1 Git User's Manual (for version 1.5.1 or newer)
2 ______________________________________________
4 This manual is designed to be readable by someone with basic unix
5 command-line skills, but no previous knowledge of git.
7 Chapter 1 gives a brief overview of git commands, without any
8 explanation; you may prefer to skip to chapter 2 on a first reading.
10 Chapters 2 and 3 explain how to fetch and study a project using
11 git--the tools you'd need to build and test a particular version of a
12 software project, to search for regressions, and so on.
14 Chapter 4 explains how to do development with git, and chapter 5 how
15 to share that development with others.
17 Further chapters cover more specialized topics.
19 Comprehensive reference documentation is available through the man
20 pages. For a command such as "git clone", just use
22 ------------------------------------------------
24 ------------------------------------------------
30 This is a quick summary of the major commands; the following chapters
31 will explain how these work in more detail.
33 [[quick-creating-a-new-repository]]
34 Creating a new repository
35 -------------------------
39 -----------------------------------------------
40 $ tar xzf project.tar.gz
43 Initialized empty Git repository in .git/
46 -----------------------------------------------
48 From a remote repository:
50 -----------------------------------------------
51 $ git clone git://example.com/pub/project.git
53 -----------------------------------------------
59 -----------------------------------------------
60 $ git branch # list all branches in this repo
61 $ git checkout test # switch working directory to branch "test"
62 $ git branch new # create branch "new" starting at current HEAD
63 $ git branch -d new # delete branch "new"
64 -----------------------------------------------
66 Instead of basing new branch on current HEAD (the default), use:
68 -----------------------------------------------
69 $ git branch new test # branch named "test"
70 $ git branch new v2.6.15 # tag named v2.6.15
71 $ git branch new HEAD^ # commit before the most recent
72 $ git branch new HEAD^^ # commit before that
73 $ git branch new test~10 # ten commits before tip of branch "test"
74 -----------------------------------------------
76 Create and switch to a new branch at the same time:
78 -----------------------------------------------
79 $ git checkout -b new v2.6.15
80 -----------------------------------------------
82 Update and examine branches from the repository you cloned from:
84 -----------------------------------------------
86 $ git branch -r # list
90 $ git checkout -b masterwork origin/master
91 -----------------------------------------------
93 Fetch a branch from a different repository, and give it a new
94 name in your repository:
96 -----------------------------------------------
97 $ git fetch git://example.com/project.git theirbranch:mybranch
98 $ git fetch git://example.com/project.git v2.6.15:mybranch
99 -----------------------------------------------
101 Keep a list of repositories you work with regularly:
103 -----------------------------------------------
104 $ git remote add example git://example.com/project.git
105 $ git remote # list remote repositories
108 $ git remote show example # get details
110 URL: git://example.com/project.git
111 Tracked remote branches
113 $ git fetch example # update branches from example
114 $ git branch -r # list all remote branches
115 -----------------------------------------------
118 [[exploring-history]]
122 -----------------------------------------------
123 $ gitk # visualize and browse history
124 $ git log # list all commits
125 $ git log src/ # ...modifying src/
126 $ git log v2.6.15..v2.6.16 # ...in v2.6.16, not in v2.6.15
127 $ git log master..test # ...in branch test, not in branch master
128 $ git log test..master # ...in branch master, but not in test
129 $ git log test...master # ...in one branch, not in both
130 $ git log -S'foo()' # ...where difference contain "foo()"
131 $ git log --since="2 weeks ago"
132 $ git log -p # show patches as well
133 $ git show # most recent commit
134 $ git diff v2.6.15..v2.6.16 # diff between two tagged versions
135 $ git diff v2.6.15..HEAD # diff with current head
136 $ git grep "foo()" # search working directory for "foo()"
137 $ git grep v2.6.15 "foo()" # search old tree for "foo()"
138 $ git show v2.6.15:a.txt # look at old version of a.txt
139 -----------------------------------------------
141 Search for regressions:
143 -----------------------------------------------
145 $ git bisect bad # current version is bad
146 $ git bisect good v2.6.13-rc2 # last known good revision
147 Bisecting: 675 revisions left to test after this
149 $ git bisect good # if this revision is good, or
150 $ git bisect bad # if this revision is bad.
