user-manual: remove some git-foo usage
[git/git.git] / Documentation / user-manual.txt
1 Git User's Manual (for version 1.5.3 or newer)
2 ______________________________________________
3
4
5 Git is a fast distributed revision control system.
6
7 This manual is designed to be readable by someone with basic UNIX
8 command-line skills, but no previous knowledge of git.
9
10 <<repositories-and-branches>> and <<exploring-git-history>> explain how
11 to fetch and study a project using git--read these chapters to learn how
12 to build and test a particular version of a software project, search for
13 regressions, and so on.
14
15 People needing to do actual development will also want to read
16 <<Developing-With-git>> and <<sharing-development>>.
17
18 Further chapters cover more specialized topics.
19
20 Comprehensive reference documentation is available through the man
21 pages, or linkgit:git-help[1] command. For example, for the command
22 "git clone <repo>", you can either use:
23
24 ------------------------------------------------
25 $ man git-clone
26 ------------------------------------------------
27
28 or:
29
30 ------------------------------------------------
31 $ git help clone
32 ------------------------------------------------
33
34 With the latter, you can use the manual viewer of your choice; see
35 linkgit:git-help[1] for more information.
36
37 See also <<git-quick-start>> for a brief overview of git commands,
38 without any explanation.
39
40 Finally, see <<todo>> for ways that you can help make this manual more
41 complete.
42
43
44 [[repositories-and-branches]]
45 Repositories and Branches
46 =========================
47
48 [[how-to-get-a-git-repository]]
49 How to get a git repository
50 ---------------------------
51
52 It will be useful to have a git repository to experiment with as you
53 read this manual.
54
55 The best way to get one is by using the linkgit:git-clone[1] command to
56 download a copy of an existing repository. If you don't already have a
57 project in mind, here are some interesting examples:
58
59 ------------------------------------------------
60 # git itself (approx. 10MB download):
61 $ git clone git://git.kernel.org/pub/scm/git/git.git
62 # the Linux kernel (approx. 150MB download):
63 $ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
64 ------------------------------------------------
65
66 The initial clone may be time-consuming for a large project, but you
67 will only need to clone once.
68
69 The clone command creates a new directory named after the project ("git"
70 or "linux-2.6" in the examples above). After you cd into this
71 directory, you will see that it contains a copy of the project files,
72 called the <<def_working_tree,working tree>>, together with a special
73 top-level directory named ".git", which contains all the information
74 about the history of the project.
75
76 [[how-to-check-out]]
77 How to check out a different version of a project
78 -------------------------------------------------
79
80 Git is best thought of as a tool for storing the history of a collection
81 of files. It stores the history as a compressed collection of
82 interrelated snapshots of the project's contents. In git each such
83 version is called a <<def_commit,commit>>.
84
85 Those snapshots aren't necessarily all arranged in a single line from
86 oldest to newest; instead, work may simultaneously proceed along
87 parallel lines of development, called <<def_branch,branches>>, which may
88 merge and diverge.
89
90 A single git repository can track development on multiple branches. It
91 does this by keeping a list of <<def_head,heads>> which reference the
92 latest commit on each branch; the linkgit:git-branch[1] command shows
93 you the list of branch heads:
94
95 ------------------------------------------------
96 $ git branch
97 * master
98 ------------------------------------------------
99
100 A freshly cloned repository contains a single branch head, by default
101 named "master", with the working directory initialized to the state of
102 the project referred to by that branch head.
103
104 Most projects also use <<def_tag,tags>>. Tags, like heads, are
105 references into the project's history, and can be listed using the
106 linkgit:git-tag[1] command:
107
108 ------------------------------------------------
109 $ git tag -l
110 v2.6.11
111 v2.6.11-tree
112 v2.6.12
113 v2.6.12-rc2
114 v2.6.12-rc3
115 v2.6.12-rc4
116 v2.6.12-rc5
117 v2.6.12-rc6
118 v2.6.13
119 ...
120 ------------------------------------------------
121
122 Tags are expected to always point at the same version of a project,
123 while heads are expected to advance as development progresses.
124
125 Create a new branch head pointing to one of these versions and check it
126 out using linkgit:git-checkout[1]:
127
128 ------------------------------------------------
129 $ git checkout -b new v2.6.13
130 ------------------------------------------------
131
132 The working directory then reflects the contents that the project had
133 when it was tagged v2.6.13, and linkgit:git-branch[1] shows two
134 branches, with an asterisk marking the currently checked-out branch:
135
136 ------------------------------------------------
137 $ git branch
138 master
139 * new
140 ------------------------------------------------
141
142 If you decide that you'd rather see version 2.6.17, you can modify
143 the current branch to point at v2.6.17 instead, with
144
145 ------------------------------------------------
146 $ git reset --hard v2.6.17
147 ------------------------------------------------
148
149 Note that if the current branch head was your only reference to a
150 particular point in history, then resetting that branch may leave you
151 with no way to find the history it used to point to; so use this command
152 carefully.
153
154 [[understanding-commits]]
155 Understanding History: Commits
156 ------------------------------
157
158 Every change in the history of a project is represented by a commit.
159 The linkgit:git-show[1] command shows the most recent commit on the
160 current branch:
161
162 ------------------------------------------------
163 $ git show
164 commit 17cf781661e6d38f737f15f53ab552f1e95960d7
165 Author: Linus Torvalds <torvalds@ppc970.osdl.org.(none)>
166 Date: Tue Apr 19 14:11:06 2005 -0700
167
168 Remove duplicate getenv(DB_ENVIRONMENT) call
169
170 Noted by Tony Luck.
171
172 diff --git a/init-db.c b/init-db.c
173 index 65898fa..b002dc6 100644
174 --- a/init-db.c
175 +++ b/init-db.c
176 @@ -7,7 +7,7 @@
177
178 int main(int argc, char **argv)
179 {
180 - char *sha1_dir = getenv(DB_ENVIRONMENT), *path;
181 + char *sha1_dir, *path;
182 int len, i;
183
184 if (mkdir(".git", 0755) < 0) {
185 ------------------------------------------------
186
187 As you can see, a commit shows who made the latest change, what they
188 did, and why.
189
190 Every commit has a 40-hexdigit id, sometimes called the "object name" or the
191 "SHA1 id", shown on the first line of the "git show" output. You can usually
192 refer to a commit by a shorter name, such as a tag or a branch name, but this
193 longer name can also be useful. Most importantly, it is a globally unique
194 name for this commit: so if you tell somebody else the object name (for
195 example in email), then you are guaranteed that name will refer to the same
196 commit in their repository that it does in yours (assuming their repository
197 has that commit at all). Since the object name is computed as a hash over the
198 contents of the commit, you are guaranteed that the commit can never change
199 without its name also changing.
200
201 In fact, in <<git-concepts>> we shall see that everything stored in git
202 history, including file data and directory contents, is stored in an object
203 with a name that is a hash of its contents.
204
205 [[understanding-reachability]]
206 Understanding history: commits, parents, and reachability
207 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
208
209 Every commit (except the very first commit in a project) also has a
210 parent commit which shows what happened before this commit.
211 Following the chain of parents will eventually take you back to the
212 beginning of the project.
213
214 However, the commits do not form a simple list; git allows lines of
215 development to diverge and then reconverge, and the point where two
216 lines of development reconverge is called a "merge". The commit
217 representing a merge can therefore have more than one parent, with
218 each parent representing the most recent commit on one of the lines
219 of development leading to that point.
220
221 The best way to see how this works is using the linkgit:gitk[1]
222 command; running gitk now on a git repository and looking for merge
223 commits will help understand how the git organizes history.
224
225 In the following, we say that commit X is "reachable" from commit Y
226 if commit X is an ancestor of commit Y. Equivalently, you could say
227 that Y is a descendant of X, or that there is a chain of parents
228 leading from commit Y to commit X.
229
230 [[history-diagrams]]
231 Understanding history: History diagrams
232 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
233
234 We will sometimes represent git history using diagrams like the one
235 below. Commits are shown as "o", and the links between them with
236 lines drawn with - / and \. Time goes left to right:
237
238
239 ................................................
240 o--o--o <-- Branch A
241 /
242 o--o--o <-- master
243 \
244 o--o--o <-- Branch B
245 ................................................
246
247 If we need to talk about a particular commit, the character "o" may
248 be replaced with another letter or number.
249
250 [[what-is-a-branch]]
251 Understanding history: What is a branch?
252 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
253
254 When we need to be precise, we will use the word "branch" to mean a line
255 of development, and "branch head" (or just "head") to mean a reference
256 to the most recent commit on a branch. In the example above, the branch
257 head named "A" is a pointer to one particular commit, but we refer to
258 the line of three commits leading up to that point as all being part of
259 "branch A".
260
261 However, when no confusion will result, we often just use the term
262 "branch" both for branches and for branch heads.
263
264 [[manipulating-branches]]
265 Manipulating branches
266 ---------------------
267
268 Creating, deleting, and modifying branches is quick and easy; here's
269 a summary of the commands:
270
271 git branch::
272 list all branches
273 git branch <branch>::
274 create a new branch named <branch>, referencing the same
275 point in history as the current branch
276 git branch <branch> <start-point>::
277 create a new branch named <branch>, referencing
278 <start-point>, which may be specified any way you like,
279 including using a branch name or a tag name
280 git branch -d <branch>::
281 delete the branch <branch>; if the branch you are deleting
282 points to a commit which is not reachable from the current
283 branch, this command will fail with a warning.
284 git branch -D <branch>::
285 even if the branch points to a commit not reachable
286 from the current branch, you may know that that commit
287 is still reachable from some other branch or tag. In that
288 case it is safe to use this command to force git to delete
289 the branch.
290 git checkout <branch>::
291 make the current branch <branch>, updating the working
292 directory to reflect the version referenced by <branch>
293 git checkout -b <new> <start-point>::
294 create a new branch <new> referencing <start-point>, and
295 check it out.
296
297 The special symbol "HEAD" can always be used to refer to the current
298 branch. In fact, git uses a file named "HEAD" in the .git directory to
299 remember which branch is current:
300
301 ------------------------------------------------
302 $ cat .git/HEAD
303 ref: refs/heads/master
304 ------------------------------------------------
305
306 [[detached-head]]
307 Examining an old version without creating a new branch
308 ------------------------------------------------------
309
310 The `git checkout` command normally expects a branch head, but will also
311 accept an arbitrary commit; for example, you can check out the commit
312 referenced by a tag:
313
314 ------------------------------------------------
315 $ git checkout v2.6.17
316 Note: moving to "v2.6.17" which isn't a local branch
317 If you want to create a new branch from this checkout, you may do so
318 (now or later) by using -b with the checkout command again. Example:
319 git checkout -b <new_branch_name>
320 HEAD is now at 427abfa... Linux v2.6.17
321 ------------------------------------------------
322
323 The HEAD then refers to the SHA1 of the commit instead of to a branch,
324 and git branch shows that you are no longer on a branch:
325
326 ------------------------------------------------
327 $ cat .git/HEAD
328 427abfa28afedffadfca9dd8b067eb6d36bac53f
329 $ git branch
330 * (no branch)
331 master
332 ------------------------------------------------
333
334 In this case we say that the HEAD is "detached".
335
336 This is an easy way to check out a particular version without having to
337 make up a name for the new branch. You can still create a new branch
338 (or tag) for this version later if you decide to.
339
340 [[examining-remote-branches]]
341 Examining branches from a remote repository
342 -------------------------------------------
343
344 The "master" branch that was created at the time you cloned is a copy
345 of the HEAD in the repository that you cloned from. That repository
346 may also have had other branches, though, and your local repository
347 keeps branches which track each of those remote branches, which you
348 can view using the "-r" option to linkgit:git-branch[1]:
349
350 ------------------------------------------------
351 $ git branch -r
352 origin/HEAD
353 origin/html
354 origin/maint
355 origin/man
356 origin/master
357 origin/next
358 origin/pu
359 origin/todo
360 ------------------------------------------------
361
362 You cannot check out these remote-tracking branches, but you can
363 examine them on a branch of your own, just as you would a tag:
364
365 ------------------------------------------------
366 $ git checkout -b my-todo-copy origin/todo
367 ------------------------------------------------
368
369 Note that the name "origin" is just the name that git uses by default
370 to refer to the repository that you cloned from.
371
372 [[how-git-stores-references]]
373 Naming branches, tags, and other references
374 -------------------------------------------
375
376 Branches, remote-tracking branches, and tags are all references to
377 commits. All references are named with a slash-separated path name
378 starting with "refs"; the names we've been using so far are actually
379 shorthand:
380
381 - The branch "test" is short for "refs/heads/test".
382 - The tag "v2.6.18" is short for "refs/tags/v2.6.18".
383 - "origin/master" is short for "refs/remotes/origin/master".
384
385 The full name is occasionally useful if, for example, there ever
386 exists a tag and a branch with the same name.
387
388 (Newly created refs are actually stored in the .git/refs directory,
389 under the path given by their name. However, for efficiency reasons
390 they may also be packed together in a single file; see
391 linkgit:git-pack-refs[1]).
392
393 As another useful shortcut, the "HEAD" of a repository can be referred
394 to just using the name of that repository. So, for example, "origin"
395 is usually a shortcut for the HEAD branch in the repository "origin".
396
397 For the complete list of paths which git checks for references, and
398 the order it uses to decide which to choose when there are multiple
399 references with the same shorthand name, see the "SPECIFYING
400 REVISIONS" section of linkgit:git-rev-parse[1].
401
402 [[Updating-a-repository-With-git-fetch]]
403 Updating a repository with git fetch
404 ------------------------------------
405
406 Eventually the developer cloned from will do additional work in her
407 repository, creating new commits and advancing the branches to point
408 at the new commits.
409
410 The command "git fetch", with no arguments, will update all of the
411 remote-tracking branches to the latest version found in her
412 repository. It will not touch any of your own branches--not even the
413 "master" branch that was created for you on clone.
414
415 [[fetching-branches]]
416 Fetching branches from other repositories
417 -----------------------------------------
418
419 You can also track branches from repositories other than the one you
420 cloned from, using linkgit:git-remote[1]:
421
422 -------------------------------------------------
423 $ git remote add linux-nfs git://linux-nfs.org/pub/nfs-2.6.git
424 $ git fetch linux-nfs
425 * refs/remotes/linux-nfs/master: storing branch 'master' ...
426 commit: bf81b46
427 -------------------------------------------------
428
429 New remote-tracking branches will be stored under the shorthand name
430 that you gave "git remote add", in this case linux-nfs:
431
432 -------------------------------------------------
433 $ git branch -r
434 linux-nfs/master
435 origin/master
436 -------------------------------------------------
437
438 If you run "git fetch <remote>" later, the tracking branches for the
439 named <remote> will be updated.
440
441 If you examine the file .git/config, you will see that git has added
442 a new stanza:
443
444 -------------------------------------------------
445 $ cat .git/config
446 ...
447 [remote "linux-nfs"]
448 url = git://linux-nfs.org/pub/nfs-2.6.git
449 fetch = +refs/heads/*:refs/remotes/linux-nfs/*
450 ...
451 -------------------------------------------------
452
453 This is what causes git to track the remote's branches; you may modify
454 or delete these configuration options by editing .git/config with a
455 text editor. (See the "CONFIGURATION FILE" section of
456 linkgit:git-config[1] for details.)
457
458 [[exploring-git-history]]
459 Exploring git history
460 =====================
461
462 Git is best thought of as a tool for storing the history of a
463 collection of files. It does this by storing compressed snapshots of
464 the contents of a file hierarchy, together with "commits" which show
465 the relationships between these snapshots.
466
467 Git provides extremely flexible and fast tools for exploring the
468 history of a project.
469
470 We start with one specialized tool that is useful for finding the
471 commit that introduced a bug into a project.
472
473 [[using-bisect]]
474 How to use bisect to find a regression
475 --------------------------------------
476
477 Suppose version 2.6.18 of your project worked, but the version at
478 "master" crashes. Sometimes the best way to find the cause of such a
479 regression is to perform a brute-force search through the project's
480 history to find the particular commit that caused the problem. The
481 linkgit:git-bisect[1] command can help you do this:
482
483 -------------------------------------------------
484 $ git bisect start
485 $ git bisect good v2.6.18
486 $ git bisect bad master
487 Bisecting: 3537 revisions left to test after this
488 [65934a9a028b88e83e2b0f8b36618fe503349f8e] BLOCK: Make USB storage depend on SCSI rather than selecting it [try #6]
489 -------------------------------------------------
490
491 If you run "git branch" at this point, you'll see that git has
492 temporarily moved you in "(no branch)". HEAD is now detached from any
493 branch and points directly to a commit (with commit id 65934...) that
494 is reachable from "master" but not from v2.6.18. Compile and test it,
495 and see whether it crashes. Assume it does crash. Then:
496
497 -------------------------------------------------
498 $ git bisect bad
499 Bisecting: 1769 revisions left to test after this
500 [7eff82c8b1511017ae605f0c99ac275a7e21b867] i2c-core: Drop useless bitmaskings
501 -------------------------------------------------
502
503 checks out an older version. Continue like this, telling git at each
504 stage whether the version it gives you is good or bad, and notice
505 that the number of revisions left to test is cut approximately in
506 half each time.
507
508 After about 13 tests (in this case), it will output the commit id of
509 the guilty commit. You can then examine the commit with
510 linkgit:git-show[1], find out who wrote it, and mail them your bug
511 report with the commit id. Finally, run
512
513 -------------------------------------------------
514 $ git bisect reset
515 -------------------------------------------------
516
517 to return you to the branch you were on before.
518
519 Note that the version which `git bisect` checks out for you at each
520 point is just a suggestion, and you're free to try a different
521 version if you think it would be a good idea. For example,
522 occasionally you may land on a commit that broke something unrelated;
523 run
524
525 -------------------------------------------------
526 $ git bisect visualize
527 -------------------------------------------------
528
529 which will run gitk and label the commit it chose with a marker that
530 says "bisect". Choose a safe-looking commit nearby, note its commit
531 id, and check it out with:
532
533 -------------------------------------------------
534 $ git reset --hard fb47ddb2db...
535 -------------------------------------------------
536
537 then test, run "bisect good" or "bisect bad" as appropriate, and
538 continue.
539
540 Instead of "git bisect visualize" and then "git reset --hard
541 fb47ddb2db...", you might just want to tell git that you want to skip
542 the current commit:
543
544 -------------------------------------------------
545 $ git bisect skip
546 -------------------------------------------------
547
548 In this case, though, git may not eventually be able to tell the first
549 bad one between some first skipped commits and a later bad commit.
550
551 There are also ways to automate the bisecting process if you have a
552 test script that can tell a good from a bad commit. See
553 linkgit:git-bisect[1] for more information about this and other "git
554 bisect" features.
555
556 [[naming-commits]]
557 Naming commits
558 --------------
559
560 We have seen several ways of naming commits already:
561
562 - 40-hexdigit object name
563 - branch name: refers to the commit at the head of the given
564 branch
565 - tag name: refers to the commit pointed to by the given tag
566 (we've seen branches and tags are special cases of
567 <<how-git-stores-references,references>>).
568 - HEAD: refers to the head of the current branch
569
570 There are many more; see the "SPECIFYING REVISIONS" section of the
571 linkgit:git-rev-parse[1] man page for the complete list of ways to
572 name revisions. Some examples:
573
574 -------------------------------------------------
575 $ git show fb47ddb2 # the first few characters of the object name
576 # are usually enough to specify it uniquely
577 $ git show HEAD^ # the parent of the HEAD commit
578 $ git show HEAD^^ # the grandparent
579 $ git show HEAD~4 # the great-great-grandparent
580 -------------------------------------------------
581
582 Recall that merge commits may have more than one parent; by default,
583 ^ and ~ follow the first parent listed in the commit, but you can
584 also choose:
585
586 -------------------------------------------------
587 $ git show HEAD^1 # show the first parent of HEAD
588 $ git show HEAD^2 # show the second parent of HEAD
589 -------------------------------------------------
590
591 In addition to HEAD, there are several other special names for
592 commits:
593
594 Merges (to be discussed later), as well as operations such as
595 `git reset`, which change the currently checked-out commit, generally
596 set ORIG_HEAD to the value HEAD had before the current operation.
