name.
2854- Git can detect errors when it reads an object, by checking that the
2855 object's name is still the SHA1 hash of its contents.
2856
2857(See <<object-details>> for the details of the object formatting and
2858SHA1 calculation.)
2859
2860There are four different types of objects: "blob", "tree", "commit", and
2861"tag".
2862
2863- A <<def_blob_object,"blob" object>> is used to store file data.
2864- A <<def_tree_object,"tree" object>> is an object that ties one or more
2865 "blob" objects into a directory structure. In addition, a tree object
2866 can refer to other tree objects, thus creating a directory hierarchy.
2867- A <<def_commit_object,"commit" object>> ties such directory hierarchies
2ef8ac1b 2868 together into a <<def_DAG,directed acyclic graph>> of revisions--each
1bbf1c79
BF
2869 commit contains the object name of exactly one tree designating the
2870 directory hierarchy at the time of the commit. In addition, a commit
2871 refers to "parent" commit objects that describe the history of how we
2872 arrived at that directory hierarchy.
2873- A <<def_tag_object,"tag" object>> symbolically identifies and can be
2874 used to sign other objects. It contains the object name and type of
2875 another object, a symbolic name (of course!) and, optionally, a
2876 signature.
b181d57f 2877
b181d57f
BF
2878The object types in some more detail:
2879
513d419c
BF
2880[[commit-object]]
2881Commit Object
2882~~~~~~~~~~~~~
b181d57f 2883
1bbf1c79
BF
2884The "commit" object links a physical state of a tree with a description
2885of how we got there and why. Use the --pretty=raw option to
5162e697 2886linkgit:git-show[1] or linkgit:git-log[1] to examine your favorite
1bbf1c79
BF
2887commit:
2888
2889------------------------------------------------
2890$ git show -s --pretty=raw 2be7fcb476
2891commit 2be7fcb4764f2dbcee52635b91fedb1b3dcf7ab4
2892tree fb3a8bdd0ceddd019615af4d57a53f43d8cee2bf
2893parent 257a84d9d02e90447b149af58b271c19405edb6a
2894author Dave Watson <dwatson@mimvista.com> 1187576872 -0400
2895committer Junio C Hamano <gitster@pobox.com> 1187591163 -0700
2896
2897 Fix misspelling of 'suppress' in docs
2898
2899 Signed-off-by: Junio C Hamano <gitster@pobox.com>
2900------------------------------------------------
2901
2902As you can see, a commit is defined by:
2903
2904- a tree: The SHA1 name of a tree object (as defined below), representing
2905 the contents of a directory at a certain point in time.
2906- parent(s): The SHA1 name of some number of commits which represent the
9e5d87d4 2907 immediately previous step(s) in the history of the project. The
1bbf1c79
BF
2908 example above has one parent; merge commits may have more than
2909 one. A commit with no parents is called a "root" commit, and
2910 represents the initial revision of a project. Each project must have
2911 at least one root. A project can also have multiple roots, though
2912 that isn't common (or necessarily a good idea).
2913- an author: The name of the person responsible for this change, together
2914 with its date.
2915- a committer: The name of the person who actually created the commit,
2916 with the date it was done. This may be different from the author, for
2917 example, if the author was someone who wrote a patch and emailed it
2918 to the person who used it to create the commit.
2919- a comment describing this commit.
2920
2921Note that a commit does not itself contain any information about what
2922actually changed; all changes are calculated by comparing the contents
2923of the tree referred to by this commit with the trees associated with
2924its parents. In particular, git does not attempt to record file renames
2925explicitly, though it can identify cases where the existence of the same
2926file data at changing paths suggests a rename. (See, for example, the
5162e697 2927-M option to linkgit:git-diff[1]).