152 -----------------------------------------------
158 Make sure git knows who to blame:
160 ------------------------------------------------
161 $ cat >>~/.gitconfig <<\EOF
163 name = Your Name Comes Here
164 email = you@yourdomain.example.com
166 ------------------------------------------------
168 Select file contents to include in the next commit, then make the
171 -----------------------------------------------
172 $ git add a.txt # updated file
173 $ git add b.txt # new file
174 $ git rm c.txt # old file
176 -----------------------------------------------
178 Or, prepare and create the commit in one step:
180 -----------------------------------------------
181 $ git commit d.txt # use latest content only of d.txt
182 $ git commit -a # use latest content of all tracked files
183 -----------------------------------------------
189 -----------------------------------------------
190 $ git merge test # merge branch "test" into the current branch
191 $ git pull git://example.com/project.git master
192 # fetch and merge in remote branch
193 $ git pull . test # equivalent to git merge test
194 -----------------------------------------------
196 [[sharing-your-changes]]
200 Importing or exporting patches:
202 -----------------------------------------------
203 $ git format-patch origin..HEAD # format a patch for each commit
204 # in HEAD but not in origin
205 $ git am mbox # import patches from the mailbox "mbox"
206 -----------------------------------------------
208 Fetch a branch in a different git repository, then merge into the
211 -----------------------------------------------
212 $ git pull git://example.com/project.git theirbranch
213 -----------------------------------------------
215 Store the fetched branch into a local branch before merging into the
218 -----------------------------------------------
219 $ git pull git://example.com/project.git theirbranch:mybranch
220 -----------------------------------------------
222 After creating commits on a local branch, update the remote
223 branch with your commits:
225 -----------------------------------------------
226 $ git push ssh://example.com/project.git mybranch:theirbranch
227 -----------------------------------------------
229 When remote and local branch are both named "test":
231 -----------------------------------------------
232 $ git push ssh://example.com/project.git test
233 -----------------------------------------------
235 Shortcut version for a frequently used remote repository:
237 -----------------------------------------------
238 $ git remote add example ssh://example.com/project.git
239 $ git push example test
240 -----------------------------------------------
242 [[repository-maintenance]]
243 Repository maintenance
244 ----------------------
246 Check for corruption:
248 -----------------------------------------------
250 -----------------------------------------------
252 Recompress, remove unused cruft:
254 -----------------------------------------------
256 -----------------------------------------------
258 [[repositories-and-branches]]
259 Repositories and Branches
260 =========================
262 [[how-to-get-a-git-repository]]
263 How to get a git repository
264 ---------------------------
266 It will be useful to have a git repository to experiment with as you
269 The best way to get one is by using the gitlink:git-clone[1] command
270 to download a copy of an existing repository for a project that you
271 are interested in. If you don't already have a project in mind, here
272 are some interesting examples:
274 ------------------------------------------------
275 # git itself (approx. 10MB download):
276 $ git clone git://git.kernel.org/pub/scm/git/git.git
277 # the linux kernel (approx. 150MB download):
278 $ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
279 ------------------------------------------------
281 The initial clone may be time-consuming for a large project, but you
282 will only need to clone once.
284 The clone command creates a new directory named after the project
285 ("git" or "linux-2.6" in the examples above). After you cd into this
286 directory, you will see that it contains a copy of the project files,
287 together with a special top-level directory named ".git", which
288 contains all the information about the history of the project.
290 In most of the following, examples will be taken from one of the two
294 How to check out a different version of a project
295 -------------------------------------------------
297 Git is best thought of as a tool for storing the history of a
298 collection of files. It stores the history as a compressed
299 collection of interrelated snapshots (versions) of the project's
302 A single git repository may contain multiple branches. It keeps track
303 of them by keeping a list of <<def_head,heads>> which reference the
304 latest version on each branch; the gitlink:git-branch[1] command shows
305 you the list of branch heads:
307 ------------------------------------------------
310 ------------------------------------------------
312 A freshly cloned repository contains a single branch head, by default
313 named "master", with the working directory initialized to the state of
314 the project referred to by that branch head.
316 Most projects also use <<def_tag,tags>>. Tags, like heads, are
317 references into the project's history, and can be listed using the
318 gitlink:git-tag[1] command:
320 ------------------------------------------------
332 ------------------------------------------------
334 Tags are expected to always point at the same version of a project,
335 while heads are expected to advance as development progresses.
337 Create a new branch head pointing to one of these versions and check it
338 out using gitlink:git-checkout[1]:
340 ------------------------------------------------
341 $ git checkout -b new v2.6.13
342 ------------------------------------------------
344 The working directory then reflects the contents that the project had
345 when it was tagged v2.6.13, and gitlink:git-branch[1] shows two
346 branches, with an asterisk marking the currently checked-out branch:
348 ------------------------------------------------
352 ------------------------------------------------
354 If you decide that you'd rather see version 2.6.17, you can modify
355 the current branch to point at v2.6.17 instead, with
357 ------------------------------------------------
358 $ git reset --hard v2.6.17
359 ------------------------------------------------
361 Note that if the current branch head was your only reference to a
362 particular point in history, then resetting that branch may leave you
363 with no way to find the history it used to point to; so use this command
366 [[understanding-commits]]
367 Understanding History: Commits
368 ------------------------------
370 Every change in the history of a project is represented by a commit.
371 The gitlink:git-show[1] command shows the most recent commit on the
374 ------------------------------------------------
376 commit 2b5f6dcce5bf94b9b119e9ed8d537098ec61c3d2
377 Author: Jamal Hadi Salim <hadi@cyberus.ca>
378 Date: Sat Dec 2 22:22:25 2006 -0800
380 [XFRM]: Fix aevent structuring to be more complete.
382 aevents can not uniquely identify an SA. We break the ABI with this
383 patch, but consensus is that since it is not yet utilized by any
384 (known) application then it is fine (better do it now than later).