597
598 The `git fetch` operation always stores the head of the last fetched
599 branch in FETCH_HEAD. For example, if you run `git fetch` without
600 specifying a local branch as the target of the operation
601
602 -------------------------------------------------
603 $ git fetch git://example.com/proj.git theirbranch
604 -------------------------------------------------
605
606 the fetched commits will still be available from FETCH_HEAD.
607
608 When we discuss merges we'll also see the special name MERGE_HEAD,
609 which refers to the other branch that we're merging in to the current
610 branch.
611
612 The linkgit:git-rev-parse[1] command is a low-level command that is
613 occasionally useful for translating some name for a commit to the object
614 name for that commit:
615
616 -------------------------------------------------
617 $ git rev-parse origin
618 e05db0fd4f31dde7005f075a84f96b360d05984b
619 -------------------------------------------------
620
621 [[creating-tags]]
622 Creating tags
623 -------------
624
625 We can also create a tag to refer to a particular commit; after
626 running
627
628 -------------------------------------------------
629 $ git tag stable-1 1b2e1d63ff
630 -------------------------------------------------
631
632 You can use stable-1 to refer to the commit 1b2e1d63ff.
633
634 This creates a "lightweight" tag. If you would also like to include a
635 comment with the tag, and possibly sign it cryptographically, then you
636 should create a tag object instead; see the linkgit:git-tag[1] man page
637 for details.
638
639 [[browsing-revisions]]
640 Browsing revisions
641 ------------------
642
643 The linkgit:git-log[1] command can show lists of commits. On its
644 own, it shows all commits reachable from the parent commit; but you
645 can also make more specific requests:
646
647 -------------------------------------------------
648 $ git log v2.5.. # commits since (not reachable from) v2.5
649 $ git log test..master # commits reachable from master but not test
650 $ git log master..test # ...reachable from test but not master
651 $ git log master...test # ...reachable from either test or master,
652 # but not both
653 $ git log --since="2 weeks ago" # commits from the last 2 weeks
654 $ git log Makefile # commits which modify Makefile
655 $ git log fs/ # ... which modify any file under fs/
656 $ git log -S'foo()' # commits which add or remove any file data
657 # matching the string 'foo()'
658 -------------------------------------------------
659
660 And of course you can combine all of these; the following finds
661 commits since v2.5 which touch the Makefile or any file under fs:
662
663 -------------------------------------------------
664 $ git log v2.5.. Makefile fs/
665 -------------------------------------------------
666
667 You can also ask git log to show patches:
668
669 -------------------------------------------------
670 $ git log -p
671 -------------------------------------------------
672
673 See the "--pretty" option in the linkgit:git-log[1] man page for more
674 display options.
675
676 Note that git log starts with the most recent commit and works
677 backwards through the parents; however, since git history can contain
678 multiple independent lines of development, the particular order that
679 commits are listed in may be somewhat arbitrary.
680
681 [[generating-diffs]]
682 Generating diffs
683 ----------------
684
685 You can generate diffs between any two versions using
686 linkgit:git-diff[1]:
687
688 -------------------------------------------------
689 $ git diff master..test
690 -------------------------------------------------
691
692 That will produce the diff between the tips of the two branches. If
693 you'd prefer to find the diff from their common ancestor to test, you
694 can use three dots instead of two:
695
696 -------------------------------------------------
697 $ git diff master...test
698 -------------------------------------------------
699
700 Sometimes what you want instead is a set of patches; for this you can
701 use linkgit:git-format-patch[1]:
702
703 -------------------------------------------------
704 $ git format-patch master..test
705 -------------------------------------------------
706
707 will generate a file with a patch for each commit reachable from test
708 but not from master.
709
710 [[viewing-old-file-versions]]
711 Viewing old file versions
712 -------------------------
713
714 You can always view an old version of a file by just checking out the
715 correct revision first. But sometimes it is more convenient to be
716 able to view an old version of a single file without checking
717 anything out; this command does that:
718
719 -------------------------------------------------
720 $ git show v2.5:fs/locks.c
721 -------------------------------------------------
722
723 Before the colon may be anything that names a commit, and after it
724 may be any path to a file tracked by git.
725
726 [[history-examples]]
727 Examples
728 --------
729
730 [[counting-commits-on-a-branch]]
731 Counting the number of commits on a branch
732 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
733
734 Suppose you want to know how many commits you've made on "mybranch"
735 since it diverged from "origin":
736
737 -------------------------------------------------
738 $ git log --pretty=oneline origin..mybranch | wc -l
739 -------------------------------------------------
740
741 Alternatively, you may often see this sort of thing done with the
742 lower-level command linkgit:git-rev-list[1], which just lists the SHA1's
743 of all the given commits:
744
745 -------------------------------------------------
746 $ git rev-list origin..mybranch | wc -l
747 -------------------------------------------------
748
749 [[checking-for-equal-branches]]
750 Check whether two branches point at the same history
751 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
752
753 Suppose you want to check whether two branches point at the same point
754 in history.
755
756 -------------------------------------------------
757 $ git diff origin..master
758 -------------------------------------------------
759
760 will tell you whether the contents of the project are the same at the
761 two branches; in theory, however, it's possible that the same project
762 contents could have been arrived at by two different historical
763 routes. You could compare the object names:
764
765 -------------------------------------------------
766 $ git rev-list origin
767 e05db0fd4f31dde7005f075a84f96b360d05984b
768 $ git rev-list master
769 e05db0fd4f31dde7005f075a84f96b360d05984b
770 -------------------------------------------------
771
772 Or you could recall that the ... operator selects all commits
773 contained reachable from either one reference or the other but not
774 both: so
775
776 -------------------------------------------------
777 $ git log origin...master
778 -------------------------------------------------
779
780 will return no commits when the two branches are equal.
781
782 [[finding-tagged-descendants]]
783 Find first tagged version including a given fix
784 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
785
786 Suppose you know that the commit e05db0fd fixed a certain problem.
787 You'd like to find the earliest tagged release that contains that
788 fix.
789
790 Of course, there may be more than one answer--if the history branched
791 after commit e05db0fd, then there could be multiple "earliest" tagged
792 releases.
793
794 You could just visually inspect the commits since e05db0fd:
795
796 -------------------------------------------------
797 $ gitk e05db0fd..
798 -------------------------------------------------
799
800 Or you can use linkgit:git-name-rev[1], which will give the commit a
801 name based on any tag it finds pointing to one of the commit's
802 descendants:
803
804 -------------------------------------------------
805 $ git name-rev --tags e05db0fd
806 e05db0fd tags/v1.5.0-rc1^0~23
807 -------------------------------------------------
808
809 The linkgit:git-describe[1] command does the opposite, naming the
810 revision using a tag on which the given commit is based:
811
812 -------------------------------------------------
813 $ git describe e05db0fd
814 v1.5.0-rc0-260-ge05db0f
815 -------------------------------------------------
816
817 but that may sometimes help you guess which tags might come after the
818 given commit.
819
820 If you just want to verify whether a given tagged version contains a
821 given commit, you could use linkgit:git-merge-base[1]:
822
823 -------------------------------------------------
824 $ git merge-base e05db0fd v1.5.0-rc1
825 e05db0fd4f31dde7005f075a84f96b360d05984b
826 -------------------------------------------------
827
828 The merge-base command finds a common ancestor of the given commits,
829 and always returns one or the other in the case where one is a
830 descendant of the other; so the above output shows that e05db0fd
831 actually is an ancestor of v1.5.0-rc1.
832
833 Alternatively, note that
834
835 -------------------------------------------------
836 $ git log v1.5.0-rc1..e05db0fd
837 -------------------------------------------------
838
839 will produce empty output if and only if v1.5.0-rc1 includes e05db0fd,
840 because it outputs only commits that are not reachable from v1.5.0-rc1.
841
842 As yet another alternative, the linkgit:git-show-branch[1] command lists
843 the commits reachable from its arguments with a display on the left-hand
844 side that indicates which arguments that commit is reachable from. So,
845 you can run something like
846
847 -------------------------------------------------
848 $ git show-branch e05db0fd v1.5.0-rc0 v1.5.0-rc1 v1.5.0-rc2
849 ! [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
850 available
851 ! [v1.5.0-rc0] GIT v1.5.0 preview
852 ! [v1.5.0-rc1] GIT v1.5.0-rc1
853 ! [v1.5.0-rc2] GIT v1.5.0-rc2
854 ...
855 -------------------------------------------------
856
857 then search for a line that looks like
858
859 -------------------------------------------------
860 + ++ [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
861 available
862 -------------------------------------------------
863
864 Which shows that e05db0fd is reachable from itself, from v1.5.0-rc1, and
865 from v1.5.0-rc2, but not from v1.5.0-rc0.
866
867 [[showing-commits-unique-to-a-branch]]
868 Showing commits unique to a given branch
869 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
870
871 Suppose you would like to see all the commits reachable from the branch
872 head named "master" but not from any other head in your repository.
873
874 We can list all the heads in this repository with
875 linkgit:git-show-ref[1]:
876
877 -------------------------------------------------
878 $ git show-ref --heads
879 bf62196b5e363d73353a9dcf094c59595f3153b7 refs/heads/core-tutorial
880 db768d5504c1bb46f63ee9d6e1772bd047e05bf9 refs/heads/maint
881 a07157ac624b2524a059a3414e99f6f44bebc1e7 refs/heads/master
882 24dbc180ea14dc1aebe09f14c8ecf32010690627 refs/heads/tutorial-2
883 1e87486ae06626c2f31eaa63d26fc0fd646c8af2 refs/heads/tutorial-fixes
884 -------------------------------------------------
885
886 We can get just the branch-head names, and remove "master", with
887 the help of the standard utilities cut and grep:
888
889 -------------------------------------------------
890 $ git show-ref --heads | cut -d' ' -f2 | grep -v '^refs/heads/master'
891 refs/heads/core-tutorial
892 refs/heads/maint
893 refs/heads/tutorial-2
894 refs/heads/tutorial-fixes
895 -------------------------------------------------
896
897 And then we can ask to see all the commits reachable from master
898 but not from these other heads:
899
900 -------------------------------------------------
901 $ gitk master --not $( git show-ref --heads | cut -d' ' -f2 |
902 grep -v '^refs/heads/master' )
903 -------------------------------------------------
904
905 Obviously, endless variations are possible; for example, to see all
906 commits reachable from some head but not from any tag in the repository:
907
908 -------------------------------------------------
909 $ gitk $( git show-ref --heads ) --not $( git show-ref --tags )
910 -------------------------------------------------
911
912 (See linkgit:git-rev-parse[1] for explanations of commit-selecting
913 syntax such as `--not`.)
914
915 [[making-a-release]]
916 Creating a changelog and tarball for a software release
917 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
918
919 The linkgit:git-archive[1] command can create a tar or zip archive from
920 any version of a project; for example:
921
922 -------------------------------------------------
923 $ git archive --format=tar --prefix=project/ HEAD | gzip >latest.tar.gz
924 -------------------------------------------------
925
926 will use HEAD to produce a tar archive in which each filename is
927 preceded by "project/".
928
929 If you're releasing a new version of a software project, you may want
930 to simultaneously make a changelog to include in the release
931 announcement.
932
933 Linus Torvalds, for example, makes new kernel releases by tagging them,
934 then running:
935
936 -------------------------------------------------
937 $ release-script 2.6.12 2.6.13-rc6 2.6.13-rc7
938 -------------------------------------------------
939
940 where release-script is a shell script that looks like:
941
942 -------------------------------------------------
943 #!/bin/sh
944 stable="$1"
945 last="$2"
946 new="$3"
947 echo "# git tag v$new"
948 echo "git archive --prefix=linux-$new/ v$new | gzip -9 > ../linux-$new.tar.gz"
949 echo "git diff v$stable v$new | gzip -9 > ../patch-$new.gz"
950 echo "git log --no-merges v$new ^v$last > ../ChangeLog-$new"
951 echo "git shortlog --no-merges v$new ^v$last > ../ShortLog"
952 echo "git diff --stat --summary -M v$last v$new > ../diffstat-$new"
953 -------------------------------------------------
954
955 and then he just cut-and-pastes the output commands after verifying that
956 they look OK.
957
958 [[Finding-comments-With-given-Content]]
959 Finding commits referencing a file with given content
960 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
961
962 Somebody hands you a copy of a file, and asks which commits modified a
963 file such that it contained the given content either before or after the
964 commit. You can find out with this:
965
966 -------------------------------------------------
967 $ git log --raw --abbrev=40 --pretty=oneline |
968 grep -B 1 `git hash-object filename`
969 -------------------------------------------------
970
971 Figuring out why this works is left as an exercise to the (advanced)
972 student. The linkgit:git-log[1], linkgit:git-diff-tree[1], and
973 linkgit:git-hash-object[1] man pages may prove helpful.
974
975 [[Developing-With-git]]
976 Developing with git
977 ===================
978
979 [[telling-git-your-name]]
980 Telling git your name
981 ---------------------
982
983 Before creating any commits, you should introduce yourself to git. The
984 easiest way to do so is to make sure the following lines appear in a
985 file named .gitconfig in your home directory:
986
987 ------------------------------------------------
988 [user]
989 name = Your Name Comes Here
990 email = you@yourdomain.example.com
991 ------------------------------------------------
992
993 (See the "CONFIGURATION FILE" section of linkgit:git-config[1] for
994 details on the configuration file.)
995
996
997 [[creating-a-new-repository]]
998 Creating a new repository
999 -------------------------
1000
1001 Creating a new repository from scratch is very easy:
1002
1003 -------------------------------------------------
1004 $ mkdir project
1005 $ cd project
1006 $ git init
1007 -------------------------------------------------
1008
1009 If you have some initial content (say, a tarball):
1010
1011 -------------------------------------------------
1012 $ tar xzvf project.tar.gz
1013 $ cd project
1014 $ git init
1015 $ git add . # include everything below ./ in the first commit:
1016 $ git commit
1017 -------------------------------------------------
1018
1019 [[how-to-make-a-commit]]
1020 How to make a commit
1021 --------------------
1022
1023 Creating a new commit takes three steps:
1024
1025 1. Making some changes to the working directory using your
1026 favorite editor.
1027 2. Telling git about your changes.
1028 3. Creating the commit using the content you told git about
1029 in step 2.
1030
1031 In practice, you can interleave and repeat steps 1 and 2 as many
1032 times as you want: in order to keep track of what you want committed
1033 at step 3, git maintains a snapshot of the tree's contents in a
1034 special staging area called "the index."
1035
1036 At the beginning, the content of the index will be identical to
1037 that of the HEAD. The command "git diff --cached", which shows
1038 the difference between the HEAD and the index, should therefore
1039 produce no output at that point.
1040
1041 Modifying the index is easy:
1042
1043 To update the index with the new contents of a modified file, use
1044
1045 -------------------------------------------------
1046 $ git add path/to/file
1047 -------------------------------------------------
1048
1049 To add the contents of a new file to the index, use
1050
1051 -------------------------------------------------
1052 $ git add path/to/file
1053 -------------------------------------------------
1054
1055 To remove a file from the index and from the working tree,
1056
1057 -------------------------------------------------
1058 $ git rm path/to/file
1059 -------------------------------------------------
1060
1061 After each step you can verify that
1062
1063 -------------------------------------------------
1064 $ git diff --cached
1065 -------------------------------------------------
1066
1067 always shows the difference between the HEAD and the index file--this
1068 is what you'd commit if you created the commit now--and that
1069
1070 -------------------------------------------------
1071 $ git diff
1072 -------------------------------------------------
1073
1074 shows the difference between the working tree and the index file.
1075
1076 Note that "git add" always adds just the current contents of a file
1077 to the index; further changes to the same file will be ignored unless
1078 you run `git add` on the file again.
1079
1080 When you're ready, just run
1081
1082 -------------------------------------------------
1083 $ git commit
1084 -------------------------------------------------
1085
1086 and git will prompt you for a commit message and then create the new
1087 commit. Check to make sure it looks like what you expected with
1088
1089 -------------------------------------------------
1090 $ git show
1091 -------------------------------------------------
1092
1093 As a special shortcut,
1094
1095 -------------------------------------------------
1096 $ git commit -a
1097 -------------------------------------------------
1098
1099 will update the index with any files that you've modified or removed
1100 and create a commit, all in one step.
1101
1102 A number of commands are useful for keeping track of what you're
1103 about to commit:
1104
1105 -------------------------------------------------
1106 $ git diff --cached # difference between HEAD and the index; what
1107 # would be committed if you ran "commit" now.
1108 $ git diff # difference between the index file and your
1109 # working directory; changes that would not
1110 # be included if you ran "commit" now.
1111 $ git diff HEAD # difference between HEAD and working tree; what
1112 # would be committed if you ran "commit -a" now.
1113 $ git status # a brief per-file summary of the above.
1114 -------------------------------------------------
1115
1116 You can also use linkgit:git-gui[1] to create commits, view changes in
1117 the index and the working tree files, and individually select diff hunks
1118 for inclusion in the index (by right-clicking on the diff hunk and
1119 choosing "Stage Hunk For Commit").
1120
1121 [[creating-good-commit-messages]]
1122 Creating good commit messages
1123 -----------------------------
1124
1125 Though not required, it's a good idea to begin the commit message
1126 with a single short (less than 50 character) line summarizing the
1127 change, followed by a blank line and then a more thorough
1128 description. Tools that turn commits into email, for example, use
1129 the first line on the Subject line and the rest of the commit in the
1130 body.
1131
1132 [[ignoring-files]]
1133 Ignoring files
1134 --------------
1135
1136 A project will often generate files that you do 'not' want to track with git.
1137 This typically includes files generated by a build process or temporary
1138 backup files made by your editor. Of course, 'not' tracking files with git
1139 is just a matter of 'not' calling `git add` on them. But it quickly becomes
1140 annoying to have these untracked files lying around; e.g. they make
1141 `git add .` practically useless, and they keep showing up in the output of
1142 `git status`.
1143
1144 You can tell git to ignore certain files by creating a file called .gitignore
1145 in the top level of your working directory, with contents such as:
1146
1147 -------------------------------------------------
1148 # Lines starting with '#' are considered comments.
1149 # Ignore any file named foo.txt.
1150 foo.txt
1151 # Ignore (generated) html files,
1152 *.html
1153 # except foo.html which is maintained by hand.
1154 !foo.html
1155 # Ignore objects and archives.
1156 *.[oa]
1157 -------------------------------------------------
1158
1159 See linkgit:gitignore[5] for a detailed explanation of the syntax. You can
1160 also place .gitignore files in other directories in your working tree, and they
1161 will apply to those directories and their subdirectories. The `.gitignore`
1162 files can be added to your repository like any other files (just run `git add
1163 .gitignore` and `git commit`, as usual), which is convenient when the exclude
1164 patterns (such as patterns matching build output files) would also make sense
1165 for other users who clone your repository.
1166
1167 If you wish the exclude patterns to affect only certain repositories
1168 (instead of every repository for a given project), you may instead put
1169 them in a file in your repository named .git/info/exclude, or in any file
1170 specified by the `core.excludesfile` configuration variable. Some git
1171 commands can also take exclude patterns directly on the command line.
1172 See linkgit:gitignore[5] for the details.
1173
1174 [[how-to-merge]]
1175 How to merge
1176 ------------
1177
1178 You can rejoin two diverging branches of development using
1179 linkgit:git-merge[1]:
1180
1181 -------------------------------------------------
1182 $ git merge branchname
1183 -------------------------------------------------
1184
1185 merges the development in the branch "branchname" into the current
1186 branch. If there are conflicts--for example, if the same file is
1187 modified in two different ways in the remote branch and the local
1188 branch--then you are warned; the output may look something like this:
1189
1190 -------------------------------------------------
1191 $ git merge next
1192 100% (4/4) done
1193 Auto-merged file.txt
1194 CONFLICT (content): Merge conflict in file.txt
1195 Automatic merge failed; fix conflicts and then commit the result.
1196 -------------------------------------------------
1197
1198 Conflict markers are left in the problematic files, and after
1199 you resolve the conflicts manually, you can update the index
1200 with the contents and run git commit, as you normally would when
1201 creating a new file.
1202
1203 If you examine the resulting commit using gitk, you will see that it
1204 has two parents, one pointing to the top of the current branch, and
1205 one to the top of the other branch.
1206
1207 [[resolving-a-merge]]
1208 Resolving a merge
1209 -----------------
1210
1211 When a merge isn't resolved automatically, git leaves the index and
1212 the working tree in a special state that gives you all the
1213 information you need to help resolve the merge.