1bbf1c79 2928
5162e697 2929A commit is usually created by linkgit:git-commit[1], which creates a
1bbf1c79
BF
2930commit whose parent is normally the current HEAD, and whose tree is
2931taken from the content currently stored in the index.
b181d57f 2932
e34caace 2933[[tree-object]]
b181d57f 2934Tree Object
971aa71f 2935~~~~~~~~~~~
b181d57f 2936
5162e697
DM
2937The ever-versatile linkgit:git-show[1] command can also be used to
2938examine tree objects, but linkgit:git-ls-tree[1] will give you more
1bbf1c79
BF
2939details:
2940
2941------------------------------------------------
2942$ git ls-tree fb3a8bdd0ce
2943100644 blob 63c918c667fa005ff12ad89437f2fdc80926e21c .gitignore
2944100644 blob 5529b198e8d14decbe4ad99db3f7fb632de0439d .mailmap
2945100644 blob 6ff87c4664981e4397625791c8ea3bbb5f2279a3 COPYING
2946040000 tree 2fb783e477100ce076f6bf57e4a6f026013dc745 Documentation
2947100755 blob 3c0032cec592a765692234f1cba47dfdcc3a9200 GIT-VERSION-GEN
2948100644 blob 289b046a443c0647624607d471289b2c7dcd470b INSTALL
2949100644 blob 4eb463797adc693dc168b926b6932ff53f17d0b1 Makefile
2950100644 blob 548142c327a6790ff8821d67c2ee1eff7a656b52 README
2951...
2952------------------------------------------------
2953
2954As you can see, a tree object contains a list of entries, each with a
2955mode, object type, SHA1 name, and name, sorted by name. It represents
2956the contents of a single directory tree.
2957
2958The object type may be a blob, representing the contents of a file, or
2959another tree, representing the contents of a subdirectory. Since trees
2960and blobs, like all other objects, are named by the SHA1 hash of their
2961contents, two trees have the same SHA1 name if and only if their
2962contents (including, recursively, the contents of all subdirectories)
2963are identical. This allows git to quickly determine the differences
2964between two related tree objects, since it can ignore any entries with
2965identical object names.
2966
2967(Note: in the presence of submodules, trees may also have commits as
6dd14366 2968entries. See <<submodules>> for documentation.)
1bbf1c79
BF
2969
2970Note that the files all have mode 644 or 755: git actually only pays
2971attention to the executable bit.
b181d57f 2972
513d419c
BF
2973[[blob-object]]
2974Blob Object
2975~~~~~~~~~~~
b181d57f 2976
5162e697 2977You can use linkgit:git-show[1] to examine the contents of a blob; take,
1bbf1c79 2978for example, the blob in the entry for "COPYING" from the tree above:
b181d57f 2979
1bbf1c79
BF
2980------------------------------------------------
2981$ git show 6ff87c4664
2982
2983 Note that the only valid version of the GPL as far as this project
2984 is concerned is _this_ particular version of the license (ie v2, not
2985 v2.2 or v3.x or whatever), unless explicitly otherwise stated.
2986...
2987------------------------------------------------
b181d57f 2988
1bbf1c79
BF
2989A "blob" object is nothing but a binary blob of data. It doesn't refer
2990to anything else or have attributes of any kind.
2991
2992Since the blob is entirely defined by its data, if two files in a
2993directory tree (or in multiple different versions of the repository)
2994have the same contents, they will share the same blob object. The object
2995is totally independent of its location in the directory tree, and
2996renaming a file does not change the object that file is associated with.
2997
2998Note that any tree or blob object can be examined using
5162e697 2999linkgit:git-show[1] with the <revision>:<path> syntax. This can
1bbf1c79
BF
3000sometimes be useful for browsing the contents of a tree that is not
3001currently checked out.
b181d57f 3002
e34caace 3003[[trust]]
b181d57f 3004Trust
971aa71f 3005~~~~~
b181d57f 3006
1bbf1c79
BF
3007If you receive the SHA1 name of a blob from one source, and its contents
3008from another (possibly untrusted) source, you can still trust that those
3009contents are correct as long as the SHA1 name agrees. This is because
3010the SHA1 is designed so that it is infeasible to find different contents
3011that produce the same hash.