386 Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
387 Signed-off-by: David S. Miller <davem@davemloft.net>
389 diff --git a/Documentation/networking/xfrm_sync.txt b/Documentation/networking/xfrm_sync.txt
390 index 8be626f..d7aac9d 100644
391 --- a/Documentation/networking/xfrm_sync.txt
392 +++ b/Documentation/networking/xfrm_sync.txt
393 @@ -47,10 +47,13 @@ aevent_id structure looks like:
395 struct xfrm_aevent_id {
396 struct xfrm_usersa_id sa_id;
397 + xfrm_address_t saddr;
402 ------------------------------------------------
404 As you can see, a commit shows who made the latest change, what they
407 Every commit has a 40-hexdigit id, sometimes called the "object name" or the
408 "SHA1 id", shown on the first line of the "git show" output. You can usually
409 refer to a commit by a shorter name, such as a tag or a branch name, but this
410 longer name can also be useful. Most importantly, it is a globally unique
411 name for this commit: so if you tell somebody else the object name (for
412 example in email), then you are guaranteed that name will refer to the same
413 commit in their repository that it does in yours (assuming their repository
414 has that commit at all). Since the object name is computed as a hash over the
415 contents of the commit, you are guaranteed that the commit can never change
416 without its name also changing.
418 In fact, in <<git-internals>> we shall see that everything stored in git
419 history, including file data and directory contents, is stored in an object
420 with a name that is a hash of its contents.
422 [[understanding-reachability]]
423 Understanding history: commits, parents, and reachability
424 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
426 Every commit (except the very first commit in a project) also has a
427 parent commit which shows what happened before this commit.
428 Following the chain of parents will eventually take you back to the
429 beginning of the project.
431 However, the commits do not form a simple list; git allows lines of
432 development to diverge and then reconverge, and the point where two
433 lines of development reconverge is called a "merge". The commit
434 representing a merge can therefore have more than one parent, with
435 each parent representing the most recent commit on one of the lines
436 of development leading to that point.
438 The best way to see how this works is using the gitlink:gitk[1]
439 command; running gitk now on a git repository and looking for merge
440 commits will help understand how the git organizes history.
442 In the following, we say that commit X is "reachable" from commit Y
443 if commit X is an ancestor of commit Y. Equivalently, you could say
444 that Y is a descendent of X, or that there is a chain of parents
445 leading from commit Y to commit X.
448 Understanding history: History diagrams
449 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
451 We will sometimes represent git history using diagrams like the one
452 below. Commits are shown as "o", and the links between them with
453 lines drawn with - / and \. Time goes left to right:
456 ................................................
462 ................................................
464 If we need to talk about a particular commit, the character "o" may
465 be replaced with another letter or number.
468 Understanding history: What is a branch?
469 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
471 When we need to be precise, we will use the word "branch" to mean a line
472 of development, and "branch head" (or just "head") to mean a reference
473 to the most recent commit on a branch. In the example above, the branch
474 head named "A" is a pointer to one particular commit, but we refer to
475 the line of three commits leading up to that point as all being part of
478 However, when no confusion will result, we often just use the term
479 "branch" both for branches and for branch heads.
481 [[manipulating-branches]]
482 Manipulating branches
483 ---------------------
485 Creating, deleting, and modifying branches is quick and easy; here's
486 a summary of the commands:
490 git branch <branch>::
491 create a new branch named <branch>, referencing the same
492 point in history as the current branch
493 git branch <branch> <start-point>::
494 create a new branch named <branch>, referencing
495 <start-point>, which may be specified any way you like,
496 including using a branch name or a tag name
497 git branch -d <branch>::
498 delete the branch <branch>; if the branch you are deleting
499 points to a commit which is not reachable from this branch,
500 this command will fail with a warning.
501 git branch -D <branch>::
502 even if the branch points to a commit not reachable
503 from the current branch, you may know that that commit
504 is still reachable from some other branch or tag. In that
505 case it is safe to use this command to force git to delete
507 git checkout <branch>::
508 make the current branch <branch>, updating the working
509 directory to reflect the version referenced by <branch>
510 git checkout -b <new> <start-point>::
511 create a new branch <new> referencing <start-point>, and
514 The special symbol "HEAD" can always be used to refer to the current
515 branch. In fact, git uses a file named "HEAD" in the .git directory to
516 remember which branch is current:
518 ------------------------------------------------
520 ref: refs/heads/master
521 ------------------------------------------------
524 Examining an old version without creating a new branch
525 ------------------------------------------------------
527 The git-checkout command normally expects a branch head, but will also
528 accept an arbitrary commit; for example, you can check out the commit
531 ------------------------------------------------
532 $ git checkout v2.6.17
533 Note: moving to "v2.6.17" which isn't a local branch
534 If you want to create a new branch from this checkout, you may do so
535 (now or later) by using -b with the checkout command again. Example:
536 git checkout -b <new_branch_name>
537 HEAD is now at 427abfa... Linux v2.6.17
538 ------------------------------------------------
540 The HEAD then refers to the SHA1 of the commit instead of to a branch,
541 and git branch shows that you are no longer on a branch:
543 ------------------------------------------------
545 427abfa28afedffadfca9dd8b067eb6d36bac53f
549 ------------------------------------------------
551 In this case we say that the HEAD is "detached".
553 This is an easy way to check out a particular version without having to
554 make up a name for the new branch. You can still create a new branch
555 (or tag) for this version later if you decide to.