1214
1215 Files with conflicts are marked specially in the index, so until you
1216 resolve the problem and update the index, linkgit:git-commit[1] will
1217 fail:
1218
1219 -------------------------------------------------
1220 $ git commit
1221 file.txt: needs merge
1222 -------------------------------------------------
1223
1224 Also, linkgit:git-status[1] will list those files as "unmerged", and the
1225 files with conflicts will have conflict markers added, like this:
1226
1227 -------------------------------------------------
1228 <<<<<<< HEAD:file.txt
1229 Hello world
1230 =======
1231 Goodbye
1232 >>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1233 -------------------------------------------------
1234
1235 All you need to do is edit the files to resolve the conflicts, and then
1236
1237 -------------------------------------------------
1238 $ git add file.txt
1239 $ git commit
1240 -------------------------------------------------
1241
1242 Note that the commit message will already be filled in for you with
1243 some information about the merge. Normally you can just use this
1244 default message unchanged, but you may add additional commentary of
1245 your own if desired.
1246
1247 The above is all you need to know to resolve a simple merge. But git
1248 also provides more information to help resolve conflicts:
1249
1250 [[conflict-resolution]]
1251 Getting conflict-resolution help during a merge
1252 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1253
1254 All of the changes that git was able to merge automatically are
1255 already added to the index file, so linkgit:git-diff[1] shows only
1256 the conflicts. It uses an unusual syntax:
1257
1258 -------------------------------------------------
1259 $ git diff
1260 diff --cc file.txt
1261 index 802992c,2b60207..0000000
1262 --- a/file.txt
1263 +++ b/file.txt
1264 @@@ -1,1 -1,1 +1,5 @@@
1265 ++<<<<<<< HEAD:file.txt
1266 +Hello world
1267 ++=======
1268 + Goodbye
1269 ++>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1270 -------------------------------------------------
1271
1272 Recall that the commit which will be committed after we resolve this
1273 conflict will have two parents instead of the usual one: one parent
1274 will be HEAD, the tip of the current branch; the other will be the
1275 tip of the other branch, which is stored temporarily in MERGE_HEAD.
1276
1277 During the merge, the index holds three versions of each file. Each of
1278 these three "file stages" represents a different version of the file:
1279
1280 -------------------------------------------------
1281 $ git show :1:file.txt # the file in a common ancestor of both branches
1282 $ git show :2:file.txt # the version from HEAD.
1283 $ git show :3:file.txt # the version from MERGE_HEAD.
1284 -------------------------------------------------
1285
1286 When you ask linkgit:git-diff[1] to show the conflicts, it runs a
1287 three-way diff between the conflicted merge results in the work tree with
1288 stages 2 and 3 to show only hunks whose contents come from both sides,
1289 mixed (in other words, when a hunk's merge results come only from stage 2,
1290 that part is not conflicting and is not shown. Same for stage 3).
1291
1292 The diff above shows the differences between the working-tree version of
1293 file.txt and the stage 2 and stage 3 versions. So instead of preceding
1294 each line by a single "+" or "-", it now uses two columns: the first
1295 column is used for differences between the first parent and the working
1296 directory copy, and the second for differences between the second parent
1297 and the working directory copy. (See the "COMBINED DIFF FORMAT" section
1298 of linkgit:git-diff-files[1] for a details of the format.)
1299
1300 After resolving the conflict in the obvious way (but before updating the
1301 index), the diff will look like:
1302
1303 -------------------------------------------------
1304 $ git diff
1305 diff --cc file.txt
1306 index 802992c,2b60207..0000000
1307 --- a/file.txt
1308 +++ b/file.txt
1309 @@@ -1,1 -1,1 +1,1 @@@
1310 - Hello world
1311 -Goodbye
1312 ++Goodbye world
1313 -------------------------------------------------
1314
1315 This shows that our resolved version deleted "Hello world" from the
1316 first parent, deleted "Goodbye" from the second parent, and added
1317 "Goodbye world", which was previously absent from both.
1318
1319 Some special diff options allow diffing the working directory against
1320 any of these stages:
1321
1322 -------------------------------------------------
1323 $ git diff -1 file.txt # diff against stage 1
1324 $ git diff --base file.txt # same as the above
1325 $ git diff -2 file.txt # diff against stage 2
1326 $ git diff --ours file.txt # same as the above
1327 $ git diff -3 file.txt # diff against stage 3
1328 $ git diff --theirs file.txt # same as the above.
1329 -------------------------------------------------
1330
1331 The linkgit:git-log[1] and linkgit:gitk[1] commands also provide special help
1332 for merges:
1333
1334 -------------------------------------------------
1335 $ git log --merge
1336 $ gitk --merge
1337 -------------------------------------------------
1338
1339 These will display all commits which exist only on HEAD or on
1340 MERGE_HEAD, and which touch an unmerged file.
1341
1342 You may also use linkgit:git-mergetool[1], which lets you merge the
1343 unmerged files using external tools such as Emacs or kdiff3.
1344
1345 Each time you resolve the conflicts in a file and update the index:
1346
1347 -------------------------------------------------
1348 $ git add file.txt
1349 -------------------------------------------------
1350
1351 the different stages of that file will be "collapsed", after which
1352 `git diff` will (by default) no longer show diffs for that file.
1353
1354 [[undoing-a-merge]]
1355 Undoing a merge
1356 ---------------
1357
1358 If you get stuck and decide to just give up and throw the whole mess
1359 away, you can always return to the pre-merge state with
1360
1361 -------------------------------------------------
1362 $ git reset --hard HEAD
1363 -------------------------------------------------
1364
1365 Or, if you've already committed the merge that you want to throw away,
1366
1367 -------------------------------------------------
1368 $ git reset --hard ORIG_HEAD
1369 -------------------------------------------------
1370
1371 However, this last command can be dangerous in some cases--never
1372 throw away a commit you have already committed if that commit may
1373 itself have been merged into another branch, as doing so may confuse
1374 further merges.
1375
1376 [[fast-forwards]]
1377 Fast-forward merges
1378 -------------------
1379
1380 There is one special case not mentioned above, which is treated
1381 differently. Normally, a merge results in a merge commit, with two
1382 parents, one pointing at each of the two lines of development that
1383 were merged.
1384
1385 However, if the current branch is a descendant of the other--so every
1386 commit present in the one is already contained in the other--then git
1387 just performs a "fast forward"; the head of the current branch is moved
1388 forward to point at the head of the merged-in branch, without any new
1389 commits being created.
1390
1391 [[fixing-mistakes]]
1392 Fixing mistakes
1393 ---------------
1394
1395 If you've messed up the working tree, but haven't yet committed your
1396 mistake, you can return the entire working tree to the last committed
1397 state with
1398
1399 -------------------------------------------------
1400 $ git reset --hard HEAD
1401 -------------------------------------------------
1402
1403 If you make a commit that you later wish you hadn't, there are two
1404 fundamentally different ways to fix the problem:
1405
1406 1. You can create a new commit that undoes whatever was done
1407 by the old commit. This is the correct thing if your
1408 mistake has already been made public.
1409
1410 2. You can go back and modify the old commit. You should
1411 never do this if you have already made the history public;
1412 git does not normally expect the "history" of a project to
1413 change, and cannot correctly perform repeated merges from
1414 a branch that has had its history changed.
1415
1416 [[reverting-a-commit]]
1417 Fixing a mistake with a new commit
1418 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1419
1420 Creating a new commit that reverts an earlier change is very easy;
1421 just pass the linkgit:git-revert[1] command a reference to the bad
1422 commit; for example, to revert the most recent commit:
1423
1424 -------------------------------------------------
1425 $ git revert HEAD
1426 -------------------------------------------------
1427
1428 This will create a new commit which undoes the change in HEAD. You
1429 will be given a chance to edit the commit message for the new commit.
1430
1431 You can also revert an earlier change, for example, the next-to-last:
1432
1433 -------------------------------------------------
1434 $ git revert HEAD^
1435 -------------------------------------------------
1436
1437 In this case git will attempt to undo the old change while leaving
1438 intact any changes made since then. If more recent changes overlap
1439 with the changes to be reverted, then you will be asked to fix
1440 conflicts manually, just as in the case of <<resolving-a-merge,
1441 resolving a merge>>.
1442
1443 [[fixing-a-mistake-by-rewriting-history]]
1444 Fixing a mistake by rewriting history
1445 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1446
1447 If the problematic commit is the most recent commit, and you have not
1448 yet made that commit public, then you may just
1449 <<undoing-a-merge,destroy it using `git reset`>>.
1450
1451 Alternatively, you
1452 can edit the working directory and update the index to fix your
1453 mistake, just as if you were going to <<how-to-make-a-commit,create a
1454 new commit>>, then run
1455
1456 -------------------------------------------------
1457 $ git commit --amend
1458 -------------------------------------------------
1459
1460 which will replace the old commit by a new commit incorporating your
1461 changes, giving you a chance to edit the old commit message first.
1462
1463 Again, you should never do this to a commit that may already have
1464 been merged into another branch; use linkgit:git-revert[1] instead in
1465 that case.
1466
1467 It is also possible to replace commits further back in the history, but
1468 this is an advanced topic to be left for
1469 <<cleaning-up-history,another chapter>>.
1470
1471 [[checkout-of-path]]
1472 Checking out an old version of a file
1473 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1474
1475 In the process of undoing a previous bad change, you may find it
1476 useful to check out an older version of a particular file using
1477 linkgit:git-checkout[1]. We've used `git checkout` before to switch
1478 branches, but it has quite different behavior if it is given a path
1479 name: the command
1480
1481 -------------------------------------------------
1482 $ git checkout HEAD^ path/to/file
1483 -------------------------------------------------
1484
1485 replaces path/to/file by the contents it had in the commit HEAD^, and
1486 also updates the index to match. It does not change branches.
1487
1488 If you just want to look at an old version of the file, without
1489 modifying the working directory, you can do that with
1490 linkgit:git-show[1]:
1491
1492 -------------------------------------------------
1493 $ git show HEAD^:path/to/file
1494 -------------------------------------------------
1495
1496 which will display the given version of the file.
1497
1498 [[interrupted-work]]
1499 Temporarily setting aside work in progress
1500 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1501
1502 While you are in the middle of working on something complicated, you
1503 find an unrelated but obvious and trivial bug. You would like to fix it
1504 before continuing. You can use linkgit:git-stash[1] to save the current
1505 state of your work, and after fixing the bug (or, optionally after doing
1506 so on a different branch and then coming back), unstash the
1507 work-in-progress changes.
1508
1509 ------------------------------------------------
1510 $ git stash save "work in progress for foo feature"
1511 ------------------------------------------------
1512
1513 This command will save your changes away to the `stash`, and
1514 reset your working tree and the index to match the tip of your
1515 current branch. Then you can make your fix as usual.
1516
1517 ------------------------------------------------
1518 ... edit and test ...
1519 $ git commit -a -m "blorpl: typofix"
1520 ------------------------------------------------
1521
1522 After that, you can go back to what you were working on with
1523 `git stash apply`:
1524
1525 ------------------------------------------------
1526 $ git stash apply
1527 ------------------------------------------------
1528
1529
1530 [[ensuring-good-performance]]
1531 Ensuring good performance
1532 -------------------------
1533
1534 On large repositories, git depends on compression to keep the history
1535 information from taking up too much space on disk or in memory.
1536
1537 This compression is not performed automatically. Therefore you
1538 should occasionally run linkgit:git-gc[1]:
1539
1540 -------------------------------------------------
1541 $ git gc
1542 -------------------------------------------------
1543
1544 to recompress the archive. This can be very time-consuming, so
1545 you may prefer to run `git gc` when you are not doing other work.
1546
1547
1548 [[ensuring-reliability]]
1549 Ensuring reliability
1550 --------------------
1551
1552 [[checking-for-corruption]]
1553 Checking the repository for corruption
1554 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1555
1556 The linkgit:git-fsck[1] command runs a number of self-consistency checks
1557 on the repository, and reports on any problems. This may take some
1558 time. The most common warning by far is about "dangling" objects:
1559
1560 -------------------------------------------------
1561 $ git fsck
1562 dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1563 dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1564 dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1565 dangling blob 218761f9d90712d37a9c5e36f406f92202db07eb
1566 dangling commit bf093535a34a4d35731aa2bd90fe6b176302f14f
1567 dangling commit 8e4bec7f2ddaa268bef999853c25755452100f8e
1568 dangling tree d50bb86186bf27b681d25af89d3b5b68382e4085
1569 dangling tree b24c2473f1fd3d91352a624795be026d64c8841f
1570 ...
1571 -------------------------------------------------
1572
1573 Dangling objects are not a problem. At worst they may take up a little
1574 extra disk space. They can sometimes provide a last-resort method for
1575 recovering lost work--see <<dangling-objects>> for details.
1576
1577 [[recovering-lost-changes]]
1578 Recovering lost changes
1579 ~~~~~~~~~~~~~~~~~~~~~~~
1580
1581 [[reflogs]]
1582 Reflogs
1583 ^^^^^^^
1584
1585 Say you modify a branch with `linkgit:git-reset[1] --hard`, and then
1586 realize that the branch was the only reference you had to that point in
1587 history.
1588
1589 Fortunately, git also keeps a log, called a "reflog", of all the
1590 previous values of each branch. So in this case you can still find the
1591 old history using, for example,
1592
1593 -------------------------------------------------
1594 $ git log master@{1}
1595 -------------------------------------------------
1596
1597 This lists the commits reachable from the previous version of the
1598 "master" branch head. This syntax can be used with any git command
1599 that accepts a commit, not just with git log. Some other examples:
1600
1601 -------------------------------------------------
1602 $ git show master@{2} # See where the branch pointed 2,
1603 $ git show master@{3} # 3, ... changes ago.
1604 $ gitk master@{yesterday} # See where it pointed yesterday,
1605 $ gitk master@{"1 week ago"} # ... or last week
1606 $ git log --walk-reflogs master # show reflog entries for master
1607 -------------------------------------------------
1608
1609 A separate reflog is kept for the HEAD, so
1610
1611 -------------------------------------------------
1612 $ git show HEAD@{"1 week ago"}
1613 -------------------------------------------------
1614
1615 will show what HEAD pointed to one week ago, not what the current branch
1616 pointed to one week ago. This allows you to see the history of what
1617 you've checked out.
1618
1619 The reflogs are kept by default for 30 days, after which they may be
1620 pruned. See linkgit:git-reflog[1] and linkgit:git-gc[1] to learn
1621 how to control this pruning, and see the "SPECIFYING REVISIONS"
1622 section of linkgit:git-rev-parse[1] for details.
1623
1624 Note that the reflog history is very different from normal git history.
1625 While normal history is shared by every repository that works on the
1626 same project, the reflog history is not shared: it tells you only about
1627 how the branches in your local repository have changed over time.
1628
1629 [[dangling-object-recovery]]
1630 Examining dangling objects
1631 ^^^^^^^^^^^^^^^^^^^^^^^^^^
1632
1633 In some situations the reflog may not be able to save you. For example,
1634 suppose you delete a branch, then realize you need the history it
1635 contained. The reflog is also deleted; however, if you have not yet
1636 pruned the repository, then you may still be able to find the lost
1637 commits in the dangling objects that `git fsck` reports. See
1638 <<dangling-objects>> for the details.
1639
1640 -------------------------------------------------
1641 $ git fsck
1642 dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1643 dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1644 dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1645 ...
1646 -------------------------------------------------
1647
1648 You can examine
1649 one of those dangling commits with, for example,
1650
1651 ------------------------------------------------
1652 $ gitk 7281251ddd --not --all
1653 ------------------------------------------------
1654
1655 which does what it sounds like: it says that you want to see the commit
1656 history that is described by the dangling commit(s), but not the
1657 history that is described by all your existing branches and tags. Thus
1658 you get exactly the history reachable from that commit that is lost.
1659 (And notice that it might not be just one commit: we only report the
1660 "tip of the line" as being dangling, but there might be a whole deep
1661 and complex commit history that was dropped.)
1662
1663 If you decide you want the history back, you can always create a new
1664 reference pointing to it, for example, a new branch:
1665
1666 ------------------------------------------------
1667 $ git branch recovered-branch 7281251ddd
1668 ------------------------------------------------
1669
1670 Other types of dangling objects (blobs and trees) are also possible, and
1671 dangling objects can arise in other situations.
1672
1673
1674 [[sharing-development]]
1675 Sharing development with others
1676 ===============================
1677
1678 [[getting-updates-With-git-pull]]
1679 Getting updates with git pull
1680 -----------------------------
1681
1682 After you clone a repository and make a few changes of your own, you
1683 may wish to check the original repository for updates and merge them
1684 into your own work.
1685
1686 We have already seen <<Updating-a-repository-With-git-fetch,how to
1687 keep remote tracking branches up to date>> with linkgit:git-fetch[1],
1688 and how to merge two branches. So you can merge in changes from the
1689 original repository's master branch with:
1690
1691 -------------------------------------------------
1692 $ git fetch
1693 $ git merge origin/master
1694 -------------------------------------------------
1695
1696 However, the linkgit:git-pull[1] command provides a way to do this in
1697 one step:
1698
1699 -------------------------------------------------
1700 $ git pull origin master
1701 -------------------------------------------------
1702
1703 In fact, if you have "master" checked out, then by default "git pull"
1704 merges from the HEAD branch of the origin repository. So often you can
1705 accomplish the above with just a simple
1706
1707 -------------------------------------------------
1708 $ git pull
1709 -------------------------------------------------
1710
1711 More generally, a branch that is created from a remote branch will pull
1712 by default from that branch. See the descriptions of the
1713 branch.<name>.remote and branch.<name>.merge options in
1714 linkgit:git-config[1], and the discussion of the `--track` option in
1715 linkgit:git-checkout[1], to learn how to control these defaults.
1716
1717 In addition to saving you keystrokes, "git pull" also helps you by
1718 producing a default commit message documenting the branch and
1719 repository that you pulled from.
1720
1721 (But note that no such commit will be created in the case of a
1722 <<fast-forwards,fast forward>>; instead, your branch will just be
1723 updated to point to the latest commit from the upstream branch.)
1724
1725 The `git pull` command can also be given "." as the "remote" repository,
1726 in which case it just merges in a branch from the current repository; so
1727 the commands
1728
1729 -------------------------------------------------
1730 $ git pull . branch
1731 $ git merge branch
1732 -------------------------------------------------
1733
1734 are roughly equivalent. The former is actually very commonly used.
1735
1736 [[submitting-patches]]
1737 Submitting patches to a project
1738 -------------------------------
1739
1740 If you just have a few changes, the simplest way to submit them may
1741 just be to send them as patches in email:
1742
1743 First, use linkgit:git-format-patch[1]; for example:
1744
1745 -------------------------------------------------
1746 $ git format-patch origin
1747 -------------------------------------------------
1748
1749 will produce a numbered series of files in the current directory, one
1750 for each patch in the current branch but not in origin/HEAD.
1751
1752 You can then import these into your mail client and send them by
1753 hand. However, if you have a lot to send at once, you may prefer to
1754 use the linkgit:git-send-email[1] script to automate the process.
1755 Consult the mailing list for your project first to determine how they
1756 prefer such patches be handled.
1757
1758 [[importing-patches]]
1759 Importing patches to a project
1760 ------------------------------
1761
1762 Git also provides a tool called linkgit:git-am[1] (am stands for
1763 "apply mailbox"), for importing such an emailed series of patches.
1764 Just save all of the patch-containing messages, in order, into a
1765 single mailbox file, say "patches.mbox", then run
1766
1767 -------------------------------------------------
1768 $ git am -3 patches.mbox
1769 -------------------------------------------------
1770
1771 Git will apply each patch in order; if any conflicts are found, it
1772 will stop, and you can fix the conflicts as described in
1773 "<<resolving-a-merge,Resolving a merge>>". (The "-3" option tells
1774 git to perform a merge; if you would prefer it just to abort and
1775 leave your tree and index untouched, you may omit that option.)
1776
1777 Once the index is updated with the results of the conflict
1778 resolution, instead of creating a new commit, just run
1779
1780 -------------------------------------------------
1781 $ git am --resolved
1782 -------------------------------------------------
1783
1784 and git will create the commit for you and continue applying the
1785 remaining patches from the mailbox.