b181d57f 3012
1bbf1c79
BF
3013Similarly, you need only trust the SHA1 name of a top-level tree object
3014to trust the contents of the entire directory that it refers to, and if
3015you receive the SHA1 name of a commit from a trusted source, then you
3016can easily verify the entire history of commits reachable through
3017parents of that commit, and all of those contents of the trees referred
3018to by those commits.
b181d57f
BF
3019
3020So to introduce some real trust in the system, the only thing you need
3021to do is to digitally sign just 'one' special note, which includes the
3022name of a top-level commit. Your digital signature shows others
3023that you trust that commit, and the immutability of the history of
3024commits tells others that they can trust the whole history.
3025
3026In other words, you can easily validate a whole archive by just
3027sending out a single email that tells the people the name (SHA1 hash)
3028of the top commit, and digitally sign that email using something
3029like GPG/PGP.
3030
3031To assist in this, git also provides the tag object...
3032
e34caace 3033[[tag-object]]
b181d57f 3034Tag Object
971aa71f 3035~~~~~~~~~~
b181d57f 3036
1bbf1c79
BF
3037A tag object contains an object, object type, tag name, the name of the
3038person ("tagger") who created the tag, and a message, which may contain
5162e697 3039a signature, as can be seen using the linkgit:git-cat-file[1]:
b181d57f 3040
1bbf1c79
BF
3041------------------------------------------------
3042$ git cat-file tag v1.5.0
3043object 437b1b20df4b356c9342dac8d38849f24ef44f27
3044type commit
3045tag v1.5.0
3046tagger Junio C Hamano <junkio@cox.net> 1171411200 +0000
3047
3048GIT 1.5.0
3049-----BEGIN PGP SIGNATURE-----
3050Version: GnuPG v1.4.6 (GNU/Linux)
3051
3052iD8DBQBF0lGqwMbZpPMRm5oRAuRiAJ9ohBLd7s2kqjkKlq1qqC57SbnmzQCdG4ui
3053nLE/L9aUXdWeTFPron96DLA=
3054=2E+0
3055-----END PGP SIGNATURE-----
3056------------------------------------------------
b181d57f 3057
5162e697
DM
3058See the linkgit:git-tag[1] command to learn how to create and verify tag
3059objects. (Note that linkgit:git-tag[1] can also be used to create
1bbf1c79 3060"lightweight tags", which are not tag objects at all, but just simple
fc74ecc1 3061references whose names begin with "refs/tags/").
b181d57f 3062
09eff7b0
BF
3063[[pack-files]]
3064How git stores objects efficiently: pack files
3065~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3066
9644ffdd
BF
3067Newly created objects are initially created in a file named after the
3068object's SHA1 hash (stored in .git/objects).
09eff7b0
BF
3069
3070Unfortunately this system becomes inefficient once a project has a
3071lot of objects. Try this on an old project:
3072
3073------------------------------------------------
3074$ git count-objects
30756930 objects, 47620 kilobytes
3076------------------------------------------------
3077
3078The first number is the number of objects which are kept in
3079individual files. The second is the amount of space taken up by
3080those "loose" objects.
3081
3082You can save space and make git faster by moving these loose objects in
3083to a "pack file", which stores a group of objects in an efficient
3084compressed format; the details of how pack files are formatted can be
3085found in link:technical/pack-format.txt[technical/pack-format.txt].
3086
3087To put the loose objects into a pack, just run git repack:
3088
3089------------------------------------------------
3090$ git repack
3091Generating pack...
3092Done counting 6020 objects.
3093Deltifying 6020 objects.
3094 100% (6020/6020) done
3095Writing 6020 objects.
3096 100% (6020/6020) done
3097Total 6020, written 6020 (delta 4070), reused 0 (delta 0)
3098Pack pack-3e54ad29d5b2e05838c75df582c65257b8d08e1c created.
3099------------------------------------------------
3100
3101You can then run
3102
3103------------------------------------------------
3104$ git prune
3105------------------------------------------------
3106
3107to remove any of the "loose" objects that are now contained in the
3108pack. This will also remove any unreferenced objects (which may be
467c0197 3109created when, for example, you use "git-reset" to remove a commit).