557 [[examining-remote-branches]]
558 Examining branches from a remote repository
559 -------------------------------------------
561 The "master" branch that was created at the time you cloned is a copy
562 of the HEAD in the repository that you cloned from. That repository
563 may also have had other branches, though, and your local repository
564 keeps branches which track each of those remote branches, which you
565 can view using the "-r" option to gitlink:git-branch[1]:
567 ------------------------------------------------
577 ------------------------------------------------
579 You cannot check out these remote-tracking branches, but you can
580 examine them on a branch of your own, just as you would a tag:
582 ------------------------------------------------
583 $ git checkout -b my-todo-copy origin/todo
584 ------------------------------------------------
586 Note that the name "origin" is just the name that git uses by default
587 to refer to the repository that you cloned from.
589 [[how-git-stores-references]]
590 Naming branches, tags, and other references
591 -------------------------------------------
593 Branches, remote-tracking branches, and tags are all references to
594 commits. All references are named with a slash-separated path name
595 starting with "refs"; the names we've been using so far are actually
598 - The branch "test" is short for "refs/heads/test".
599 - The tag "v2.6.18" is short for "refs/tags/v2.6.18".
600 - "origin/master" is short for "refs/remotes/origin/master".
602 The full name is occasionally useful if, for example, there ever
603 exists a tag and a branch with the same name.
605 As another useful shortcut, if the repository "origin" posesses only
606 a single branch, you can refer to that branch as just "origin".
608 More generally, if you have defined a remote repository named
609 "example", you can refer to the branch in that repository as
610 "example". And for a repository with multiple branches, this will
611 refer to the branch designated as the "HEAD" branch.
613 For the complete list of paths which git checks for references, and
614 the order it uses to decide which to choose when there are multiple
615 references with the same shorthand name, see the "SPECIFYING
616 REVISIONS" section of gitlink:git-rev-parse[1].
618 [[Updating-a-repository-with-git-fetch]]
619 Updating a repository with git fetch
620 ------------------------------------
622 Eventually the developer cloned from will do additional work in her
623 repository, creating new commits and advancing the branches to point
626 The command "git fetch", with no arguments, will update all of the
627 remote-tracking branches to the latest version found in her
628 repository. It will not touch any of your own branches--not even the
629 "master" branch that was created for you on clone.
631 [[fetching-branches]]
632 Fetching branches from other repositories
633 -----------------------------------------
635 You can also track branches from repositories other than the one you
636 cloned from, using gitlink:git-remote[1]:
638 -------------------------------------------------
639 $ git remote add linux-nfs git://linux-nfs.org/pub/nfs-2.6.git
640 $ git fetch linux-nfs
641 * refs/remotes/linux-nfs/master: storing branch 'master' ...
643 -------------------------------------------------
645 New remote-tracking branches will be stored under the shorthand name
646 that you gave "git remote add", in this case linux-nfs:
648 -------------------------------------------------
652 -------------------------------------------------
654 If you run "git fetch <remote>" later, the tracking branches for the
655 named <remote> will be updated.
657 If you examine the file .git/config, you will see that git has added
660 -------------------------------------------------
664 url = git://linux-nfs.org/pub/nfs-2.6.git
665 fetch = +refs/heads/*:refs/remotes/linux-nfs/*
667 -------------------------------------------------
669 This is what causes git to track the remote's branches; you may modify
670 or delete these configuration options by editing .git/config with a
671 text editor. (See the "CONFIGURATION FILE" section of
672 gitlink:git-config[1] for details.)
674 [[exploring-git-history]]
675 Exploring git history
676 =====================
678 Git is best thought of as a tool for storing the history of a
679 collection of files. It does this by storing compressed snapshots of
680 the contents of a file heirarchy, together with "commits" which show
681 the relationships between these snapshots.
683 Git provides extremely flexible and fast tools for exploring the
684 history of a project.
686 We start with one specialized tool that is useful for finding the
687 commit that introduced a bug into a project.
690 How to use bisect to find a regression
691 --------------------------------------
693 Suppose version 2.6.18 of your project worked, but the version at
694 "master" crashes. Sometimes the best way to find the cause of such a
695 regression is to perform a brute-force search through the project's
696 history to find the particular commit that caused the problem. The
697 gitlink:git-bisect[1] command can help you do this:
699 -------------------------------------------------
701 $ git bisect good v2.6.18
702 $ git bisect bad master
703 Bisecting: 3537 revisions left to test after this
704 [65934a9a028b88e83e2b0f8b36618fe503349f8e] BLOCK: Make USB storage depend on SCSI rather than selecting it [try #6]
705 -------------------------------------------------
707 If you run "git branch" at this point, you'll see that git has
708 temporarily moved you to a new branch named "bisect". This branch
709 points to a commit (with commit id 65934...) that is reachable from
710 v2.6.19 but not from v2.6.18. Compile and test it, and see whether
711 it crashes. Assume it does crash. Then:
713 -------------------------------------------------
715 Bisecting: 1769 revisions left to test after this
716 [7eff82c8b1511017ae605f0c99ac275a7e21b867] i2c-core: Drop useless bitmaskings
717 -------------------------------------------------
719 checks out an older version. Continue like this, telling git at each
720 stage whether the version it gives you is good or bad, and notice
721 that the number of revisions left to test is cut approximately in
724 After about 13 tests (in this case), it will output the commit id of
725 the guilty commit. You can then examine the commit with
726 gitlink:git-show[1], find out who wrote it, and mail them your bug
727 report with the commit id. Finally, run
729 -------------------------------------------------
731 -------------------------------------------------
733 to return you to the branch you were on before and delete the
734 temporary "bisect" branch.