1786
1787 The final result will be a series of commits, one for each patch in
1788 the original mailbox, with authorship and commit log message each
1789 taken from the message containing each patch.
1790
1791 [[public-repositories]]
1792 Public git repositories
1793 -----------------------
1794
1795 Another way to submit changes to a project is to tell the maintainer
1796 of that project to pull the changes from your repository using
1797 linkgit:git-pull[1]. In the section "<<getting-updates-With-git-pull,
1798 Getting updates with `git pull`>>" we described this as a way to get
1799 updates from the "main" repository, but it works just as well in the
1800 other direction.
1801
1802 If you and the maintainer both have accounts on the same machine, then
1803 you can just pull changes from each other's repositories directly;
1804 commands that accept repository URLs as arguments will also accept a
1805 local directory name:
1806
1807 -------------------------------------------------
1808 $ git clone /path/to/repository
1809 $ git pull /path/to/other/repository
1810 -------------------------------------------------
1811
1812 or an ssh URL:
1813
1814 -------------------------------------------------
1815 $ git clone ssh://yourhost/~you/repository
1816 -------------------------------------------------
1817
1818 For projects with few developers, or for synchronizing a few private
1819 repositories, this may be all you need.
1820
1821 However, the more common way to do this is to maintain a separate public
1822 repository (usually on a different host) for others to pull changes
1823 from. This is usually more convenient, and allows you to cleanly
1824 separate private work in progress from publicly visible work.
1825
1826 You will continue to do your day-to-day work in your personal
1827 repository, but periodically "push" changes from your personal
1828 repository into your public repository, allowing other developers to
1829 pull from that repository. So the flow of changes, in a situation
1830 where there is one other developer with a public repository, looks
1831 like this:
1832
1833 you push
1834 your personal repo ------------------> your public repo
1835 ^ |
1836 | |
1837 | you pull | they pull
1838 | |
1839 | |
1840 | they push V
1841 their public repo <------------------- their repo
1842
1843 We explain how to do this in the following sections.
1844
1845 [[setting-up-a-public-repository]]
1846 Setting up a public repository
1847 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1848
1849 Assume your personal repository is in the directory ~/proj. We
1850 first create a new clone of the repository and tell `git daemon` that it
1851 is meant to be public:
1852
1853 -------------------------------------------------
1854 $ git clone --bare ~/proj proj.git
1855 $ touch proj.git/git-daemon-export-ok
1856 -------------------------------------------------
1857
1858 The resulting directory proj.git contains a "bare" git repository--it is
1859 just the contents of the ".git" directory, without any files checked out
1860 around it.
1861
1862 Next, copy proj.git to the server where you plan to host the
1863 public repository. You can use scp, rsync, or whatever is most
1864 convenient.
1865
1866 [[exporting-via-git]]
1867 Exporting a git repository via the git protocol
1868 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1869
1870 This is the preferred method.
1871
1872 If someone else administers the server, they should tell you what
1873 directory to put the repository in, and what git:// URL it will appear
1874 at. You can then skip to the section
1875 "<<pushing-changes-to-a-public-repository,Pushing changes to a public
1876 repository>>", below.
1877
1878 Otherwise, all you need to do is start linkgit:git-daemon[1]; it will
1879 listen on port 9418. By default, it will allow access to any directory
1880 that looks like a git directory and contains the magic file
1881 git-daemon-export-ok. Passing some directory paths as `git daemon`
1882 arguments will further restrict the exports to those paths.
1883
1884 You can also run `git daemon` as an inetd service; see the
1885 linkgit:git-daemon[1] man page for details. (See especially the
1886 examples section.)
1887
1888 [[exporting-via-http]]
1889 Exporting a git repository via http
1890 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1891
1892 The git protocol gives better performance and reliability, but on a
1893 host with a web server set up, http exports may be simpler to set up.
1894
1895 All you need to do is place the newly created bare git repository in
1896 a directory that is exported by the web server, and make some
1897 adjustments to give web clients some extra information they need:
1898
1899 -------------------------------------------------
1900 $ mv proj.git /home/you/public_html/proj.git
1901 $ cd proj.git
1902 $ git --bare update-server-info
1903 $ mv hooks/post-update.sample hooks/post-update
1904 -------------------------------------------------
1905
1906 (For an explanation of the last two lines, see
1907 linkgit:git-update-server-info[1] and linkgit:githooks[5].)
1908
1909 Advertise the URL of proj.git. Anybody else should then be able to
1910 clone or pull from that URL, for example with a command line like:
1911
1912 -------------------------------------------------
1913 $ git clone http://yourserver.com/~you/proj.git
1914 -------------------------------------------------
1915
1916 (See also
1917 link:howto/setup-git-server-over-http.txt[setup-git-server-over-http]
1918 for a slightly more sophisticated setup using WebDAV which also
1919 allows pushing over http.)
1920
1921 [[pushing-changes-to-a-public-repository]]
1922 Pushing changes to a public repository
1923 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1924
1925 Note that the two techniques outlined above (exporting via
1926 <<exporting-via-http,http>> or <<exporting-via-git,git>>) allow other
1927 maintainers to fetch your latest changes, but they do not allow write
1928 access, which you will need to update the public repository with the
1929 latest changes created in your private repository.
1930
1931 The simplest way to do this is using linkgit:git-push[1] and ssh; to
1932 update the remote branch named "master" with the latest state of your
1933 branch named "master", run
1934
1935 -------------------------------------------------
1936 $ git push ssh://yourserver.com/~you/proj.git master:master
1937 -------------------------------------------------
1938
1939 or just
1940
1941 -------------------------------------------------
1942 $ git push ssh://yourserver.com/~you/proj.git master
1943 -------------------------------------------------
1944
1945 As with `git fetch`, `git push` will complain if this does not result in a
1946 <<fast-forwards,fast forward>>; see the following section for details on
1947 handling this case.
1948
1949 Note that the target of a "push" is normally a
1950 <<def_bare_repository,bare>> repository. You can also push to a
1951 repository that has a checked-out working tree, but the working tree
1952 will not be updated by the push. This may lead to unexpected results if
1953 the branch you push to is the currently checked-out branch!
1954
1955 As with `git fetch`, you may also set up configuration options to
1956 save typing; so, for example, after
1957
1958 -------------------------------------------------
1959 $ cat >>.git/config <<EOF
1960 [remote "public-repo"]
1961 url = ssh://yourserver.com/~you/proj.git
1962 EOF
1963 -------------------------------------------------
1964
1965 you should be able to perform the above push with just
1966
1967 -------------------------------------------------
1968 $ git push public-repo master
1969 -------------------------------------------------
1970
1971 See the explanations of the remote.<name>.url, branch.<name>.remote,
1972 and remote.<name>.push options in linkgit:git-config[1] for
1973 details.
1974
1975 [[forcing-push]]
1976 What to do when a push fails
1977 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1978
1979 If a push would not result in a <<fast-forwards,fast forward>> of the
1980 remote branch, then it will fail with an error like:
1981
1982 -------------------------------------------------
1983 error: remote 'refs/heads/master' is not an ancestor of
1984 local 'refs/heads/master'.
1985 Maybe you are not up-to-date and need to pull first?
1986 error: failed to push to 'ssh://yourserver.com/~you/proj.git'
1987 -------------------------------------------------
1988
1989 This can happen, for example, if you:
1990
1991 - use `git reset --hard` to remove already-published commits, or
1992 - use `git commit --amend` to replace already-published commits
1993 (as in <<fixing-a-mistake-by-rewriting-history>>), or
1994 - use `git rebase` to rebase any already-published commits (as
1995 in <<using-git-rebase>>).
1996
1997 You may force `git push` to perform the update anyway by preceding the
1998 branch name with a plus sign:
1999
2000 -------------------------------------------------
2001 $ git push ssh://yourserver.com/~you/proj.git +master
2002 -------------------------------------------------
2003
2004 Normally whenever a branch head in a public repository is modified, it
2005 is modified to point to a descendant of the commit that it pointed to
2006 before. By forcing a push in this situation, you break that convention.
2007 (See <<problems-With-rewriting-history>>.)
2008
2009 Nevertheless, this is a common practice for people that need a simple
2010 way to publish a work-in-progress patch series, and it is an acceptable
2011 compromise as long as you warn other developers that this is how you
2012 intend to manage the branch.
2013
2014 It's also possible for a push to fail in this way when other people have
2015 the right to push to the same repository. In that case, the correct
2016 solution is to retry the push after first updating your work: either by a
2017 pull, or by a fetch followed by a rebase; see the
2018 <<setting-up-a-shared-repository,next section>> and
2019 linkgit:gitcvs-migration[7] for more.
2020
2021 [[setting-up-a-shared-repository]]
2022 Setting up a shared repository
2023 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2024
2025 Another way to collaborate is by using a model similar to that
2026 commonly used in CVS, where several developers with special rights
2027 all push to and pull from a single shared repository. See
2028 linkgit:gitcvs-migration[7] for instructions on how to
2029 set this up.
2030
2031 However, while there is nothing wrong with git's support for shared
2032 repositories, this mode of operation is not generally recommended,
2033 simply because the mode of collaboration that git supports--by
2034 exchanging patches and pulling from public repositories--has so many
2035 advantages over the central shared repository:
2036
2037 - Git's ability to quickly import and merge patches allows a
2038 single maintainer to process incoming changes even at very
2039 high rates. And when that becomes too much, `git pull` provides
2040 an easy way for that maintainer to delegate this job to other
2041 maintainers while still allowing optional review of incoming
2042 changes.
2043 - Since every developer's repository has the same complete copy
2044 of the project history, no repository is special, and it is
2045 trivial for another developer to take over maintenance of a
2046 project, either by mutual agreement, or because a maintainer
2047 becomes unresponsive or difficult to work with.
2048 - The lack of a central group of "committers" means there is
2049 less need for formal decisions about who is "in" and who is
2050 "out".
2051
2052 [[setting-up-gitweb]]
2053 Allowing web browsing of a repository
2054 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2055
2056 The gitweb cgi script provides users an easy way to browse your
2057 project's files and history without having to install git; see the file
2058 gitweb/INSTALL in the git source tree for instructions on setting it up.
2059
2060 [[sharing-development-examples]]
2061 Examples
2062 --------
2063
2064 [[maintaining-topic-branches]]
2065 Maintaining topic branches for a Linux subsystem maintainer
2066 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2067
2068 This describes how Tony Luck uses git in his role as maintainer of the
2069 IA64 architecture for the Linux kernel.
2070
2071 He uses two public branches:
2072
2073 - A "test" tree into which patches are initially placed so that they
2074 can get some exposure when integrated with other ongoing development.
2075 This tree is available to Andrew for pulling into -mm whenever he
2076 wants.
2077
2078 - A "release" tree into which tested patches are moved for final sanity
2079 checking, and as a vehicle to send them upstream to Linus (by sending
2080 him a "please pull" request.)
2081
2082 He also uses a set of temporary branches ("topic branches"), each
2083 containing a logical grouping of patches.
2084
2085 To set this up, first create your work tree by cloning Linus's public
2086 tree:
2087
2088 -------------------------------------------------
2089 $ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git work
2090 $ cd work
2091 -------------------------------------------------
2092
2093 Linus's tree will be stored in the remote branch named origin/master,
2094 and can be updated using linkgit:git-fetch[1]; you can track other
2095 public trees using linkgit:git-remote[1] to set up a "remote" and
2096 linkgit:git-fetch[1] to keep them up-to-date; see
2097 <<repositories-and-branches>>.
2098
2099 Now create the branches in which you are going to work; these start out
2100 at the current tip of origin/master branch, and should be set up (using
2101 the --track option to linkgit:git-branch[1]) to merge changes in from
2102 Linus by default.
2103
2104 -------------------------------------------------
2105 $ git branch --track test origin/master
2106 $ git branch --track release origin/master
2107 -------------------------------------------------
2108
2109 These can be easily kept up to date using linkgit:git-pull[1].
2110
2111 -------------------------------------------------
2112 $ git checkout test && git pull
2113 $ git checkout release && git pull
2114 -------------------------------------------------
2115
2116 Important note! If you have any local changes in these branches, then
2117 this merge will create a commit object in the history (with no local
2118 changes git will simply do a "Fast forward" merge). Many people dislike
2119 the "noise" that this creates in the Linux history, so you should avoid
2120 doing this capriciously in the "release" branch, as these noisy commits
2121 will become part of the permanent history when you ask Linus to pull
2122 from the release branch.
2123
2124 A few configuration variables (see linkgit:git-config[1]) can
2125 make it easy to push both branches to your public tree. (See
2126 <<setting-up-a-public-repository>>.)
2127
2128 -------------------------------------------------
2129 $ cat >> .git/config <<EOF
2130 [remote "mytree"]
2131 url = master.kernel.org:/pub/scm/linux/kernel/git/aegl/linux-2.6.git
2132 push = release
2133 push = test
2134 EOF
2135 -------------------------------------------------
2136
2137 Then you can push both the test and release trees using
2138 linkgit:git-push[1]:
2139
2140 -------------------------------------------------
2141 $ git push mytree
2142 -------------------------------------------------
2143
2144 or push just one of the test and release branches using:
2145
2146 -------------------------------------------------
2147 $ git push mytree test
2148 -------------------------------------------------
2149
2150 or
2151
2152 -------------------------------------------------
2153 $ git push mytree release
2154 -------------------------------------------------
2155
2156 Now to apply some patches from the community. Think of a short
2157 snappy name for a branch to hold this patch (or related group of
2158 patches), and create a new branch from the current tip of Linus's
2159 branch:
2160
2161 -------------------------------------------------
2162 $ git checkout -b speed-up-spinlocks origin
2163 -------------------------------------------------
2164
2165 Now you apply the patch(es), run some tests, and commit the change(s). If
2166 the patch is a multi-part series, then you should apply each as a separate
2167 commit to this branch.
2168
2169 -------------------------------------------------
2170 $ ... patch ... test ... commit [ ... patch ... test ... commit ]*
2171 -------------------------------------------------
2172
2173 When you are happy with the state of this change, you can pull it into the
2174 "test" branch in preparation to make it public:
2175
2176 -------------------------------------------------
2177 $ git checkout test && git pull . speed-up-spinlocks
2178 -------------------------------------------------
2179
2180 It is unlikely that you would have any conflicts here ... but you might if you
2181 spent a while on this step and had also pulled new versions from upstream.
2182
2183 Some time later when enough time has passed and testing done, you can pull the
2184 same branch into the "release" tree ready to go upstream. This is where you
2185 see the value of keeping each patch (or patch series) in its own branch. It
2186 means that the patches can be moved into the "release" tree in any order.
2187
2188 -------------------------------------------------
2189 $ git checkout release && git pull . speed-up-spinlocks
2190 -------------------------------------------------
2191
2192 After a while, you will have a number of branches, and despite the
2193 well chosen names you picked for each of them, you may forget what
2194 they are for, or what status they are in. To get a reminder of what
2195 changes are in a specific branch, use:
2196
2197 -------------------------------------------------
2198 $ git log linux..branchname | git shortlog
2199 -------------------------------------------------
2200
2201 To see whether it has already been merged into the test or release branches,
2202 use:
2203
2204 -------------------------------------------------
2205 $ git log test..branchname
2206 -------------------------------------------------
2207
2208 or
2209
2210 -------------------------------------------------
2211 $ git log release..branchname
2212 -------------------------------------------------
2213
2214 (If this branch has not yet been merged, you will see some log entries.
2215 If it has been merged, then there will be no output.)
2216
2217 Once a patch completes the great cycle (moving from test to release,
2218 then pulled by Linus, and finally coming back into your local
2219 "origin/master" branch), the branch for this change is no longer needed.
2220 You detect this when the output from:
2221
2222 -------------------------------------------------
2223 $ git log origin..branchname
2224 -------------------------------------------------
2225
2226 is empty. At this point the branch can be deleted:
2227
2228 -------------------------------------------------
2229 $ git branch -d branchname
2230 -------------------------------------------------
2231
2232 Some changes are so trivial that it is not necessary to create a separate
2233 branch and then merge into each of the test and release branches. For
2234 these changes, just apply directly to the "release" branch, and then
2235 merge that into the "test" branch.
2236
2237 To create diffstat and shortlog summaries of changes to include in a "please
2238 pull" request to Linus you can use:
2239
2240 -------------------------------------------------
2241 $ git diff --stat origin..release
2242 -------------------------------------------------
2243
2244 and
2245
2246 -------------------------------------------------
2247 $ git log -p origin..release | git shortlog
2248 -------------------------------------------------
2249
2250 Here are some of the scripts that simplify all this even further.
2251
2252 -------------------------------------------------
2253 ==== update script ====
2254 # Update a branch in my GIT tree. If the branch to be updated
2255 # is origin, then pull from kernel.org. Otherwise merge
2256 # origin/master branch into test|release branch
2257
2258 case "$1" in
2259 test|release)
2260 git checkout $1 && git pull . origin
2261 ;;
2262 origin)
2263 before=$(git rev-parse refs/remotes/origin/master)
2264 git fetch origin
2265 after=$(git rev-parse refs/remotes/origin/master)
2266 if [ $before != $after ]
2267 then
2268 git log $before..$after | git shortlog
2269 fi
2270 ;;
2271 *)
2272 echo "Usage: $0 origin|test|release" 1>&2
2273 exit 1
2274 ;;
2275 esac
2276 -------------------------------------------------
2277
2278 -------------------------------------------------
2279 ==== merge script ====
2280 # Merge a branch into either the test or release branch
2281
2282 pname=$0
2283
2284 usage()
2285 {
2286 echo "Usage: $pname branch test|release" 1>&2
2287 exit 1
2288 }
2289
2290 git show-ref -q --verify -- refs/heads/"$1" || {
2291 echo "Can't see branch <$1>" 1>&2
2292 usage
2293 }
2294
2295 case "$2" in
2296 test|release)
2297 if [ $(git log $2..$1 | wc -c) -eq 0 ]
2298 then
2299 echo $1 already merged into $2 1>&2
2300 exit 1
2301 fi
2302 git checkout $2 && git pull . $1
2303 ;;
2304 *)
2305 usage
2306 ;;
2307 esac
2308 -------------------------------------------------
2309
2310 -------------------------------------------------
2311 ==== status script ====
2312 # report on status of my ia64 GIT tree
2313
2314 gb=$(tput setab 2)
2315 rb=$(tput setab 1)
2316 restore=$(tput setab 9)
2317
2318 if [ `git rev-list test..release | wc -c` -gt 0 ]
2319 then
2320 echo $rb Warning: commits in release that are not in test $restore
2321 git log test..release
2322 fi
2323
2324 for branch in `git show-ref --heads | sed 's|^.*/||'`
2325 do
2326 if [ $branch = test -o $branch = release ]
2327 then
2328 continue
2329 fi
2330
2331 echo -n $gb ======= $branch ====== $restore " "
2332 status=
2333 for ref in test release origin/master
2334 do
2335 if [ `git rev-list $ref..$branch | wc -c` -gt 0 ]
2336 then
2337 status=$status${ref:0:1}
2338 fi
2339 done
2340 case $status in
2341 trl)
2342 echo $rb Need to pull into test $restore
2343 ;;
2344 rl)
2345 echo "In test"
2346 ;;
2347 l)
2348 echo "Waiting for linus"
2349 ;;
2350 "")
2351 echo $rb All done $restore
2352 ;;
2353 *)
2354 echo $rb "<$status>" $restore
2355 ;;
2356 esac
2357 git log origin/master..$branch | git shortlog
2358 done
2359 -------------------------------------------------
2360
2361
2362 [[cleaning-up-history]]
2363 Rewriting history and maintaining patch series
2364 ==============================================
2365
2366 Normally commits are only added to a project, never taken away or
2367 replaced. Git is designed with this assumption, and violating it will
2368 cause git's merge machinery (for example) to do the wrong thing.
2369
2370 However, there is a situation in which it can be useful to violate this
2371 assumption.
2372
2373 [[patch-series]]
2374 Creating the perfect patch series
2375 ---------------------------------
2376
2377 Suppose you are a contributor to a large project, and you want to add a
2378 complicated feature, and to present it to the other developers in a way
2379 that makes it easy for them to read your changes, verify that they are
2380 correct, and understand why you made each change.
2381
2382 If you present all of your changes as a single patch (or commit), they
2383 may find that it is too much to digest all at once.