09eff7b0
BF
3110You can verify that the loose objects are gone by looking at the
3111.git/objects directory or by running
3112
3113------------------------------------------------
3114$ git count-objects
31150 objects, 0 kilobytes
3116------------------------------------------------
3117
3118Although the object files are gone, any commands that refer to those
3119objects will work exactly as they did before.
3120
5162e697 3121The linkgit:git-gc[1] command performs packing, pruning, and more for
09eff7b0
BF
3122you, so is normally the only high-level command you need.
3123
3124[[dangling-objects]]
3125Dangling objects
3126~~~~~~~~~~~~~~~~
3127
5162e697 3128The linkgit:git-fsck[1] command will sometimes complain about dangling
09eff7b0
BF
3129objects. They are not a problem.
3130
3131The most common cause of dangling objects is that you've rebased a
3132branch, or you have pulled from somebody else who rebased a branch--see
3133<<cleaning-up-history>>. In that case, the old head of the original
3134branch still exists, as does everything it pointed to. The branch
3135pointer itself just doesn't, since you replaced it with another one.
3136
3137There are also other situations that cause dangling objects. For
467c0197 3138example, a "dangling blob" may arise because you did a "git-add" of a
09eff7b0
BF
3139file, but then, before you actually committed it and made it part of the
3140bigger picture, you changed something else in that file and committed
2ef8ac1b 3141that *updated* thing--the old state that you added originally ends up
09eff7b0
BF
3142not being pointed to by any commit or tree, so it's now a dangling blob
3143object.
3144
3145Similarly, when the "recursive" merge strategy runs, and finds that
3146there are criss-cross merges and thus more than one merge base (which is
3147fairly unusual, but it does happen), it will generate one temporary
3148midway tree (or possibly even more, if you had lots of criss-crossing
3149merges and more than two merge bases) as a temporary internal merge
3150base, and again, those are real objects, but the end result will not end
3151up pointing to them, so they end up "dangling" in your repository.
3152
3153Generally, dangling objects aren't anything to worry about. They can
3154even be very useful: if you screw something up, the dangling objects can
3155be how you recover your old tree (say, you did a rebase, and realized
2ef8ac1b 3156that you really didn't want to--you can look at what dangling objects
09eff7b0
BF
3157you have, and decide to reset your head to some old dangling state).
3158
3159For commits, you can just use:
3160
3161------------------------------------------------
3162$ gitk <dangling-commit-sha-goes-here> --not --all
3163------------------------------------------------
3164
3165This asks for all the history reachable from the given commit but not
3166from any branch, tag, or other reference. If you decide it's something
3167you want, you can always create a new reference to it, e.g.,
3168
3169------------------------------------------------
3170$ git branch recovered-branch <dangling-commit-sha-goes-here>
3171------------------------------------------------
3172
3173For blobs and trees, you can't do the same, but you can still examine
3174them. You can just do
3175
3176------------------------------------------------
3177$ git show <dangling-blob/tree-sha-goes-here>
3178------------------------------------------------
3179
3180to show what the contents of the blob were (or, for a tree, basically
3181what the "ls" for that directory was), and that may give you some idea
3182of what the operation was that left that dangling object.
3183
3184Usually, dangling blobs and trees aren't very interesting. They're
3185almost always the result of either being a half-way mergebase (the blob
3186will often even have the conflict markers from a merge in it, if you
3187have had conflicting merges that you fixed up by hand), or simply
467c0197 3188because you interrupted a "git-fetch" with ^C or something like that,
09eff7b0
BF
3189leaving _some_ of the new objects in the object database, but just
3190dangling and useless.
3191
3192Anyway, once you are sure that you're not interested in any dangling
3193state, you can just prune all unreachable objects:
3194
3195------------------------------------------------
3196$ git prune
3197------------------------------------------------
3198
3199and they'll be gone. But you should only run "git prune" on a quiescent
2ef8ac1b 3200repository--it's kind of like doing a filesystem fsck recovery: you
09eff7b0
BF
3201don't want to do that while the filesystem is mounted.