736 Note that the version which git-bisect checks out for you at each
737 point is just a suggestion, and you're free to try a different
738 version if you think it would be a good idea. For example,
739 occasionally you may land on a commit that broke something unrelated;
742 -------------------------------------------------
743 $ git bisect visualize
744 -------------------------------------------------
746 which will run gitk and label the commit it chose with a marker that
747 says "bisect". Chose a safe-looking commit nearby, note its commit
748 id, and check it out with:
750 -------------------------------------------------
751 $ git reset --hard fb47ddb2db...
752 -------------------------------------------------
754 then test, run "bisect good" or "bisect bad" as appropriate, and
761 We have seen several ways of naming commits already:
763 - 40-hexdigit object name
764 - branch name: refers to the commit at the head of the given
766 - tag name: refers to the commit pointed to by the given tag
767 (we've seen branches and tags are special cases of
768 <<how-git-stores-references,references>>).
769 - HEAD: refers to the head of the current branch
771 There are many more; see the "SPECIFYING REVISIONS" section of the
772 gitlink:git-rev-parse[1] man page for the complete list of ways to
773 name revisions. Some examples:
775 -------------------------------------------------
776 $ git show fb47ddb2 # the first few characters of the object name
777 # are usually enough to specify it uniquely
778 $ git show HEAD^ # the parent of the HEAD commit
779 $ git show HEAD^^ # the grandparent
780 $ git show HEAD~4 # the great-great-grandparent
781 -------------------------------------------------
783 Recall that merge commits may have more than one parent; by default,
784 ^ and ~ follow the first parent listed in the commit, but you can
787 -------------------------------------------------
788 $ git show HEAD^1 # show the first parent of HEAD
789 $ git show HEAD^2 # show the second parent of HEAD
790 -------------------------------------------------
792 In addition to HEAD, there are several other special names for
795 Merges (to be discussed later), as well as operations such as
796 git-reset, which change the currently checked-out commit, generally
797 set ORIG_HEAD to the value HEAD had before the current operation.
799 The git-fetch operation always stores the head of the last fetched
800 branch in FETCH_HEAD. For example, if you run git fetch without
801 specifying a local branch as the target of the operation
803 -------------------------------------------------
804 $ git fetch git://example.com/proj.git theirbranch
805 -------------------------------------------------
807 the fetched commits will still be available from FETCH_HEAD.
809 When we discuss merges we'll also see the special name MERGE_HEAD,
810 which refers to the other branch that we're merging in to the current
813 The gitlink:git-rev-parse[1] command is a low-level command that is
814 occasionally useful for translating some name for a commit to the object
815 name for that commit:
817 -------------------------------------------------
818 $ git rev-parse origin
819 e05db0fd4f31dde7005f075a84f96b360d05984b
820 -------------------------------------------------
826 We can also create a tag to refer to a particular commit; after
829 -------------------------------------------------
830 $ git tag stable-1 1b2e1d63ff
831 -------------------------------------------------
833 You can use stable-1 to refer to the commit 1b2e1d63ff.
835 This creates a "lightweight" tag. If the tag is a tag you wish to
836 share with others, and possibly sign cryptographically, then you
837 should create a tag object instead; see the gitlink:git-tag[1] man
840 [[browsing-revisions]]
844 The gitlink:git-log[1] command can show lists of commits. On its
845 own, it shows all commits reachable from the parent commit; but you
846 can also make more specific requests:
848 -------------------------------------------------
849 $ git log v2.5.. # commits since (not reachable from) v2.5
850 $ git log test..master # commits reachable from master but not test
851 $ git log master..test # ...reachable from test but not master
852 $ git log master...test # ...reachable from either test or master,
854 $ git log --since="2 weeks ago" # commits from the last 2 weeks
855 $ git log Makefile # commits which modify Makefile
856 $ git log fs/ # ... which modify any file under fs/
857 $ git log -S'foo()' # commits which add or remove any file data
858 # matching the string 'foo()'
859 -------------------------------------------------
861 And of course you can combine all of these; the following finds
862 commits since v2.5 which touch the Makefile or any file under fs:
864 -------------------------------------------------
865 $ git log v2.5.. Makefile fs/
866 -------------------------------------------------
868 You can also ask git log to show patches:
870 -------------------------------------------------
872 -------------------------------------------------
874 See the "--pretty" option in the gitlink:git-log[1] man page for more
877 Note that git log starts with the most recent commit and works
878 backwards through the parents; however, since git history can contain
879 multiple independent lines of development, the particular order that
880 commits are listed in may be somewhat arbitrary.
886 You can generate diffs between any two versions using
889 -------------------------------------------------
890 $ git diff master..test
891 -------------------------------------------------
893 Sometimes what you want instead is a set of patches:
895 -------------------------------------------------
896 $ git format-patch master..test
897 -------------------------------------------------
899 will generate a file with a patch for each commit reachable from test
900 but not from master. Note that if master also has commits which are
901 not reachable from test, then the combined result of these patches
902 will not be the same as the diff produced by the git-diff example.