2384
2385 If you present them with the entire history of your work, complete with
2386 mistakes, corrections, and dead ends, they may be overwhelmed.
2387
2388 So the ideal is usually to produce a series of patches such that:
2389
2390 1. Each patch can be applied in order.
2391
2392 2. Each patch includes a single logical change, together with a
2393 message explaining the change.
2394
2395 3. No patch introduces a regression: after applying any initial
2396 part of the series, the resulting project still compiles and
2397 works, and has no bugs that it didn't have before.
2398
2399 4. The complete series produces the same end result as your own
2400 (probably much messier!) development process did.
2401
2402 We will introduce some tools that can help you do this, explain how to
2403 use them, and then explain some of the problems that can arise because
2404 you are rewriting history.
2405
2406 [[using-git-rebase]]
2407 Keeping a patch series up to date using git rebase
2408 --------------------------------------------------
2409
2410 Suppose that you create a branch "mywork" on a remote-tracking branch
2411 "origin", and create some commits on top of it:
2412
2413 -------------------------------------------------
2414 $ git checkout -b mywork origin
2415 $ vi file.txt
2416 $ git commit
2417 $ vi otherfile.txt
2418 $ git commit
2419 ...
2420 -------------------------------------------------
2421
2422 You have performed no merges into mywork, so it is just a simple linear
2423 sequence of patches on top of "origin":
2424
2425 ................................................
2426 o--o--o <-- origin
2427 \
2428 o--o--o <-- mywork
2429 ................................................
2430
2431 Some more interesting work has been done in the upstream project, and
2432 "origin" has advanced:
2433
2434 ................................................
2435 o--o--O--o--o--o <-- origin
2436 \
2437 a--b--c <-- mywork
2438 ................................................
2439
2440 At this point, you could use "pull" to merge your changes back in;
2441 the result would create a new merge commit, like this:
2442
2443 ................................................
2444 o--o--O--o--o--o <-- origin
2445 \ \
2446 a--b--c--m <-- mywork
2447 ................................................
2448
2449 However, if you prefer to keep the history in mywork a simple series of
2450 commits without any merges, you may instead choose to use
2451 linkgit:git-rebase[1]:
2452
2453 -------------------------------------------------
2454 $ git checkout mywork
2455 $ git rebase origin
2456 -------------------------------------------------
2457
2458 This will remove each of your commits from mywork, temporarily saving
2459 them as patches (in a directory named ".git/rebase-apply"), update mywork to
2460 point at the latest version of origin, then apply each of the saved
2461 patches to the new mywork. The result will look like:
2462
2463
2464 ................................................
2465 o--o--O--o--o--o <-- origin
2466 \
2467 a'--b'--c' <-- mywork
2468 ................................................
2469
2470 In the process, it may discover conflicts. In that case it will stop
2471 and allow you to fix the conflicts; after fixing conflicts, use `git add`
2472 to update the index with those contents, and then, instead of
2473 running `git commit`, just run
2474
2475 -------------------------------------------------
2476 $ git rebase --continue
2477 -------------------------------------------------
2478
2479 and git will continue applying the rest of the patches.
2480
2481 At any point you may use the `--abort` option to abort this process and
2482 return mywork to the state it had before you started the rebase:
2483
2484 -------------------------------------------------
2485 $ git rebase --abort
2486 -------------------------------------------------
2487
2488 [[rewriting-one-commit]]
2489 Rewriting a single commit
2490 -------------------------
2491
2492 We saw in <<fixing-a-mistake-by-rewriting-history>> that you can replace the
2493 most recent commit using
2494
2495 -------------------------------------------------
2496 $ git commit --amend
2497 -------------------------------------------------
2498
2499 which will replace the old commit by a new commit incorporating your
2500 changes, giving you a chance to edit the old commit message first.
2501
2502 You can also use a combination of this and linkgit:git-rebase[1] to
2503 replace a commit further back in your history and recreate the
2504 intervening changes on top of it. First, tag the problematic commit
2505 with
2506
2507 -------------------------------------------------
2508 $ git tag bad mywork~5
2509 -------------------------------------------------
2510
2511 (Either gitk or `git log` may be useful for finding the commit.)
2512
2513 Then check out that commit, edit it, and rebase the rest of the series
2514 on top of it (note that we could check out the commit on a temporary
2515 branch, but instead we're using a <<detached-head,detached head>>):
2516
2517 -------------------------------------------------
2518 $ git checkout bad
2519 $ # make changes here and update the index
2520 $ git commit --amend
2521 $ git rebase --onto HEAD bad mywork
2522 -------------------------------------------------
2523
2524 When you're done, you'll be left with mywork checked out, with the top
2525 patches on mywork reapplied on top of your modified commit. You can
2526 then clean up with
2527
2528 -------------------------------------------------
2529 $ git tag -d bad
2530 -------------------------------------------------
2531
2532 Note that the immutable nature of git history means that you haven't really
2533 "modified" existing commits; instead, you have replaced the old commits with
2534 new commits having new object names.
2535
2536 [[reordering-patch-series]]
2537 Reordering or selecting from a patch series
2538 -------------------------------------------
2539
2540 Given one existing commit, the linkgit:git-cherry-pick[1] command
2541 allows you to apply the change introduced by that commit and create a
2542 new commit that records it. So, for example, if "mywork" points to a
2543 series of patches on top of "origin", you might do something like:
2544
2545 -------------------------------------------------
2546 $ git checkout -b mywork-new origin
2547 $ gitk origin..mywork &
2548 -------------------------------------------------
2549
2550 and browse through the list of patches in the mywork branch using gitk,
2551 applying them (possibly in a different order) to mywork-new using
2552 cherry-pick, and possibly modifying them as you go using `git commit --amend`.
2553 The linkgit:git-gui[1] command may also help as it allows you to
2554 individually select diff hunks for inclusion in the index (by
2555 right-clicking on the diff hunk and choosing "Stage Hunk for Commit").
2556
2557 Another technique is to use `git format-patch` to create a series of
2558 patches, then reset the state to before the patches:
2559
2560 -------------------------------------------------
2561 $ git format-patch origin
2562 $ git reset --hard origin
2563 -------------------------------------------------
2564
2565 Then modify, reorder, or eliminate patches as preferred before applying
2566 them again with linkgit:git-am[1].
2567
2568 [[patch-series-tools]]
2569 Other tools
2570 -----------
2571
2572 There are numerous other tools, such as StGIT, which exist for the
2573 purpose of maintaining a patch series. These are outside of the scope of
2574 this manual.
2575
2576 [[problems-With-rewriting-history]]
2577 Problems with rewriting history
2578 -------------------------------
2579
2580 The primary problem with rewriting the history of a branch has to do
2581 with merging. Suppose somebody fetches your branch and merges it into
2582 their branch, with a result something like this:
2583
2584 ................................................
2585 o--o--O--o--o--o <-- origin
2586 \ \
2587 t--t--t--m <-- their branch:
2588 ................................................
2589
2590 Then suppose you modify the last three commits:
2591
2592 ................................................
2593 o--o--o <-- new head of origin
2594 /
2595 o--o--O--o--o--o <-- old head of origin
2596 ................................................
2597
2598 If we examined all this history together in one repository, it will
2599 look like:
2600
2601 ................................................
2602 o--o--o <-- new head of origin
2603 /
2604 o--o--O--o--o--o <-- old head of origin
2605 \ \
2606 t--t--t--m <-- their branch:
2607 ................................................
2608
2609 Git has no way of knowing that the new head is an updated version of
2610 the old head; it treats this situation exactly the same as it would if
2611 two developers had independently done the work on the old and new heads
2612 in parallel. At this point, if someone attempts to merge the new head
2613 in to their branch, git will attempt to merge together the two (old and
2614 new) lines of development, instead of trying to replace the old by the
2615 new. The results are likely to be unexpected.
2616
2617 You may still choose to publish branches whose history is rewritten,
2618 and it may be useful for others to be able to fetch those branches in
2619 order to examine or test them, but they should not attempt to pull such
2620 branches into their own work.
2621
2622 For true distributed development that supports proper merging,
2623 published branches should never be rewritten.
2624
2625 [[bisect-merges]]
2626 Why bisecting merge commits can be harder than bisecting linear history
2627 -----------------------------------------------------------------------
2628
2629 The linkgit:git-bisect[1] command correctly handles history that
2630 includes merge commits. However, when the commit that it finds is a
2631 merge commit, the user may need to work harder than usual to figure out
2632 why that commit introduced a problem.
2633
2634 Imagine this history:
2635
2636 ................................................
2637 ---Z---o---X---...---o---A---C---D
2638 \ /
2639 o---o---Y---...---o---B
2640 ................................................
2641
2642 Suppose that on the upper line of development, the meaning of one
2643 of the functions that exists at Z is changed at commit X. The
2644 commits from Z leading to A change both the function's
2645 implementation and all calling sites that exist at Z, as well
2646 as new calling sites they add, to be consistent. There is no
2647 bug at A.
2648
2649 Suppose that in the meantime on the lower line of development somebody
2650 adds a new calling site for that function at commit Y. The
2651 commits from Z leading to B all assume the old semantics of that
2652 function and the callers and the callee are consistent with each
2653 other. There is no bug at B, either.
2654
2655 Suppose further that the two development lines merge cleanly at C,
2656 so no conflict resolution is required.
2657
2658 Nevertheless, the code at C is broken, because the callers added
2659 on the lower line of development have not been converted to the new
2660 semantics introduced on the upper line of development. So if all
2661 you know is that D is bad, that Z is good, and that
2662 linkgit:git-bisect[1] identifies C as the culprit, how will you
2663 figure out that the problem is due to this change in semantics?
2664
2665 When the result of a `git bisect` is a non-merge commit, you should
2666 normally be able to discover the problem by examining just that commit.
2667 Developers can make this easy by breaking their changes into small
2668 self-contained commits. That won't help in the case above, however,
2669 because the problem isn't obvious from examination of any single
2670 commit; instead, a global view of the development is required. To
2671 make matters worse, the change in semantics in the problematic
2672 function may be just one small part of the changes in the upper
2673 line of development.
2674
2675 On the other hand, if instead of merging at C you had rebased the
2676 history between Z to B on top of A, you would have gotten this
2677 linear history:
2678
2679 ................................................................
2680 ---Z---o---X--...---o---A---o---o---Y*--...---o---B*--D*
2681 ................................................................
2682
2683 Bisecting between Z and D* would hit a single culprit commit Y*,
2684 and understanding why Y* was broken would probably be easier.
2685
2686 Partly for this reason, many experienced git users, even when
2687 working on an otherwise merge-heavy project, keep the history
2688 linear by rebasing against the latest upstream version before
2689 publishing.
2690
2691 [[advanced-branch-management]]
2692 Advanced branch management
2693 ==========================
2694
2695 [[fetching-individual-branches]]
2696 Fetching individual branches
2697 ----------------------------
2698
2699 Instead of using linkgit:git-remote[1], you can also choose just
2700 to update one branch at a time, and to store it locally under an
2701 arbitrary name:
2702
2703 -------------------------------------------------
2704 $ git fetch origin todo:my-todo-work
2705 -------------------------------------------------
2706
2707 The first argument, "origin", just tells git to fetch from the
2708 repository you originally cloned from. The second argument tells git
2709 to fetch the branch named "todo" from the remote repository, and to
2710 store it locally under the name refs/heads/my-todo-work.
2711
2712 You can also fetch branches from other repositories; so
2713
2714 -------------------------------------------------
2715 $ git fetch git://example.com/proj.git master:example-master
2716 -------------------------------------------------
2717
2718 will create a new branch named "example-master" and store in it the
2719 branch named "master" from the repository at the given URL. If you
2720 already have a branch named example-master, it will attempt to
2721 <<fast-forwards,fast-forward>> to the commit given by example.com's
2722 master branch. In more detail:
2723
2724 [[fetch-fast-forwards]]
2725 git fetch and fast-forwards
2726 ---------------------------
2727
2728 In the previous example, when updating an existing branch, "git fetch"
2729 checks to make sure that the most recent commit on the remote
2730 branch is a descendant of the most recent commit on your copy of the
2731 branch before updating your copy of the branch to point at the new
2732 commit. Git calls this process a <<fast-forwards,fast forward>>.
2733
2734 A fast forward looks something like this:
2735
2736 ................................................
2737 o--o--o--o <-- old head of the branch
2738 \
2739 o--o--o <-- new head of the branch
2740 ................................................
2741
2742
2743 In some cases it is possible that the new head will *not* actually be
2744 a descendant of the old head. For example, the developer may have
2745 realized she made a serious mistake, and decided to backtrack,
2746 resulting in a situation like:
2747
2748 ................................................
2749 o--o--o--o--a--b <-- old head of the branch
2750 \
2751 o--o--o <-- new head of the branch
2752 ................................................
2753
2754 In this case, "git fetch" will fail, and print out a warning.
2755
2756 In that case, you can still force git to update to the new head, as
2757 described in the following section. However, note that in the
2758 situation above this may mean losing the commits labeled "a" and "b",
2759 unless you've already created a reference of your own pointing to
2760 them.
2761
2762 [[forcing-fetch]]
2763 Forcing git fetch to do non-fast-forward updates
2764 ------------------------------------------------
2765
2766 If git fetch fails because the new head of a branch is not a
2767 descendant of the old head, you may force the update with:
2768
2769 -------------------------------------------------
2770 $ git fetch git://example.com/proj.git +master:refs/remotes/example/master
2771 -------------------------------------------------
2772
2773 Note the addition of the "+" sign. Alternatively, you can use the "-f"
2774 flag to force updates of all the fetched branches, as in:
2775
2776 -------------------------------------------------
2777 $ git fetch -f origin
2778 -------------------------------------------------
2779
2780 Be aware that commits that the old version of example/master pointed at
2781 may be lost, as we saw in the previous section.
2782
2783 [[remote-branch-configuration]]
2784 Configuring remote branches
2785 ---------------------------
2786
2787 We saw above that "origin" is just a shortcut to refer to the
2788 repository that you originally cloned from. This information is
2789 stored in git configuration variables, which you can see using
2790 linkgit:git-config[1]:
2791
2792 -------------------------------------------------
2793 $ git config -l
2794 core.repositoryformatversion=0
2795 core.filemode=true
2796 core.logallrefupdates=true
2797 remote.origin.url=git://git.kernel.org/pub/scm/git/git.git
2798 remote.origin.fetch=+refs/heads/*:refs/remotes/origin/*
2799 branch.master.remote=origin
2800 branch.master.merge=refs/heads/master
2801 -------------------------------------------------
2802
2803 If there are other repositories that you also use frequently, you can
2804 create similar configuration options to save typing; for example,
2805 after
2806
2807 -------------------------------------------------
2808 $ git config remote.example.url git://example.com/proj.git
2809 -------------------------------------------------
2810
2811 then the following two commands will do the same thing:
2812
2813 -------------------------------------------------
2814 $ git fetch git://example.com/proj.git master:refs/remotes/example/master
2815 $ git fetch example master:refs/remotes/example/master
2816 -------------------------------------------------
2817
2818 Even better, if you add one more option:
2819
2820 -------------------------------------------------
2821 $ git config remote.example.fetch master:refs/remotes/example/master
2822 -------------------------------------------------
2823
2824 then the following commands will all do the same thing:
2825
2826 -------------------------------------------------
2827 $ git fetch git://example.com/proj.git master:refs/remotes/example/master
2828 $ git fetch example master:refs/remotes/example/master
2829 $ git fetch example
2830 -------------------------------------------------
2831
2832 You can also add a "+" to force the update each time:
2833
2834 -------------------------------------------------
2835 $ git config remote.example.fetch +master:ref/remotes/example/master
2836 -------------------------------------------------
2837
2838 Don't do this unless you're sure you won't mind "git fetch" possibly
2839 throwing away commits on 'example/master'.
2840
2841 Also note that all of the above configuration can be performed by
2842 directly editing the file .git/config instead of using
2843 linkgit:git-config[1].
2844
2845 See linkgit:git-config[1] for more details on the configuration
2846 options mentioned above.
2847
2848
2849 [[git-concepts]]
2850 Git concepts
2851 ============
2852
2853 Git is built on a small number of simple but powerful ideas. While it
2854 is possible to get things done without understanding them, you will find
2855 git much more intuitive if you do.
2856
2857 We start with the most important, the <<def_object_database,object
2858 database>> and the <<def_index,index>>.
2859
2860 [[the-object-database]]
2861 The Object Database
2862 -------------------
2863
2864
2865 We already saw in <<understanding-commits>> that all commits are stored
2866 under a 40-digit "object name". In fact, all the information needed to
2867 represent the history of a project is stored in objects with such names.
2868 In each case the name is calculated by taking the SHA1 hash of the
2869 contents of the object. The SHA1 hash is a cryptographic hash function.
2870 What that means to us is that it is impossible to find two different
2871 objects with the same name. This has a number of advantages; among
2872 others:
2873
2874 - Git can quickly determine whether two objects are identical or not,
2875 just by comparing names.
2876 - Since object names are computed the same way in every repository, the
2877 same content stored in two repositories will always be stored under
2878 the same name.
2879 - Git can detect errors when it reads an object, by checking that the
2880 object's name is still the SHA1 hash of its contents.
2881
2882 (See <<object-details>> for the details of the object formatting and
2883 SHA1 calculation.)
2884
2885 There are four different types of objects: "blob", "tree", "commit", and
2886 "tag".
2887
2888 - A <<def_blob_object,"blob" object>> is used to store file data.
2889 - A <<def_tree_object,"tree" object>> ties one or more
2890 "blob" objects into a directory structure. In addition, a tree object
2891 can refer to other tree objects, thus creating a directory hierarchy.
2892 - A <<def_commit_object,"commit" object>> ties such directory hierarchies
2893 together into a <<def_DAG,directed acyclic graph>> of revisions--each
2894 commit contains the object name of exactly one tree designating the
2895 directory hierarchy at the time of the commit. In addition, a commit
2896 refers to "parent" commit objects that describe the history of how we
2897 arrived at that directory hierarchy.
2898 - A <<def_tag_object,"tag" object>> symbolically identifies and can be
2899 used to sign other objects. It contains the object name and type of
2900 another object, a symbolic name (of course!) and, optionally, a
2901 signature.
2902
2903 The object types in some more detail:
2904
2905 [[commit-object]]
2906 Commit Object
2907 ~~~~~~~~~~~~~
2908
2909 The "commit" object links a physical state of a tree with a description
2910 of how we got there and why. Use the --pretty=raw option to
2911 linkgit:git-show[1] or linkgit:git-log[1] to examine your favorite
2912 commit:
2913
2914 ------------------------------------------------
2915 $ git show -s --pretty=raw 2be7fcb476
2916 commit 2be7fcb4764f2dbcee52635b91fedb1b3dcf7ab4
2917 tree fb3a8bdd0ceddd019615af4d57a53f43d8cee2bf
2918 parent 257a84d9d02e90447b149af58b271c19405edb6a
2919 author Dave Watson <dwatson@mimvista.com> 1187576872 -0400
2920 committer Junio C Hamano <gitster@pobox.com> 1187591163 -0700
2921
2922 Fix misspelling of 'suppress' in docs
2923
2924 Signed-off-by: Junio C Hamano <gitster@pobox.com>
2925 ------------------------------------------------
2926
2927 As you can see, a commit is defined by:
2928
2929 - a tree: The SHA1 name of a tree object (as defined below), representing
2930 the contents of a directory at a certain point in time.
2931 - parent(s): The SHA1 name of some number of commits which represent the
2932 immediately previous step(s) in the history of the project. The
2933 example above has one parent; merge commits may have more than
2934 one. A commit with no parents is called a "root" commit, and
2935 represents the initial revision of a project. Each project must have
2936 at least one root. A project can also have multiple roots, though
2937 that isn't common (or necessarily a good idea).
2938 - an author: The name of the person responsible for this change, together
2939 with its date.
2940 - a committer: The name of the person who actually created the commit,
2941 with the date it was done. This may be different from the author, for
2942 example, if the author was someone who wrote a patch and emailed it
2943 to the person who used it to create the commit.
2944 - a comment describing this commit.
2945
2946 Note that a commit does not itself contain any information about what
2947 actually changed; all changes are calculated by comparing the contents
2948 of the tree referred to by this commit with the trees associated with
2949 its parents. In particular, git does not attempt to record file renames
2950 explicitly, though it can identify cases where the existence of the same
2951 file data at changing paths suggests a rename. (See, for example, the
2952 -M option to linkgit:git-diff[1]).