3202
2ef8ac1b 3203(The same is true of "git-fsck" itself, btw, but since
09eff7b0
BF
3204git-fsck never actually *changes* the repository, it just reports
3205on what it found, git-fsck itself is never "dangerous" to run.
3206Running it while somebody is actually changing the repository can cause
3207confusing and scary messages, but it won't actually do anything bad. In
3208contrast, running "git prune" while somebody is actively changing the
3209repository is a *BAD* idea).
b181d57f 3210
1cdade2c
BF
3211[[recovering-from-repository-corruption]]
3212Recovering from repository corruption
3213~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3214
3215By design, git treats data trusted to it with caution. However, even in
3216the absence of bugs in git itself, it is still possible that hardware or
3217operating system errors could corrupt data.
3218
3219The first defense against such problems is backups. You can back up a
3220git directory using clone, or just using cp, tar, or any other backup
3221mechanism.
3222
3223As a last resort, you can search for the corrupted objects and attempt
3224to replace them by hand. Back up your repository before attempting this
3225in case you corrupt things even more in the process.
3226
3227We'll assume that the problem is a single missing or corrupted blob,
9e5d87d4 3228which is sometimes a solvable problem. (Recovering missing trees and
1cdade2c
BF
3229especially commits is *much* harder).
3230
3231Before starting, verify that there is corruption, and figure out where
5162e697 3232it is with linkgit:git-fsck[1]; this may be time-consuming.
1cdade2c
BF
3233
3234Assume the output looks like this:
3235
3236------------------------------------------------
b1889c36 3237$ git fsck --full
1cdade2c
BF
3238broken link from tree 2d9263c6d23595e7cb2a21e5ebbb53655278dff8
3239 to blob 4b9458b3786228369c63936db65827de3cc06200
3240missing blob 4b9458b3786228369c63936db65827de3cc06200
3241------------------------------------------------
3242
3243(Typically there will be some "dangling object" messages too, but they
3244aren't interesting.)
3245
3246Now you know that blob 4b9458b3 is missing, and that the tree 2d9263c6
3247points to it. If you could find just one copy of that missing blob
3248object, possibly in some other repository, you could move it into
3249.git/objects/4b/9458b3... and be done. Suppose you can't. You can
5162e697 3250still examine the tree that pointed to it with linkgit:git-ls-tree[1],
1cdade2c
BF
3251which might output something like:
3252
3253------------------------------------------------
3254$ git ls-tree 2d9263c6d23595e7cb2a21e5ebbb53655278dff8
3255100644 blob 8d14531846b95bfa3564b58ccfb7913a034323b8 .gitignore
3256100644 blob ebf9bf84da0aab5ed944264a5db2a65fe3a3e883 .mailmap
3257100644 blob ca442d313d86dc67e0a2e5d584b465bd382cbf5c COPYING
3258...
3259100644 blob 4b9458b3786228369c63936db65827de3cc06200 myfile
3260...
3261------------------------------------------------
3262
3263So now you know that the missing blob was the data for a file named
3264"myfile". And chances are you can also identify the directory--let's
3265say it's in "somedirectory". If you're lucky the missing copy might be
3266the same as the copy you have checked out in your working tree at
3267"somedirectory/myfile"; you can test whether that's right with
5162e697 3268linkgit:git-hash-object[1]:
1cdade2c
BF
3269
3270------------------------------------------------
3271$ git hash-object -w somedirectory/myfile
3272------------------------------------------------
3273
3274which will create and store a blob object with the contents of
3275somedirectory/myfile, and output the sha1 of that object. if you're
3276extremely lucky it might be 4b9458b3786228369c63936db65827de3cc06200, in
3277which case you've guessed right, and the corruption is fixed!
3278
3279Otherwise, you need more information. How do you tell which version of
3280the file has been lost?