904 [[viewing-old-file-versions]]
905 Viewing old file versions
906 -------------------------
908 You can always view an old version of a file by just checking out the
909 correct revision first. But sometimes it is more convenient to be
910 able to view an old version of a single file without checking
911 anything out; this command does that:
913 -------------------------------------------------
914 $ git show v2.5:fs/locks.c
915 -------------------------------------------------
917 Before the colon may be anything that names a commit, and after it
918 may be any path to a file tracked by git.
924 [[checking-for-equal-branches]]
925 Check whether two branches point at the same history
926 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
928 Suppose you want to check whether two branches point at the same point
931 -------------------------------------------------
932 $ git diff origin..master
933 -------------------------------------------------
935 will tell you whether the contents of the project are the same at the
936 two branches; in theory, however, it's possible that the same project
937 contents could have been arrived at by two different historical
938 routes. You could compare the object names:
940 -------------------------------------------------
941 $ git rev-list origin
942 e05db0fd4f31dde7005f075a84f96b360d05984b
943 $ git rev-list master
944 e05db0fd4f31dde7005f075a84f96b360d05984b
945 -------------------------------------------------
947 Or you could recall that the ... operator selects all commits
948 contained reachable from either one reference or the other but not
951 -------------------------------------------------
952 $ git log origin...master
953 -------------------------------------------------
955 will return no commits when the two branches are equal.
957 [[finding-tagged-descendants]]
958 Find first tagged version including a given fix
959 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
961 Suppose you know that the commit e05db0fd fixed a certain problem.
962 You'd like to find the earliest tagged release that contains that
965 Of course, there may be more than one answer--if the history branched
966 after commit e05db0fd, then there could be multiple "earliest" tagged
969 You could just visually inspect the commits since e05db0fd:
971 -------------------------------------------------
973 -------------------------------------------------
975 Or you can use gitlink:git-name-rev[1], which will give the commit a
976 name based on any tag it finds pointing to one of the commit's
979 -------------------------------------------------
980 $ git name-rev --tags e05db0fd
981 e05db0fd tags/v1.5.0-rc1^0~23
982 -------------------------------------------------
984 The gitlink:git-describe[1] command does the opposite, naming the
985 revision using a tag on which the given commit is based:
987 -------------------------------------------------
988 $ git describe e05db0fd
989 v1.5.0-rc0-260-ge05db0f
990 -------------------------------------------------
992 but that may sometimes help you guess which tags might come after the
995 If you just want to verify whether a given tagged version contains a
996 given commit, you could use gitlink:git-merge-base[1]:
998 -------------------------------------------------
999 $ git merge-base e05db0fd v1.5.0-rc1
1000 e05db0fd4f31dde7005f075a84f96b360d05984b
1001 -------------------------------------------------
1003 The merge-base command finds a common ancestor of the given commits,
1004 and always returns one or the other in the case where one is a
1005 descendant of the other; so the above output shows that e05db0fd
1006 actually is an ancestor of v1.5.0-rc1.
1008 Alternatively, note that
1010 -------------------------------------------------
1011 $ git log v1.5.0-rc1..e05db0fd
1012 -------------------------------------------------
1014 will produce empty output if and only if v1.5.0-rc1 includes e05db0fd,
1015 because it outputs only commits that are not reachable from v1.5.0-rc1.
1017 As yet another alternative, the gitlink:git-show-branch[1] command lists
1018 the commits reachable from its arguments with a display on the left-hand
1019 side that indicates which arguments that commit is reachable from. So,
1020 you can run something like
1022 -------------------------------------------------
1023 $ git show-branch e05db0fd v1.5.0-rc0 v1.5.0-rc1 v1.5.0-rc2
1024 ! [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
1026 ! [v1.5.0-rc0] GIT v1.5.0 preview
1027 ! [v1.5.0-rc1] GIT v1.5.0-rc1
1028 ! [v1.5.0-rc2] GIT v1.5.0-rc2
1030 -------------------------------------------------
1032 then search for a line that looks like
1034 -------------------------------------------------
1035 + ++ [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
1037 -------------------------------------------------
1039 Which shows that e05db0fd is reachable from itself, from v1.5.0-rc1, and
1040 from v1.5.0-rc2, but not from v1.5.0-rc0.
1043 [[Developing-with-git]]
1047 [[telling-git-your-name]]
1048 Telling git your name
1049 ---------------------
1051 Before creating any commits, you should introduce yourself to git. The
1052 easiest way to do so is to make sure the following lines appear in a
1053 file named .gitconfig in your home directory:
1055 ------------------------------------------------
1057 name = Your Name Comes Here
1058 email = you@yourdomain.example.com
1059 ------------------------------------------------
1061 (See the "CONFIGURATION FILE" section of gitlink:git-config[1] for
1062 details on the configuration file.)
1065 [[creating-a-new-repository]]
1066 Creating a new repository
1067 -------------------------
1069 Creating a new repository from scratch is very easy:
1071 -------------------------------------------------
1075 -------------------------------------------------
1077 If you have some initial content (say, a tarball):
1079 -------------------------------------------------
1080 $ tar -xzvf project.tar.gz
1083 $ git add . # include everything below ./ in the first commit:
1085 -------------------------------------------------
1087 [[how-to-make-a-commit]]
1088 How to make a commit
1089 --------------------
1091 Creating a new commit takes three steps:
1093 1. Making some changes to the working directory using your
1095 2. Telling git about your changes.