2953
2954 A commit is usually created by linkgit:git-commit[1], which creates a
2955 commit whose parent is normally the current HEAD, and whose tree is
2956 taken from the content currently stored in the index.
2957
2958 [[tree-object]]
2959 Tree Object
2960 ~~~~~~~~~~~
2961
2962 The ever-versatile linkgit:git-show[1] command can also be used to
2963 examine tree objects, but linkgit:git-ls-tree[1] will give you more
2964 details:
2965
2966 ------------------------------------------------
2967 $ git ls-tree fb3a8bdd0ce
2968 100644 blob 63c918c667fa005ff12ad89437f2fdc80926e21c .gitignore
2969 100644 blob 5529b198e8d14decbe4ad99db3f7fb632de0439d .mailmap
2970 100644 blob 6ff87c4664981e4397625791c8ea3bbb5f2279a3 COPYING
2971 040000 tree 2fb783e477100ce076f6bf57e4a6f026013dc745 Documentation
2972 100755 blob 3c0032cec592a765692234f1cba47dfdcc3a9200 GIT-VERSION-GEN
2973 100644 blob 289b046a443c0647624607d471289b2c7dcd470b INSTALL
2974 100644 blob 4eb463797adc693dc168b926b6932ff53f17d0b1 Makefile
2975 100644 blob 548142c327a6790ff8821d67c2ee1eff7a656b52 README
2976 ...
2977 ------------------------------------------------
2978
2979 As you can see, a tree object contains a list of entries, each with a
2980 mode, object type, SHA1 name, and name, sorted by name. It represents
2981 the contents of a single directory tree.
2982
2983 The object type may be a blob, representing the contents of a file, or
2984 another tree, representing the contents of a subdirectory. Since trees
2985 and blobs, like all other objects, are named by the SHA1 hash of their
2986 contents, two trees have the same SHA1 name if and only if their
2987 contents (including, recursively, the contents of all subdirectories)
2988 are identical. This allows git to quickly determine the differences
2989 between two related tree objects, since it can ignore any entries with
2990 identical object names.
2991
2992 (Note: in the presence of submodules, trees may also have commits as
2993 entries. See <<submodules>> for documentation.)
2994
2995 Note that the files all have mode 644 or 755: git actually only pays
2996 attention to the executable bit.
2997
2998 [[blob-object]]
2999 Blob Object
3000 ~~~~~~~~~~~
3001
3002 You can use linkgit:git-show[1] to examine the contents of a blob; take,
3003 for example, the blob in the entry for "COPYING" from the tree above:
3004
3005 ------------------------------------------------
3006 $ git show 6ff87c4664
3007
3008 Note that the only valid version of the GPL as far as this project
3009 is concerned is _this_ particular version of the license (ie v2, not
3010 v2.2 or v3.x or whatever), unless explicitly otherwise stated.
3011 ...
3012 ------------------------------------------------
3013
3014 A "blob" object is nothing but a binary blob of data. It doesn't refer
3015 to anything else or have attributes of any kind.
3016
3017 Since the blob is entirely defined by its data, if two files in a
3018 directory tree (or in multiple different versions of the repository)
3019 have the same contents, they will share the same blob object. The object
3020 is totally independent of its location in the directory tree, and
3021 renaming a file does not change the object that file is associated with.
3022
3023 Note that any tree or blob object can be examined using
3024 linkgit:git-show[1] with the <revision>:<path> syntax. This can
3025 sometimes be useful for browsing the contents of a tree that is not
3026 currently checked out.
3027
3028 [[trust]]
3029 Trust
3030 ~~~~~
3031
3032 If you receive the SHA1 name of a blob from one source, and its contents
3033 from another (possibly untrusted) source, you can still trust that those
3034 contents are correct as long as the SHA1 name agrees. This is because
3035 the SHA1 is designed so that it is infeasible to find different contents
3036 that produce the same hash.
3037
3038 Similarly, you need only trust the SHA1 name of a top-level tree object
3039 to trust the contents of the entire directory that it refers to, and if
3040 you receive the SHA1 name of a commit from a trusted source, then you
3041 can easily verify the entire history of commits reachable through
3042 parents of that commit, and all of those contents of the trees referred
3043 to by those commits.
3044
3045 So to introduce some real trust in the system, the only thing you need
3046 to do is to digitally sign just 'one' special note, which includes the
3047 name of a top-level commit. Your digital signature shows others
3048 that you trust that commit, and the immutability of the history of
3049 commits tells others that they can trust the whole history.
3050
3051 In other words, you can easily validate a whole archive by just
3052 sending out a single email that tells the people the name (SHA1 hash)
3053 of the top commit, and digitally sign that email using something
3054 like GPG/PGP.
3055
3056 To assist in this, git also provides the tag object...
3057
3058 [[tag-object]]
3059 Tag Object
3060 ~~~~~~~~~~
3061
3062 A tag object contains an object, object type, tag name, the name of the
3063 person ("tagger") who created the tag, and a message, which may contain
3064 a signature, as can be seen using linkgit:git-cat-file[1]:
3065
3066 ------------------------------------------------
3067 $ git cat-file tag v1.5.0
3068 object 437b1b20df4b356c9342dac8d38849f24ef44f27
3069 type commit
3070 tag v1.5.0
3071 tagger Junio C Hamano <junkio@cox.net> 1171411200 +0000
3072
3073 GIT 1.5.0
3074 -----BEGIN PGP SIGNATURE-----
3075 Version: GnuPG v1.4.6 (GNU/Linux)
3076
3077 iD8DBQBF0lGqwMbZpPMRm5oRAuRiAJ9ohBLd7s2kqjkKlq1qqC57SbnmzQCdG4ui
3078 nLE/L9aUXdWeTFPron96DLA=
3079 =2E+0
3080 -----END PGP SIGNATURE-----
3081 ------------------------------------------------
3082
3083 See the linkgit:git-tag[1] command to learn how to create and verify tag
3084 objects. (Note that linkgit:git-tag[1] can also be used to create
3085 "lightweight tags", which are not tag objects at all, but just simple
3086 references whose names begin with "refs/tags/").
3087
3088 [[pack-files]]
3089 How git stores objects efficiently: pack files
3090 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3091
3092 Newly created objects are initially created in a file named after the
3093 object's SHA1 hash (stored in .git/objects).
3094
3095 Unfortunately this system becomes inefficient once a project has a
3096 lot of objects. Try this on an old project:
3097
3098 ------------------------------------------------
3099 $ git count-objects
3100 6930 objects, 47620 kilobytes
3101 ------------------------------------------------
3102
3103 The first number is the number of objects which are kept in
3104 individual files. The second is the amount of space taken up by
3105 those "loose" objects.
3106
3107 You can save space and make git faster by moving these loose objects in
3108 to a "pack file", which stores a group of objects in an efficient
3109 compressed format; the details of how pack files are formatted can be
3110 found in link:technical/pack-format.txt[technical/pack-format.txt].
3111
3112 To put the loose objects into a pack, just run git repack:
3113
3114 ------------------------------------------------
3115 $ git repack
3116 Generating pack...
3117 Done counting 6020 objects.
3118 Deltifying 6020 objects.
3119 100% (6020/6020) done
3120 Writing 6020 objects.
3121 100% (6020/6020) done
3122 Total 6020, written 6020 (delta 4070), reused 0 (delta 0)
3123 Pack pack-3e54ad29d5b2e05838c75df582c65257b8d08e1c created.
3124 ------------------------------------------------
3125
3126 You can then run
3127
3128 ------------------------------------------------
3129 $ git prune
3130 ------------------------------------------------
3131
3132 to remove any of the "loose" objects that are now contained in the
3133 pack. This will also remove any unreferenced objects (which may be
3134 created when, for example, you use "git reset" to remove a commit).
3135 You can verify that the loose objects are gone by looking at the
3136 .git/objects directory or by running
3137
3138 ------------------------------------------------
3139 $ git count-objects
3140 0 objects, 0 kilobytes
3141 ------------------------------------------------
3142
3143 Although the object files are gone, any commands that refer to those
3144 objects will work exactly as they did before.
3145
3146 The linkgit:git-gc[1] command performs packing, pruning, and more for
3147 you, so is normally the only high-level command you need.
3148
3149 [[dangling-objects]]
3150 Dangling objects
3151 ~~~~~~~~~~~~~~~~
3152
3153 The linkgit:git-fsck[1] command will sometimes complain about dangling
3154 objects. They are not a problem.
3155
3156 The most common cause of dangling objects is that you've rebased a
3157 branch, or you have pulled from somebody else who rebased a branch--see
3158 <<cleaning-up-history>>. In that case, the old head of the original
3159 branch still exists, as does everything it pointed to. The branch
3160 pointer itself just doesn't, since you replaced it with another one.
3161
3162 There are also other situations that cause dangling objects. For
3163 example, a "dangling blob" may arise because you did a "git add" of a
3164 file, but then, before you actually committed it and made it part of the
3165 bigger picture, you changed something else in that file and committed
3166 that *updated* thing--the old state that you added originally ends up
3167 not being pointed to by any commit or tree, so it's now a dangling blob
3168 object.
3169
3170 Similarly, when the "recursive" merge strategy runs, and finds that
3171 there are criss-cross merges and thus more than one merge base (which is
3172 fairly unusual, but it does happen), it will generate one temporary
3173 midway tree (or possibly even more, if you had lots of criss-crossing
3174 merges and more than two merge bases) as a temporary internal merge
3175 base, and again, those are real objects, but the end result will not end
3176 up pointing to them, so they end up "dangling" in your repository.
3177
3178 Generally, dangling objects aren't anything to worry about. They can
3179 even be very useful: if you screw something up, the dangling objects can
3180 be how you recover your old tree (say, you did a rebase, and realized
3181 that you really didn't want to--you can look at what dangling objects
3182 you have, and decide to reset your head to some old dangling state).
3183
3184 For commits, you can just use:
3185
3186 ------------------------------------------------
3187 $ gitk <dangling-commit-sha-goes-here> --not --all
3188 ------------------------------------------------
3189
3190 This asks for all the history reachable from the given commit but not
3191 from any branch, tag, or other reference. If you decide it's something
3192 you want, you can always create a new reference to it, e.g.,
3193
3194 ------------------------------------------------
3195 $ git branch recovered-branch <dangling-commit-sha-goes-here>
3196 ------------------------------------------------
3197
3198 For blobs and trees, you can't do the same, but you can still examine
3199 them. You can just do
3200
3201 ------------------------------------------------
3202 $ git show <dangling-blob/tree-sha-goes-here>
3203 ------------------------------------------------
3204
3205 to show what the contents of the blob were (or, for a tree, basically
3206 what the "ls" for that directory was), and that may give you some idea
3207 of what the operation was that left that dangling object.
3208
3209 Usually, dangling blobs and trees aren't very interesting. They're
3210 almost always the result of either being a half-way mergebase (the blob
3211 will often even have the conflict markers from a merge in it, if you
3212 have had conflicting merges that you fixed up by hand), or simply
3213 because you interrupted a "git fetch" with ^C or something like that,
3214 leaving _some_ of the new objects in the object database, but just
3215 dangling and useless.
3216
3217 Anyway, once you are sure that you're not interested in any dangling
3218 state, you can just prune all unreachable objects:
3219
3220 ------------------------------------------------
3221 $ git prune
3222 ------------------------------------------------
3223
3224 and they'll be gone. But you should only run "git prune" on a quiescent
3225 repository--it's kind of like doing a filesystem fsck recovery: you
3226 don't want to do that while the filesystem is mounted.
3227
3228 (The same is true of "git fsck" itself, btw, but since
3229 `git fsck` never actually *changes* the repository, it just reports
3230 on what it found, `git fsck` itself is never 'dangerous' to run.
3231 Running it while somebody is actually changing the repository can cause
3232 confusing and scary messages, but it won't actually do anything bad. In
3233 contrast, running "git prune" while somebody is actively changing the
3234 repository is a *BAD* idea).
3235
3236 [[recovering-from-repository-corruption]]
3237 Recovering from repository corruption
3238 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3239
3240 By design, git treats data trusted to it with caution. However, even in
3241 the absence of bugs in git itself, it is still possible that hardware or
3242 operating system errors could corrupt data.
3243
3244 The first defense against such problems is backups. You can back up a
3245 git directory using clone, or just using cp, tar, or any other backup
3246 mechanism.
3247
3248 As a last resort, you can search for the corrupted objects and attempt
3249 to replace them by hand. Back up your repository before attempting this
3250 in case you corrupt things even more in the process.
3251
3252 We'll assume that the problem is a single missing or corrupted blob,
3253 which is sometimes a solvable problem. (Recovering missing trees and
3254 especially commits is *much* harder).
3255
3256 Before starting, verify that there is corruption, and figure out where
3257 it is with linkgit:git-fsck[1]; this may be time-consuming.
3258
3259 Assume the output looks like this:
3260
3261 ------------------------------------------------
3262 $ git fsck --full
3263 broken link from tree 2d9263c6d23595e7cb2a21e5ebbb53655278dff8
3264 to blob 4b9458b3786228369c63936db65827de3cc06200
3265 missing blob 4b9458b3786228369c63936db65827de3cc06200
3266 ------------------------------------------------
3267
3268 (Typically there will be some "dangling object" messages too, but they
3269 aren't interesting.)
3270
3271 Now you know that blob 4b9458b3 is missing, and that the tree 2d9263c6
3272 points to it. If you could find just one copy of that missing blob
3273 object, possibly in some other repository, you could move it into
3274 .git/objects/4b/9458b3... and be done. Suppose you can't. You can
3275 still examine the tree that pointed to it with linkgit:git-ls-tree[1],
3276 which might output something like:
3277
3278 ------------------------------------------------
3279 $ git ls-tree 2d9263c6d23595e7cb2a21e5ebbb53655278dff8
3280 100644 blob 8d14531846b95bfa3564b58ccfb7913a034323b8 .gitignore
3281 100644 blob ebf9bf84da0aab5ed944264a5db2a65fe3a3e883 .mailmap
3282 100644 blob ca442d313d86dc67e0a2e5d584b465bd382cbf5c COPYING
3283 ...
3284 100644 blob 4b9458b3786228369c63936db65827de3cc06200 myfile
3285 ...
3286 ------------------------------------------------
3287
3288 So now you know that the missing blob was the data for a file named
3289 "myfile". And chances are you can also identify the directory--let's
3290 say it's in "somedirectory". If you're lucky the missing copy might be
3291 the same as the copy you have checked out in your working tree at
3292 "somedirectory/myfile"; you can test whether that's right with
3293 linkgit:git-hash-object[1]:
3294
3295 ------------------------------------------------
3296 $ git hash-object -w somedirectory/myfile
3297 ------------------------------------------------
3298
3299 which will create and store a blob object with the contents of
3300 somedirectory/myfile, and output the sha1 of that object. if you're
3301 extremely lucky it might be 4b9458b3786228369c63936db65827de3cc06200, in
3302 which case you've guessed right, and the corruption is fixed!
3303
3304 Otherwise, you need more information. How do you tell which version of
3305 the file has been lost?
3306
3307 The easiest way to do this is with:
3308
3309 ------------------------------------------------
3310 $ git log --raw --all --full-history -- somedirectory/myfile
3311 ------------------------------------------------
3312
3313 Because you're asking for raw output, you'll now get something like
3314
3315 ------------------------------------------------
3316 commit abc
3317 Author:
3318 Date:
3319 ...
3320 :100644 100644 4b9458b... newsha... M somedirectory/myfile
3321
3322
3323 commit xyz
3324 Author:
3325 Date:
3326
3327 ...
3328 :100644 100644 oldsha... 4b9458b... M somedirectory/myfile
3329 ------------------------------------------------
3330
3331 This tells you that the immediately preceding version of the file was
3332 "newsha", and that the immediately following version was "oldsha".
3333 You also know the commit messages that went with the change from oldsha
3334 to 4b9458b and with the change from 4b9458b to newsha.
3335
3336 If you've been committing small enough changes, you may now have a good
3337 shot at reconstructing the contents of the in-between state 4b9458b.
3338
3339 If you can do that, you can now recreate the missing object with
3340
3341 ------------------------------------------------
3342 $ git hash-object -w <recreated-file>
3343 ------------------------------------------------
3344
3345 and your repository is good again!
3346
3347 (Btw, you could have ignored the fsck, and started with doing a
3348
3349 ------------------------------------------------
3350 $ git log --raw --all
3351 ------------------------------------------------
3352
3353 and just looked for the sha of the missing object (4b9458b..) in that
3354 whole thing. It's up to you - git does *have* a lot of information, it is
3355 just missing one particular blob version.
3356
3357 [[the-index]]
3358 The index
3359 -----------
3360
3361 The index is a binary file (generally kept in .git/index) containing a
3362 sorted list of path names, each with permissions and the SHA1 of a blob
3363 object; linkgit:git-ls-files[1] can show you the contents of the index:
3364
3365 -------------------------------------------------
3366 $ git ls-files --stage
3367 100644 63c918c667fa005ff12ad89437f2fdc80926e21c 0 .gitignore
3368 100644 5529b198e8d14decbe4ad99db3f7fb632de0439d 0 .mailmap
3369 100644 6ff87c4664981e4397625791c8ea3bbb5f2279a3 0 COPYING
3370 100644 a37b2152bd26be2c2289e1f57a292534a51a93c7 0 Documentation/.gitignore
3371 100644 fbefe9a45b00a54b58d94d06eca48b03d40a50e0 0 Documentation/Makefile
3372 ...
3373 100644 2511aef8d89ab52be5ec6a5e46236b4b6bcd07ea 0 xdiff/xtypes.h
3374 100644 2ade97b2574a9f77e7ae4002a4e07a6a38e46d07 0 xdiff/xutils.c
3375 100644 d5de8292e05e7c36c4b68857c1cf9855e3d2f70a 0 xdiff/xutils.h
3376 -------------------------------------------------
3377
3378 Note that in older documentation you may see the index called the
3379 "current directory cache" or just the "cache". It has three important
3380 properties:
3381
3382 1. The index contains all the information necessary to generate a single
3383 (uniquely determined) tree object.
3384 +
3385 For example, running linkgit:git-commit[1] generates this tree object
3386 from the index, stores it in the object database, and uses it as the
3387 tree object associated with the new commit.
3388
3389 2. The index enables fast comparisons between the tree object it defines
3390 and the working tree.
3391 +
3392 It does this by storing some additional data for each entry (such as
3393 the last modified time). This data is not displayed above, and is not
3394 stored in the created tree object, but it can be used to determine
3395 quickly which files in the working directory differ from what was
3396 stored in the index, and thus save git from having to read all of the
3397 data from such files to look for changes.
3398
3399 3. It can efficiently represent information about merge conflicts
3400 between different tree objects, allowing each pathname to be
3401 associated with sufficient information about the trees involved that
3402 you can create a three-way merge between them.
3403 +
3404 We saw in <<conflict-resolution>> that during a merge the index can
3405 store multiple versions of a single file (called "stages"). The third
3406 column in the linkgit:git-ls-files[1] output above is the stage
3407 number, and will take on values other than 0 for files with merge
3408 conflicts.
3409
3410 The index is thus a sort of temporary staging area, which is filled with
3411 a tree which you are in the process of working on.
3412
3413 If you blow the index away entirely, you generally haven't lost any
3414 information as long as you have the name of the tree that it described.
3415
3416 [[submodules]]
3417 Submodules
3418 ==========
3419
3420 Large projects are often composed of smaller, self-contained modules. For
3421 example, an embedded Linux distribution's source tree would include every
3422 piece of software in the distribution with some local modifications; a movie
3423 player might need to build against a specific, known-working version of a
3424 decompression library; several independent programs might all share the same
3425 build scripts.
3426
3427 With centralized revision control systems this is often accomplished by
3428 including every module in one single repository. Developers can check out
3429 all modules or only the modules they need to work with. They can even modify
3430 files across several modules in a single commit while moving things around
3431 or updating APIs and translations.
3432
3433 Git does not allow partial checkouts, so duplicating this approach in Git
3434 would force developers to keep a local copy of modules they are not
3435 interested in touching. Commits in an enormous checkout would be slower
3436 than you'd expect as Git would have to scan every directory for changes.