3281
3282The easiest way to do this is with:
3283
3284------------------------------------------------
3285$ git log --raw --all --full-history -- somedirectory/myfile
3286------------------------------------------------
3287
3288Because you're asking for raw output, you'll now get something like
3289
3290------------------------------------------------
3291commit abc
3292Author:
3293Date:
3294...
3295:100644 100644 4b9458b... newsha... M somedirectory/myfile
3296
3297
3298commit xyz
3299Author:
3300Date:
3301
3302...
3303:100644 100644 oldsha... 4b9458b... M somedirectory/myfile
3304------------------------------------------------
3305
3306This tells you that the immediately preceding version of the file was
3307"newsha", and that the immediately following version was "oldsha".
3308You also know the commit messages that went with the change from oldsha
3309to 4b9458b and with the change from 4b9458b to newsha.
3310
3311If you've been committing small enough changes, you may now have a good
3312shot at reconstructing the contents of the in-between state 4b9458b.
3313
3314If you can do that, you can now recreate the missing object with
3315
3316------------------------------------------------
3317$ git hash-object -w <recreated-file>
3318------------------------------------------------
3319
3320and your repository is good again!
3321
1ab58e8d 3322(Btw, you could have ignored the fsck, and started with doing a
1cdade2c
BF
3323
3324------------------------------------------------
3325$ git log --raw --all
3326------------------------------------------------
3327
1ab58e8d
JH
3328and just looked for the sha of the missing object (4b9458b..) in that
3329whole thing. It's up to you - git does *have* a lot of information, it is
1cdade2c
BF
3330just missing one particular blob version.
3331
e34caace 3332[[the-index]]
1c097891
BF
3333The index
3334-----------
3335
3336The index is a binary file (generally kept in .git/index) containing a
3337sorted list of path names, each with permissions and the SHA1 of a blob
5162e697 3338object; linkgit:git-ls-files[1] can show you the contents of the index:
b181d57f 3339
1c097891
BF
3340-------------------------------------------------
3341$ git ls-files --stage
3342100644 63c918c667fa005ff12ad89437f2fdc80926e21c 0 .gitignore
3343100644 5529b198e8d14decbe4ad99db3f7fb632de0439d 0 .mailmap
3344100644 6ff87c4664981e4397625791c8ea3bbb5f2279a3 0 COPYING
3345100644 a37b2152bd26be2c2289e1f57a292534a51a93c7 0 Documentation/.gitignore
3346100644 fbefe9a45b00a54b58d94d06eca48b03d40a50e0 0 Documentation/Makefile
3347...
3348100644 2511aef8d89ab52be5ec6a5e46236b4b6bcd07ea 0 xdiff/xtypes.h
3349100644 2ade97b2574a9f77e7ae4002a4e07a6a38e46d07 0 xdiff/xutils.c
3350100644 d5de8292e05e7c36c4b68857c1cf9855e3d2f70a 0 xdiff/xutils.h
3351-------------------------------------------------
3352
3353Note that in older documentation you may see the index called the
3354"current directory cache" or just the "cache". It has three important
3355properties:
3356
33571. The index contains all the information necessary to generate a single
3358(uniquely determined) tree object.
3359+
5162e697 3360For example, running linkgit:git-commit[1] generates this tree object
1c097891
BF
3361from the index, stores it in the object database, and uses it as the
3362tree object associated with the new commit.
3363
33642. The index enables fast comparisons between the tree object it defines
3365and the working tree.
3366+
3367It does this by storing some additional data for each entry (such as
3368the last modified time). This data is not displayed above, and is not
3369stored in the created tree object, but it can be used to determine
3370quickly which files in the working directory differ from what was
3371stored in the index, and thus save git from having to read all of the
3372data from such files to look for changes.
3373
33743. It can efficiently represent information about merge conflicts
3375between different tree objects, allowing each pathname to be
b181d57f 3376associated with sufficient information about the trees involved that
1c097891
BF
3377you can create a three-way merge between them.
3378+
3379We saw in <<confli