1096 3. Creating the commit using the content you told git about
1099 In practice, you can interleave and repeat steps 1 and 2 as many
1100 times as you want: in order to keep track of what you want committed
1101 at step 3, git maintains a snapshot of the tree's contents in a
1102 special staging area called "the index."
1104 At the beginning, the content of the index will be identical to
1105 that of the HEAD. The command "git diff --cached", which shows
1106 the difference between the HEAD and the index, should therefore
1107 produce no output at that point.
1109 Modifying the index is easy:
1111 To update the index with the new contents of a modified file, use
1113 -------------------------------------------------
1114 $ git add path/to/file
1115 -------------------------------------------------
1117 To add the contents of a new file to the index, use
1119 -------------------------------------------------
1120 $ git add path/to/file
1121 -------------------------------------------------
1123 To remove a file from the index and from the working tree,
1125 -------------------------------------------------
1126 $ git rm path/to/file
1127 -------------------------------------------------
1129 After each step you can verify that
1131 -------------------------------------------------
1133 -------------------------------------------------
1135 always shows the difference between the HEAD and the index file--this
1136 is what you'd commit if you created the commit now--and that
1138 -------------------------------------------------
1140 -------------------------------------------------
1142 shows the difference between the working tree and the index file.
1144 Note that "git add" always adds just the current contents of a file
1145 to the index; further changes to the same file will be ignored unless
1146 you run git-add on the file again.
1148 When you're ready, just run
1150 -------------------------------------------------
1152 -------------------------------------------------
1154 and git will prompt you for a commit message and then create the new
1155 commit. Check to make sure it looks like what you expected with
1157 -------------------------------------------------
1159 -------------------------------------------------
1161 As a special shortcut,
1163 -------------------------------------------------
1165 -------------------------------------------------
1167 will update the index with any files that you've modified or removed
1168 and create a commit, all in one step.
1170 A number of commands are useful for keeping track of what you're
1173 -------------------------------------------------
1174 $ git diff --cached # difference between HEAD and the index; what
1175 # would be commited if you ran "commit" now.
1176 $ git diff # difference between the index file and your
1177 # working directory; changes that would not
1178 # be included if you ran "commit" now.
1179 $ git status # a brief per-file summary of the above.
1180 -------------------------------------------------
1182 [[creating-good-commit-messages]]
1183 Creating good commit messages
1184 -----------------------------
1186 Though not required, it's a good idea to begin the commit message
1187 with a single short (less than 50 character) line summarizing the
1188 change, followed by a blank line and then a more thorough
1189 description. Tools that turn commits into email, for example, use
1190 the first line on the Subject line and the rest of the commit in the
1197 You can rejoin two diverging branches of development using
1198 gitlink:git-merge[1]:
1200 -------------------------------------------------
1201 $ git merge branchname
1202 -------------------------------------------------
1204 merges the development in the branch "branchname" into the current
1205 branch. If there are conflicts--for example, if the same file is
1206 modified in two different ways in the remote branch and the local
1207 branch--then you are warned; the output may look something like this:
1209 -------------------------------------------------
1212 Auto-merged file.txt
1213 CONFLICT (content): Merge conflict in file.txt
1214 Automatic merge failed; fix conflicts and then commit the result.
1215 -------------------------------------------------
1217 Conflict markers are left in the problematic files, and after
1218 you resolve the conflicts manually, you can update the index
1219 with the contents and run git commit, as you normally would when
1220 creating a new file.
1222 If you examine the resulting commit using gitk, you will see that it
1223 has two parents, one pointing to the top of the current branch, and
1224 one to the top of the other branch.
1228 [[resolving-a-merge]]
1232 When a merge isn't resolved automatically, git leaves the index and
1233 the working tree in a special state that gives you all the
1234 information you need to help resolve the merge.
1236 Files with conflicts are marked specially in the index, so until you
1237 resolve the problem and update the index, gitlink:git-commit[1] will
1240 -------------------------------------------------
1242 file.txt: needs merge
1243 -------------------------------------------------
1245 Also, gitlink:git-status[1] will list those files as "unmerged", and the
1246 files with conflicts will have conflict markers added, like this:
1248 -------------------------------------------------
1249 <<<<<<< HEAD:file.txt
1253 >>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1254 -------------------------------------------------
1256 All you need to do is edit the files to resolve the conflicts, and then
1258 -------------------------------------------------
1261 -------------------------------------------------
1263 Note that the commit message will already be filled in for you with
1264 some information about the merge. Normally you can just use this
1265 default message unchanged, but you may add additional commentary of
1266 your own if desired.