3437 If modules have a lot of local history, clones would take forever.
3438
3439 On the plus side, distributed revision control systems can much better
3440 integrate with external sources. In a centralized model, a single arbitrary
3441 snapshot of the external project is exported from its own revision control
3442 and then imported into the local revision control on a vendor branch. All
3443 the history is hidden. With distributed revision control you can clone the
3444 entire external history and much more easily follow development and re-merge
3445 local changes.
3446
3447 Git's submodule support allows a repository to contain, as a subdirectory, a
3448 checkout of an external project. Submodules maintain their own identity;
3449 the submodule support just stores the submodule repository location and
3450 commit ID, so other developers who clone the containing project
3451 ("superproject") can easily clone all the submodules at the same revision.
3452 Partial checkouts of the superproject are possible: you can tell Git to
3453 clone none, some or all of the submodules.
3454
3455 The linkgit:git-submodule[1] command is available since Git 1.5.3. Users
3456 with Git 1.5.2 can look up the submodule commits in the repository and
3457 manually check them out; earlier versions won't recognize the submodules at
3458 all.
3459
3460 To see how submodule support works, create (for example) four example
3461 repositories that can be used later as a submodule:
3462
3463 -------------------------------------------------
3464 $ mkdir ~/git
3465 $ cd ~/git
3466 $ for i in a b c d
3467 do
3468 mkdir $i
3469 cd $i
3470 git init
3471 echo "module $i" > $i.txt
3472 git add $i.txt
3473 git commit -m "Initial commit, submodule $i"
3474 cd ..
3475 done
3476 -------------------------------------------------
3477
3478 Now create the superproject and add all the submodules:
3479
3480 -------------------------------------------------
3481 $ mkdir super
3482 $ cd super
3483 $ git init
3484 $ for i in a b c d
3485 do
3486 git submodule add ~/git/$i $i
3487 done
3488 -------------------------------------------------
3489
3490 NOTE: Do not use local URLs here if you plan to publish your superproject!
3491
3492 See what files `git submodule` created:
3493
3494 -------------------------------------------------
3495 $ ls -a
3496 . .. .git .gitmodules a b c d
3497 -------------------------------------------------
3498
3499 The `git submodule add <repo> <path>` command does a couple of things:
3500
3501 - It clones the submodule from <repo> to the given <path> under the
3502 current directory and by default checks out the master branch.
3503 - It adds the submodule's clone path to the linkgit:gitmodules[5] file and
3504 adds this file to the index, ready to be committed.
3505 - It adds the submodule's current commit ID to the index, ready to be
3506 committed.
3507
3508 Commit the superproject:
3509
3510 -------------------------------------------------
3511 $ git commit -m "Add submodules a, b, c and d."
3512 -------------------------------------------------
3513
3514 Now clone the superproject:
3515
3516 -------------------------------------------------
3517 $ cd ..
3518 $ git clone super cloned
3519 $ cd cloned
3520 -------------------------------------------------
3521
3522 The submodule directories are there, but they're empty:
3523
3524 -------------------------------------------------
3525 $ ls -a a
3526 . ..
3527 $ git submodule status
3528 -d266b9873ad50488163457f025db7cdd9683d88b a
3529 -e81d457da15309b4fef4249aba9b50187999670d b
3530 -c1536a972b9affea0f16e0680ba87332dc059146 c
3531 -d96249ff5d57de5de093e6baff9e0aafa5276a74 d
3532 -------------------------------------------------
3533
3534 NOTE: The commit object names shown above would be different for you, but they
3535 should match the HEAD commit object names of your repositories. You can check
3536 it by running `git ls-remote ../a`.
3537
3538 Pulling down the submodules is a two-step process. First run `git submodule
3539 init` to add the submodule repository URLs to `.git/config`:
3540
3541 -------------------------------------------------
3542 $ git submodule init
3543 -------------------------------------------------
3544
3545 Now use `git submodule update` to clone the repositories and check out the
3546 commits specified in the superproject:
3547
3548 -------------------------------------------------
3549 $ git submodule update
3550 $ cd a
3551 $ ls -a
3552 . .. .git a.txt
3553 -------------------------------------------------
3554
3555 One major difference between `git submodule update` and `git submodule add` is
3556 that `git submodule update` checks out a specific commit, rather than the tip
3557 of a branch. It's like checking out a tag: the head is detached, so you're not
3558 working on a branch.
3559
3560 -------------------------------------------------
3561 $ git branch
3562 * (no branch)
3563 master
3564 -------------------------------------------------
3565
3566 If you want to make a change within a submodule and you have a detached head,
3567 then you should create or checkout a branch, make your changes, publish the
3568 change within the submodule, and then update the superproject to reference the
3569 new commit:
3570
3571 -------------------------------------------------
3572 $ git checkout master
3573 -------------------------------------------------
3574
3575 or
3576
3577 -------------------------------------------------
3578 $ git checkout -b fix-up
3579 -------------------------------------------------
3580
3581 then
3582
3583 -------------------------------------------------
3584 $ echo "adding a line again" >> a.txt
3585 $ git commit -a -m "Updated the submodule from within the superproject."
3586 $ git push
3587 $ cd ..
3588 $ git diff
3589 diff --git a/a b/a
3590 index d266b98..261dfac 160000
3591 --- a/a
3592 +++ b/a
3593 @@ -1 +1 @@
3594 -Subproject commit d266b9873ad50488163457f025db7cdd9683d88b
3595 +Subproject commit 261dfac35cb99d380eb966e102c1197139f7fa24
3596 $ git add a
3597 $ git commit -m "Updated submodule a."
3598 $ git push
3599 -------------------------------------------------
3600
3601 You have to run `git submodule update` after `git pull` if you want to update
3602 submodules, too.
3603
3604 Pitfalls with submodules
3605 ------------------------
3606
3607 Always publish the submodule change before publishing the change to the
3608 superproject that references it. If you forget to publish the submodule change,
3609 others won't be able to clone the repository:
3610
3611 -------------------------------------------------
3612 $ cd ~/git/super/a
3613 $ echo i added another line to this file >> a.txt
3614 $ git commit -a -m "doing it wrong this time"
3615 $ cd ..
3616 $ git add a
3617 $ git commit -m "Updated submodule a again."
3618 $ git push
3619 $ cd ~/git/cloned
3620 $ git pull
3621 $ git submodule update
3622 error: pathspec '261dfac35cb99d380eb966e102c1197139f7fa24' did not match any file(s) known to git.
3623 Did you forget to 'git add'?
3624 Unable to checkout '261dfac35cb99d380eb966e102c1197139f7fa24' in submodule path 'a'
3625 -------------------------------------------------
3626
3627 You also should not rewind branches in a submodule beyond commits that were
3628 ever recorded in any superproject.
3629
3630 It's not safe to run `git submodule update` if you've made and committed
3631 changes within a submodule without checking out a branch first. They will be
3632 silently overwritten:
3633
3634 -------------------------------------------------
3635 $ cat a.txt
3636 module a
3637 $ echo line added from private2 >> a.txt
3638 $ git commit -a -m "line added inside private2"
3639 $ cd ..
3640 $ git submodule update
3641 Submodule path 'a': checked out 'd266b9873ad50488163457f025db7cdd9683d88b'
3642 $ cd a
3643 $ cat a.txt
3644 module a
3645 -------------------------------------------------
3646
3647 NOTE: The changes are still visible in the submodule's reflog.
3648
3649 This is not the case if you did not commit your changes.
3650
3651 [[low-level-operations]]
3652 Low-level git operations
3653 ========================
3654
3655 Many of the higher-level commands were originally implemented as shell
3656 scripts using a smaller core of low-level git commands. These can still
3657 be useful when doing unusual things with git, or just as a way to
3658 understand its inner workings.
3659
3660 [[object-manipulation]]
3661 Object access and manipulation
3662 ------------------------------
3663
3664 The linkgit:git-cat-file[1] command can show the contents of any object,
3665 though the higher-level linkgit:git-show[1] is usually more useful.
3666
3667 The linkgit:git-commit-tree[1] command allows constructing commits with
3668 arbitrary parents and trees.
3669
3670 A tree can be created with linkgit:git-write-tree[1] and its data can be
3671 accessed by linkgit:git-ls-tree[1]. Two trees can be compared with
3672 linkgit:git-diff-tree[1].
3673
3674 A tag is created with linkgit:git-mktag[1], and the signature can be
3675 verified by linkgit:git-verify-tag[1], though it is normally simpler to
3676 use linkgit:git-tag[1] for both.
3677
3678 [[the-workflow]]
3679 The Workflow
3680 ------------
3681
3682 High-level operations such as linkgit:git-commit[1],
3683 linkgit:git-checkout[1] and linkgit:git-reset[1] work by moving data
3684 between the working tree, the index, and the object database. Git
3685 provides low-level operations which perform each of these steps
3686 individually.
3687
3688 Generally, all "git" operations work on the index file. Some operations
3689 work *purely* on the index file (showing the current state of the
3690 index), but most operations move data between the index file and either
3691 the database or the working directory. Thus there are four main
3692 combinations:
3693
3694 [[working-directory-to-index]]
3695 working directory -> index
3696 ~~~~~~~~~~~~~~~~~~~~~~~~~~
3697
3698 The linkgit:git-update-index[1] command updates the index with
3699 information from the working directory. You generally update the
3700 index information by just specifying the filename you want to update,
3701 like so:
3702
3703 -------------------------------------------------
3704 $ git update-index filename
3705 -------------------------------------------------
3706
3707 but to avoid common mistakes with filename globbing etc, the command
3708 will not normally add totally new entries or remove old entries,
3709 i.e. it will normally just update existing cache entries.
3710
3711 To tell git that yes, you really do realize that certain files no
3712 longer exist, or that new files should be added, you
3713 should use the `--remove` and `--add` flags respectively.
3714
3715 NOTE! A `--remove` flag does 'not' mean that subsequent filenames will
3716 necessarily be removed: if the files still exist in your directory
3717 structure, the index will be updated with their new status, not
3718 removed. The only thing `--remove` means is that update-index will be
3719 considering a removed file to be a valid thing, and if the file really
3720 does not exist any more, it will update the index accordingly.
3721
3722 As a special case, you can also do `git update-index --refresh`, which
3723 will refresh the "stat" information of each index to match the current
3724 stat information. It will 'not' update the object status itself, and
3725 it will only update the fields that are used to quickly test whether
3726 an object still matches its old backing store object.
3727
3728 The previously introduced linkgit:git-add[1] is just a wrapper for
3729 linkgit:git-update-index[1].
3730
3731 [[index-to-object-database]]
3732 index -> object database
3733 ~~~~~~~~~~~~~~~~~~~~~~~~
3734
3735 You write your current index file to a "tree" object with the program
3736
3737 -------------------------------------------------
3738 $ git write-tree
3739 -------------------------------------------------
3740
3741 that doesn't come with any options--it will just write out the
3742 current index into the set of tree objects that describe that state,
3743 and it will return the name of the resulting top-level tree. You can
3744 use that tree to re-generate the index at any time by going in the
3745 other direction:
3746
3747 [[object-database-to-index]]
3748 object database -> index
3749 ~~~~~~~~~~~~~~~~~~~~~~~~
3750
3751 You read a "tree" file from the object database, and use that to
3752 populate (and overwrite--don't do this if your index contains any
3753 unsaved state that you might want to restore later!) your current
3754 index. Normal operation is just
3755
3756 -------------------------------------------------
3757 $ git read-tree <sha1 of tree>
3758 -------------------------------------------------
3759
3760 and your index file will now be equivalent to the tree that you saved
3761 earlier. However, that is only your 'index' file: your working
3762 directory contents have not been modified.
3763
3764 [[index-to-working-directory]]
3765 index -> working directory
3766 ~~~~~~~~~~~~~~~~~~~~~~~~~~
3767
3768 You update your working directory from the index by "checking out"
3769 files. This is not a very common operation, since normally you'd just
3770 keep your files updated, and rather than write to your working
3771 directory, you'd tell the index files about the changes in your
3772 working directory (i.e. `git update-index`).
3773
3774 However, if you decide to jump to a new version, or check out somebody
3775 else's version, or just restore a previous tree, you'd populate your
3776 index file with read-tree, and then you need to check out the result
3777 with
3778
3779 -------------------------------------------------
3780 $ git checkout-index filename
3781 -------------------------------------------------
3782
3783 or, if you want to check out all of the index, use `-a`.
3784
3785 NOTE! `git checkout-index` normally refuses to overwrite old files, so
3786 if you have an old version of the tree already checked out, you will
3787 need to use the "-f" flag ('before' the "-a" flag or the filename) to
3788 'force' the checkout.
3789
3790
3791 Finally, there are a few odds and ends which are not purely moving
3792 from one representation to the other:
3793
3794 [[tying-it-all-together]]
3795 Tying it all together
3796 ~~~~~~~~~~~~~~~~~~~~~
3797
3798 To commit a tree you have instantiated with "git write-tree", you'd
3799 create a "commit" object that refers to that tree and the history
3800 behind it--most notably the "parent" commits that preceded it in
3801 history.
3802
3803 Normally a "commit" has one parent: the previous state of the tree
3804 before a certain change was made. However, sometimes it can have two
3805 or more parent commits, in which case we call it a "merge", due to the
3806 fact that such a commit brings together ("merges") two or more
3807 previous states represented by other commits.
3808
3809 In other words, while a "tree" represents a particular directory state
3810 of a working directory, a "commit" represents that state in "time",
3811 and explains how we got there.
3812
3813 You create a commit object by giving it the tree that describes the
3814 state at the time of the commit, and a list of parents:
3815
3816 -------------------------------------------------
3817 $ git commit-tree <tree> -p <parent> [-p <parent2> ..]
3818 -------------------------------------------------
3819
3820 and then giving the reason for the commit on stdin (either through
3821 redirection from a pipe or file, or by just typing it at the tty).
3822
3823 `git commit-tree` will return the name of the object that represents
3824 that commit, and you should save it away for later use. Normally,
3825 you'd commit a new `HEAD` state, and while git doesn't care where you
3826 save the note about that state, in practice we tend to just write the
3827 result to the file pointed at by `.git/HEAD`, so that we can always see
3828 what the last committed state was.
3829
3830 Here is an ASCII art by Jon Loeliger that illustrates how
3831 various pieces fit together.
3832
3833 ------------
3834
3835 commit-tree
3836 commit obj
3837 +----+
3838 | |
3839 | |
3840 V V
3841 +-----------+
3842 | Object DB |
3843 | Backing |
3844 | Store |
3845 +-----------+
3846 ^
3847 write-tree | |
3848 tree obj | |
3849 | | read-tree
3850 | | tree obj
3851 V
3852 +-----------+
3853 | Index |
3854 | "cache" |
3855 +-----------+
3856 update-index ^
3857 blob obj | |
3858 | |
3859 checkout-index -u | | checkout-index
3860 stat | | blob obj
3861 V
3862 +-----------+
3863 | Working |
3864 | Directory |
3865 +-----------+
3866
3867 ------------
3868
3869
3870 [[examining-the-data]]
3871 Examining the data
3872 ------------------
3873
3874 You can examine the data represented in the object database and the
3875 index with various helper tools. For every object, you can use
3876 linkgit:git-cat-file[1] to examine details about the
3877 object:
3878
3879 -------------------------------------------------
3880 $ git cat-file -t <objectname>
3881 -------------------------------------------------
3882
3883 shows the type of the object, and once you have the type (which is
3884 usually implicit in where you find the object), you can use
3885
3886 -------------------------------------------------
3887 $ git cat-file blob|tree|commit|tag <objectname>
3888 -------------------------------------------------
3889
3890 to show its contents. NOTE! Trees have binary content, and as a result
3891 there is a special helper for showing that content, called
3892 `git ls-tree`, which turns the binary content into a more easily
3893 readable form.
3894
3895 It's especially instructive to look at "commit" objects, since those
3896 tend to be small and fairly self-explanatory. In particular, if you
3897 follow the convention of having the top commit name in `.git/HEAD`,
3898 you can do
3899
3900 -------------------------------------------------
3901 $ git cat-file commit HEAD
3902 -------------------------------------------------
3903
3904 to see what the top commit was.
3905
3906 [[merging-multiple-trees]]
3907 Merging multiple trees
3908 ----------------------
3909
3910 Git helps you do a three-way merge, which you can expand to n-way by
3911 repeating the merge procedure arbitrary times until you finally
3912 "commit" the state. The normal situation is that you'd only do one
3913 three-way merge (two parents), and commit it, but if you like to, you
3914 can do multiple parents in one go.
3915
3916 To do a three-way merge, you need the two sets of "commit" objects
3917 that you want to merge, use those to find the closest common parent (a
3918 third "commit" object), and then use those commit objects to find the
3919 state of the directory ("tree" object) at these points.
3920
3921 To get the "base" for the merge, you first look up the common parent
3922 of two commits with
3923
3924 -------------------------------------------------
3925 $ git merge-base <commit1> <commit2>
3926 -------------------------------------------------
3927
3928 which will return you the commit they are both based on. You should
3929 now look up the "tree" objects of those commits, which you can easily
3930 do with (for example)
3931
3932 -------------------------------------------------
3933 $ git cat-file commit <commitname> | head -1
3934 -------------------------------------------------
3935
3936 since the tree object information is always the first line in a commit
3937 object.
3938
3939 Once you know the three trees you are going to merge (the one "original"
3940 tree, aka the common tree, and the two "result" trees, aka the branches
3941 you want to merge), you do a "merge" read into the index. This will
3942 complain if it has to throw away your old index contents, so you should
3943 make sure that you've committed those--in fact you would normally
3944 always do a merge against your last commit (which should thus match what
3945 you have in your current index anyway).
3946
3947 To do the merge, do
3948
3949 -------------------------------------------------
3950 $ git read-tree -m -u <origtree> <yourtree> <targettree>
3951 -------------------------------------------------
3952
3953 which will do all trivial merge operations for you directly in the
3954 index file, and you can just write the result out with
3955 `git write-tree`.
3956
3957
3958 [[merging-multiple-trees-2]]
3959 Merging multiple trees, continued
3960 ---------------------------------
3961
3962 Sadly, many merges aren't trivial. If there are files that have
3963 been added, moved or removed, or if both branches have modified the
3964 same file, you will be left with an index tree that contains "merge
3965 entries" in it. Such an index tree can 'NOT' be written out to a tree
3966 object, and you will have to resolve any such merge clashes using
3967 other tools before you can write out the result.
3968
3969 You can examine such index state with `git ls-files --unmerged`
3970 command. An example:
3971
3972 ------------------------------------------------
3973 $ git read-tree -m $orig HEAD $target
3974 $ git ls-files --unmerged
3975 100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1 hello.c
3976 100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2 hello.c
3977 100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello.c
3978 ------------------------------------------------
3979
3980 Each line of the `git ls-files --unmerged` output begins with
3981 the blob mode bits, blob SHA1, 'stage number', and the
3982 filename. The 'stage number' is git's way to say which tree it
3983 came from: stage 1 corresponds to `$orig` tree, stage 2 `HEAD`
3984 tree, and stage3 `$target` tree.
3985
3986 Earlier we said that trivial merges are done inside
3987 `git read-tree -m`. For example, if the file did not change
3988 from `$orig` to `HEAD` nor `$target`, or if the file changed
3989 from `$orig` to `HEAD` and `$orig` to `$target` the same way,
3990 obviously the final outcome is what is in `HEAD`. What the
3991 above example shows is that file `hello.c` was changed from
3992 `$orig` to `HEAD` and `$orig` to `$target` in a different way.
3993 You could resolve this by running your favorite 3-way merge
3994 program, e.g. `diff3`, `merge`, or git's own merge-file, on
3995 the blob objects from these three stages yourself, like this:
3996
3997 ------------------------------------------------
3998 $ git cat-file blob 263414f... >hello.c~1
3999 $ git cat-file blob 06fa6a2... >hello.c~2
4000 $ git cat-file blob cc44c73... >hello.c~3
4001 $ git merge-file hello.c~2 hello.c~1 hello.c~3
4002 ------------------------------------------------
4003
4004 This would leave the merge result in `hello.c~2` file, along
4005 with conflict markers if there are conflicts. After verifying
4006 the merge result makes sense, you can tell git what the final
4007 merge result for this file is by:
4008
4009 -------------------------------------------------
4010 $ mv -f hello.c~2 hello.c
4011 $ git update-index hello.c
4012 -------------------------------------------------
4013
4014 When a path is in the "unmerged" state, running `git update-index` for
4015 that path tells git to mark the path resolved.
4016
4017 The above is the description of a git merge at the lowest level,
4018 to help you understand what conceptually happens under the hood.
4019 In practice, nobody, not even git itself, runs `git cat-file` three times
4020 for this. There is a `git merge-index` program that extracts the
4021 stages to temporary files and calls a "merge" script on it:
4022
4023 -------------------------------------------------
4024 $ git merge-index git-merge-one-file hello.c
4025 -------------------------------------------------
4026
4027 and that is what higher level `git merge -s resolve` is implemented with.
4028
4029 [[hacking-git]]
4030 Hacking git
4031 ===========
4032
4033 This chapter covers internal details of the git implementation which
4034 probably only git developers need to understand.
4035
4036 [[object-details]]
4037 Object storage format
4038 ---------------------
4039
4040 All objects have a statically determined "type" which identifies the
4041 format of the object (i.e. how it is used, and how it can refer to other
4042 objects). There are currently four different object types: "blob",
4043 "tree", "commit", and "tag".
4044
4045 Regardless of object type, all objects share the following
4046 characteristics: they are all deflated with zlib, and have a header
4047 that not only specifies their type, but also provides size information
4048 about the data in the object. It's worth noting that the SHA1 hash
4049 that is used to name the object is the hash of the original data
4050 plus this header, so `sha1sum` 'file' does not match the object name
4051 for 'file'.
4052 (Historical note: in the dawn of the age of git the hash
4053 was the sha1 of the 'compressed' object.)
4054
4055 As a result, the general consistency of an object can always be tested
4056 independently of the contents or the type of the object: all objects can
4057 be validated by verifying that (a) their hashes match the content of the
4058 file and (b) the object successfully inflates to a stream of bytes that
4059 forms a sequence of <ascii type without space> {plus} <space> {plus} <ascii decimal
4060 size> {plus} <byte\0> {plus} <binary object data>.
4061
4062 The structured objects can further have their structure and
4063 connectivity to other objects verified. This is generally done with
4064 the `git fsck` program, which generates a full dependency graph
4065 of all objects, and verifies their internal consistency (in addition
4066 to just verifying their superficial consistency through the hash).
4067
4068 [[birdview-on-the-source-code]]
4069 A birds-eye view of Git's source code
4070 -------------------------------------
4071
4072 It is not always easy for new developers to find their way through Git's
4073 source code. This section gives you a little guidance to show where to
4074 start.
4075
4076 A good place to start is with the contents of the initial commit, with:
4077
4078 ----------------------------------------------------
4079 $ git checkout e83c5163
4080 ----------------------------------------------------
4081
4082 The initial revision lays the foundation for almost everything git has
4083 today, but is small enough to read in one sitting.
4084
4085 Note that terminology has changed since that revision. For example, the
4086 README in that revision uses the word "changeset" to describe what we
4087 now call a <<def_commit_object,commit>>.
4088
4089 Also, we do not call it "cache" any more, but rather "index"; however, the
4090 file is still called `cache.h`. Remark: Not much reason to change it now,
4091 especially since there is no good single name for it anyway, because it is
4092 basically _the_ header file which is included by _all_ of Git's C sources.
4093
4094 If you grasp the ideas in that initial commit, you should check out a
4095 more recent version and skim `cache.h`, `object.h` and `commit.h`.
4096
4097 In the early days, Git (in the tradition of UNIX) was a bunch of programs
4098 which were extremely simple, and which you used in scripts, piping the
4099 output of one into another. This turned out to be good for initial
4100 development, since it was easier to test new things. However, recently
4101 many of these parts have become builtins, and some of the core has been
4102 "libified", i.e. put into libgit.a for performance, portability reasons,
4103 and to avoid code duplication.
4104
4105 By now, you know what the index is (and find the corresponding data
4106 structures in `cache.h`), and that there are just a couple of object types
4107 (blobs, trees, commits and tags) which inherit their common structure from
4108 `struct object`, which is their first member (and thus, you can cast e.g.
4109 `(struct object *)commit` to achieve the _same_ as `&commit->object`, i.e.
4110 get at the object name and flags).
4111
4112 Now is a good point to take a break to let this information sink in.
4113
4114 Next step: get familiar with the object naming. Read <<naming-commits>>.
4115 There are quite a few ways to name an object (and not only revisions!).
4116 All of these are handled in `sha1_name.c`. Just have a quick look at
4117 the function `get_sha1()`. A lot of the special handling is done by
4118 functions like `get_sha1_basic()` or the likes.
4119
4120 This is just to get you into the groove for the most libified part of Git:
4121 the revision walker.
4122
4123 Basically, the initial version of `git log` was a shell script:
4124
4125 ----------------------------------------------------------------
4126 $ git-rev-list --pretty $(git-rev-parse --default HEAD "$@") | \
4127 LESS=-S ${PAGER:-less}
4128 ----------------------------------------------------------------
4129
4130 What does this mean?
4131
4132 `git rev-list` is the original version of the revision walker, which
4133 _always_ printed a list of revisions to stdout. It is still functional,
4134 and needs to, since most new Git programs start out as scripts using
4135 `git rev-list`.
4136
4137 `git rev-parse` is not as important any more; it was only used to filter out
4138 options that were relevant for the different plumbing commands that were
4139 called by the script.
4140
4141 Most of what `git rev-list` did is contained in `revision.c` and
4142 `revision.h`. It wraps the options in a struct named `rev_info`, which
4143 controls how and what revisions are walked, and more.
4144
4145 The original job of `git rev-parse` is now taken by the function
4146 `setup_revisions()`, which parses the revisions and the common command line
4147 options for the revision walker. This information is stored in the struct
4148 `rev_info` for later consumption. You can do your own command line option
4149 parsing after calling `setup_revisions()`. After that, you have to call
4150 `prepare_revision_walk()` for initialization, and then you can get the
4151 commits one by one with the function `get_revision()`.
4152
4153 If you are interested in more details of the revision walking process,
4154 just have a look at the first implementation of `cmd_log()`; call
4155 `git show v1.3.0{tilde}155^2{tilde}4` and scroll down to that function (note that you
4156 no longer need to call `setup_pager()` directly).
4157
4158 Nowadays, `git log` is a builtin, which means that it is _contained_ in the
4159 command `git`. The source side of a builtin is
4160
4161 - a function called `cmd_<bla>`, typically defined in `builtin-<bla>.c`,
4162 and declared in `builtin.h`,
4163
4164 - an entry in the `commands[]` array in `git.c`, and
4165
4166 - an entry in `BUILTIN_OBJECTS` in the `Makefile`.
4167
4168 Sometimes, more than one builtin is contained in one source file. For
4169 example, `cmd_whatchanged()` and `cmd_log()` both reside in `builtin-log.c`,
4170 since they share quite a bit of code. In that case, the commands which are
4171 _not_ named like the `.c` file in which they live have to be listed in
4172 `BUILT_INS` in the `Makefile`.
4173
4174 `git log` looks more complicated in C than it does in the original script,
4175 but that allows for a much greater flexibility and performance.
4176
4177 Here again it is a good point to take a pause.
4178
4179 Lesson three is: study the code. Really, it is the best way to learn about
4180 the organization of Git (after you know the basic concepts).
4181
4182 So, think about something which you are interested in, say, "how can I
4183 access a blob just knowing the object name of it?". The first step is to
4184 find a Git command with which you can do it. In this example, it is either
4185 `git show` or `git cat-file`.
4186
4187 For the sake of clarity, let's stay with `git cat-file`, because it
4188
4189 - is plumbing, and
4190
4191 - was around even in the initial commit (it literally went only through
4192 some 20 revisions as `cat-file.c`, was renamed to `builtin-cat-file.c`
4193 when made a builtin, and then saw less than 10 versions).
4194
4195 So, look into `builtin-cat-file.c`, search for `cmd_cat_file()` and look what
4196 it does.
4197
4198 ------------------------------------------------------------------
4199 git_config(git_default_config);
4200 if (argc != 3)
4201 usage("git cat-file [-t|-s|-e|-p|<type>] <sha1>");
4202 if (get_sha1(argv[2], sha1))
4203 die("Not a valid object name %s", argv[2]);
4204 ------------------------------------------------------------------
4205
4206 Let's skip over the obvious details; the only really interesting part
4207 here is the call to `get_sha1()`. It tries to interpret `argv[2]` as an
4208 object name, and if it refers to an object which is present in the current
4209 repository, it writes the resulting SHA-1 into the variable `sha1`.
4210
4211 Two things are interesting here:
4212
4213 - `get_sha1()` returns 0 on _success_. This might surprise some new
4214 Git hackers, but there is a long tradition in UNIX to return different
4215 negative numbers in case of different errors--and 0 on success.
4216
4217 - the variable `sha1` in the function signature of `get_sha1()` is `unsigned
4218 char \*`, but is actually expected to be a pointer to `unsigned
4219 char[20]`. This variable will contain the 160-bit SHA-1 of the given
4220 commit. Note that whenever a SHA-1 is passed as `unsigned char \*`, it
4221 is the binary representation, as opposed to the ASCII representation in
4222 hex characters, which is passed as `char *`.
4223
4224 You will see both of these things throughout the code.
4225
4226 Now, for the meat:
4227
4228 -----------------------------------------------------------------------------
4229 case 0:
4230 buf = read_object_with_reference(sha1, argv[1], &size, NULL);
4231 -----------------------------------------------------------------------------
4232
4233 This is how you read a blob (actually, not only a blob, but any type of
4234 object). To know how the function `read_object_with_reference()` actually
4235 works, find the source code for it (something like `git grep
4236 read_object_with | grep ":[a-z]"` in the git repository), and read
4237 the source.
4238
4239 To find out how the result can be used, just read on in `cmd_cat_file()`:
4240
4241 -----------------------------------
4242 write_or_die(1, buf, size);
4243 -----------------------------------
4244
4245 Sometimes, you do not know where to look for a feature. In many such cases,
4246 it helps to search through the output of `git log`, and then `git show` the
4247 corresponding commit.
4248
4249 Example: If you know that there was some test case for `git bundle`, but
4250 do not remember where it was (yes, you _could_ `git grep bundle t/`, but that
4251 does not illustrate the point!):
4252
4253 ------------------------
4254 $ git log --no-merges t/
4255 ------------------------
4256
4257 In the pager (`less`), just search for "bundle", go a few lines back,
4258 and see that it is in commit 18449ab0... Now just copy this object name,
4259 and paste it into the command line
4260
4261 -------------------
4262 $ git show 18449ab0
4263 -------------------
4264
4265 Voila.
4266
4267 Another example: Find out what to do in order to make some script a
4268 builtin:
4269
4270 -------------------------------------------------
4271 $ git log --no-merges --diff-filter=A builtin-*.c
4272 -------------------------------------------------
4273
4274 You see, Git is actually the best tool to find out about the source of Git
4275 itself!
4276
4277 [[glossary]]
4278 GIT Glossary
4279 ============
4280
4281 include::glossary-content.txt[]
4282
4283 [[git-quick-start]]
4284 Appendix A: Git Quick Reference
4285 ===============================
4286
4287 This is a quick summary of the major commands; the previous chapters
4288 explain how these work in more detail.
4289
4290 [[quick-creating-a-new-repository]]
4291 Creating a new repository
4292 -------------------------
4293
4294 From a tarball:
4295
4296 -----------------------------------------------
4297 $ tar xzf project.tar.gz
4298 $ cd project
4299 $ git init
4300 Initialized empty Git repository in .git/
4301 $ git add .
4302 $ git commit
4303 -----------------------------------------------
4304
4305 From a remote repository:
4306
4307 -----------------------------------------------
4308 $ git clone git://example.com/pub/project.git
4309 $ cd project
4310 -----------------------------------------------
4311
4312 [[managing-branches]]
4313 Managing branches
4314 -----------------
4315
4316 -----------------------------------------------
4317 $ git branch # list all local branches in this repo
4318 $ git checkout test # switch working directory to branch "test"
4319 $ git branch new # create branch "new" starting at current HEAD
4320 $ git branch -d new # delete branch "new"
4321 -----------------------------------------------
4322
4323 Instead of basing a new branch on current HEAD (the default), use:
4324
4325 -----------------------------------------------
4326 $ git branch new test # branch named "test"
4327 $ git branch new v2.6.15 # tag named v2.6.15
4328 $ git branch new HEAD^ # commit before the most recent
4329 $ git branch new HEAD^^ # commit before that
4330 $ git branch new test~10 # ten commits before tip of branch "test"
4331 -----------------------------------------------
4332
4333 Create and switch to a new branch at the same time:
4334
4335 -----------------------------------------------
4336 $ git checkout -b new v2.6.15
4337 -----------------------------------------------
4338
4339 Update and examine branches from the repository you cloned from:
4340
4341 -----------------------------------------------
4342 $ git fetch # update
4343 $ git branch -r # list
4344 origin/master
4345 origin/next
4346 ...
4347 $ git checkout -b masterwork origin/master
4348 -----------------------------------------------
4349
4350 Fetch a branch from a different repository, and give it a new
4351 name in your repository:
4352
4353 -----------------------------------------------
4354 $ git fetch git://example.com/project.git theirbranch:mybranch
4355 $ git fetch git://example.com/project.git v2.6.15:mybranch
4356 -----------------------------------------------
4357
4358 Keep a list of repositories you work with regularly:
4359
4360 -----------------------------------------------
4361 $ git remote add example git://example.com/project.git
4362 $ git remote # list remote repositories
4363 example
4364 origin
4365 $ git remote show example # get details
4366 * remote example
4367 URL: git://example.com/project.git
4368 Tracked remote branches
4369 master
4370 next
4371 ...
4372 $ git fetch example # update branches from example
4373 $ git branch -r # list all remote branches
4374 -----------------------------------------------
4375
4376
4377 [[exploring-history]]
4378 Exploring history
4379 -----------------
4380
4381 -----------------------------------------------
4382 $ gitk # visualize and browse history
4383 $ git log # list all commits
4384 $ git log src/ # ...modifying src/
4385 $ git log v2.6.15..v2.6.16 # ...in v2.6.16, not in v2.6.15
4386 $ git log master..test # ...in branch test, not in branch master
4387 $ git log test..master # ...in branch master, but not in test
4388 $ git log test...master # ...in one branch, not in both
4389 $ git log -S'foo()' # ...where difference contain "foo()"
4390 $ git log --since="2 weeks ago"
4391 $ git log -p # show patches as well
4392 $ git show # most recent commit
4393 $ git diff v2.6.15..v2.6.16 # diff between two tagged versions
4394 $ git diff v2.6.15..HEAD # diff with current head
4395 $ git grep "foo()" # search working directory for "foo()"
4396 $ git grep v2.6.15 "foo()" # search old tree for "foo()"
4397 $ git show v2.6.15:a.txt # look at old version of a.txt
4398 -----------------------------------------------
4399
4400 Search for regressions:
4401
4402 -----------------------------------------------
4403 $ git bisect start
4404 $ git bisect bad # current version is bad
4405 $ git bisect good v2.6.13-rc2 # last known good revision
4406 Bisecting: 675 revisions left to test after this
4407 # test here, then:
4408 $ git bisect good # if this revision is good, or
4409 $ git bisect bad # if this revision is bad.
4410 # repeat until done.
4411 -----------------------------------------------
4412
4413 [[making-changes]]
4414 Making changes
4415 --------------
4416
4417 Make sure git knows who to blame:
4418
4419 ------------------------------------------------
4420 $ cat >>~/.gitconfig <<\EOF
4421 [user]
4422 name = Your Name Comes Here
4423 email = you@yourdomain.example.com
4424 EOF
4425 ------------------------------------------------
4426
4427 Select file contents to include in the next commit, then make the
4428 commit:
4429
4430 -----------------------------------------------
4431 $ git add a.txt # updated file
4432 $ git add b.txt # new file
4433 $ git rm c.txt # old file
4434 $ git commit
4435 -----------------------------------------------
4436
4437 Or, prepare and create the commit in one step:
4438
4439 -----------------------------------------------
4440 $ git commit d.txt # use latest content only of d.txt
4441 $ git commit -a # use latest content of all tracked files
4442 -----------------------------------------------
4443
4444 [[merging]]
4445 Merging
4446 -------
4447
4448 -----------------------------------------------
4449 $ git merge test # merge branch "test" into the current branch
4450 $ git pull git://example.com/project.git master
4451 # fetch and merge in remote branch
4452 $ git pull . test # equivalent to git merge test
4453 -----------------------------------------------
4454
4455 [[sharing-your-changes]]
4456 Sharing your changes
4457 --------------------
4458
4459 Importing or exporting patches:
4460
4461 -----------------------------------------------
4462 $ git format-patch origin..HEAD # format a patch for each commit
4463 # in HEAD but not in origin
4464 $ git am mbox # import patches from the mailbox "mbox"
4465 -----------------------------------------------
4466
4467 Fetch a branch in a different git repository, then merge into the
4468 current branch:
4469
4470 -----------------------------------------------
4471 $ git pull git://example.com/project.git theirbranch
4472 -----------------------------------------------
4473
4474 Store the fetched branch into a local branch before merging into the
4475 current branch:
4476
4477 -----------------------------------------------
4478 $ git pull git://example.com/project.git theirbranch:mybranch
4479 -----------------------------------------------
4480
4481 After creating commits on a local branch, update the remote
4482 branch with your commits:
4483
4484 -----------------------------------------------
4485 $ git push ssh://example.com/project.git mybranch:theirbranch
4486 -----------------------------------------------
4487
4488 When remote and local branch are both named "test":
4489
4490 -----------------------------------------------
4491 $ git push ssh://example.com/project.git test
4492 -----------------------------------------------
4493
4494 Shortcut version for a frequently used remote repository:
4495
4496 -----------------------------------------------
4497 $ git remote add example ssh://example.com/project.git
4498 $ git push example test
4499 -----------------------------------------------
4500
4501 [[repository-maintenance]]
4502 Repository maintenance
4503 ----------------------
4504
4505 Check for corruption:
4506
4507 -----------------------------------------------
4508 $ git fsck
4509 -----------------------------------------------
4510
4511 Recompress, remove unused cruft:
4512
4513 -----------------------------------------------
4514 $ git gc
4515 -----------------------------------------------
4516
4517
4518 [[todo]]
4519 Appendix B: Notes and todo list for this manual
4520 ===============================================
4521
4522 This is a work in progress.
4523
4524 The basic requirements:
4525
4526 - It must be readable in order, from beginning to end, by someone
4527 intelligent with a basic grasp of the UNIX command line, but without
4528 any special knowledge of git. If necessary, any other prerequisites
4529 should be specifically mentioned as they arise.
4530 - Whenever possible, section headings should clearly describe the task
4531 they explain how to do, in language that requires no more knowledge
4532 than necessary: for example, "importing patches into a project" rather
4533 than "the `git am` command"
4534
4535 Think about how to create a clear chapter dependency graph that will
4536 allow people to get to important topics without necessarily reading
4537 everything in between.
4538
4539 Scan Documentation/ for other stuff left out; in particular:
4540
4541 - howto's
4542 - some of technical/?
4543 - hooks
4544 - list of commands in linkgit:git[1]
4545
4546 Scan email archives for other stuff left out
4547
4548 Scan man pages to see if any assume more background than this manual
4549 provides.
4550
4551 Simplify beginning by suggesting disconnected head instead of
4552 temporary branch creation?
4553
4554 Add more good examples. Entire sections of just cookbook examples
4555 might be a good idea; maybe make an "advanced examples" section a
4556 standard end-of-chapter section?
4557
4558 Include cross-references to the glossary, where appropriate.
4559
4560 Document shallow clones? See draft 1.5.0 release notes for some
4561 documentation.
4562
4563 Add a section on working with other version control systems, including
4564 CVS, Subversion, and just imports of series of release tarballs.
4565
4566 More details on gitweb?
4567
4568 Write a chapter on using plumbing and writing scripts.
4569
4570 Alternates, clone -reference, etc.
4571
4572 More on recovery from repository corruption. See:
4573 http://marc.theaimsgroup.com/?l=git&m=117263864820799&w=2
4574 http://marc.theaimsgroup.com/?l=git&m=117147855503798&w=2