1268 The above is all you need to know to resolve a simple merge. But git
1269 also provides more information to help resolve conflicts:
1271 [[conflict-resolution]]
1272 Getting conflict-resolution help during a merge
1273 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1275 All of the changes that git was able to merge automatically are
1276 already added to the index file, so gitlink:git-diff[1] shows only
1277 the conflicts. It uses an unusual syntax:
1279 -------------------------------------------------
1282 index 802992c,2b60207..0000000
1285 @@@ -1,1 -1,1 +1,5 @@@
1286 ++<<<<<<< HEAD:file.txt
1290 ++>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1291 -------------------------------------------------
1293 Recall that the commit which will be commited after we resolve this
1294 conflict will have two parents instead of the usual one: one parent
1295 will be HEAD, the tip of the current branch; the other will be the
1296 tip of the other branch, which is stored temporarily in MERGE_HEAD.
1298 During the merge, the index holds three versions of each file. Each of
1299 these three "file stages" represents a different version of the file:
1301 -------------------------------------------------
1302 $ git show :1:file.txt # the file in a common ancestor of both branches
1303 $ git show :2:file.txt # the version from HEAD, but including any
1304 # nonconflicting changes from MERGE_HEAD
1305 $ git show :3:file.txt # the version from MERGE_HEAD, but including any
1306 # nonconflicting changes from HEAD.
1307 -------------------------------------------------
1309 Since the stage 2 and stage 3 versions have already been updated with
1310 nonconflicting changes, the only remaining differences between them are
1311 the important ones; thus gitlink:git-diff[1] can use the information in
1312 the index to show only those conflicts.
1314 The diff above shows the differences between the working-tree version of
1315 file.txt and the stage 2 and stage 3 versions. So instead of preceding
1316 each line by a single "+" or "-", it now uses two columns: the first
1317 column is used for differences between the first parent and the working
1318 directory copy, and the second for differences between the second parent
1319 and the working directory copy. (See the "COMBINED DIFF FORMAT" section
1320 of gitlink:git-diff-files[1] for a details of the format.)
1322 After resolving the conflict in the obvious way (but before updating the
1323 index), the diff will look like:
1325 -------------------------------------------------
1328 index 802992c,2b60207..0000000
1331 @@@ -1,1 -1,1 +1,1 @@@
1335 -------------------------------------------------
1337 This shows that our resolved version deleted "Hello world" from the
1338 first parent, deleted "Goodbye" from the second parent, and added
1339 "Goodbye world", which was previously absent from both.
1341 Some special diff options allow diffing the working directory against
1342 any of these stages:
1344 -------------------------------------------------
1345 $ git diff -1 file.txt # diff against stage 1
1346 $ git diff --base file.txt # same as the above
1347 $ git diff -2 file.txt # diff against stage 2
1348 $ git diff --ours file.txt # same as the above
1349 $ git diff -3 file.txt # diff against stage 3
1350 $ git diff --theirs file.txt # same as the above.
1351 -------------------------------------------------
1353 The gitlink:git-log[1] and gitk[1] commands also provide special help
1356 -------------------------------------------------
1359 -------------------------------------------------
1361 These will display all commits which exist only on HEAD or on
1362 MERGE_HEAD, and which touch an unmerged file.
1364 Each time you resolve the conflicts in a file and update the index:
1366 -------------------------------------------------
1368 -------------------------------------------------
1370 the different stages of that file will be "collapsed", after which
1371 git-diff will (by default) no longer show diffs for that file.
1377 If you get stuck and decide to just give up and throw the whole mess
1378 away, you can always return to the pre-merge state with
1380 -------------------------------------------------
1381 $ git reset --hard HEAD
1382 -------------------------------------------------
1384 Or, if you've already commited the merge that you want to throw away,
1386 -------------------------------------------------
1387 $ git reset --hard ORIG_HEAD
1388 -------------------------------------------------
1390 However, this last command can be dangerous in some cases--never
1391 throw away a commit you have already committed if that commit may
1392 itself have been merged into another branch, as doing so may confuse
1399 There is one special case not mentioned above, which is treated
1400 differently. Normally, a merge results in a merge commit, with two
1401 parents, one pointing at each of the two lines of development that
1404 However, if the current branch is a descendant of the other--so every
1405 commit present in the one is already contained in the other--then git
1406 just performs a "fast forward"; the head of the current branch is moved
1407 forward to point at the head of the merged-in branch, without any new
1408 commits being created.
1414 If you've messed up the working tree, but haven't yet committed your
1415 mistake, you can return the entire working tree to the last committed
1418 -------------------------------------------------
1419 $ git reset --hard HEAD
1420 -------------------------------------------------
1422 If you make a commit that you later wish you hadn't, there are two
1423 fundamentally different ways to fix the problem:
1425 1. You can create a new commit that undoes whatever was done
1426 by the previous commit. This is the correct thing if your
1427 mistake has already been made public.
1429 2. You can go back and modify the old commit. You should
1430 never do this if you have already made the history public;
1431 git does not normally expect the "history" of a project to
1432 change, and cannot correctly perform repeated merges from
1433 a branch that has had its history changed.
1435 [[reverting-a-commit]]
1436 Fixing a mistake with a new commit
1437 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1439 Creating a new commit that reverts an earlier change is very easy;
1440 just pass the gitlink:git-revert[1] command a reference to the bad
1441 commit; for example, to revert the most recent commit:
1443 -------------------------------------------------
1445 -------------------------------------------------
1447 This will create a new commit which undoes the change in HEAD. You
1448 will be given a chance to edit the commit message for the new commit.
1450 You can also revert an earlier change, for example, the next-to-last:
1452 -------------------------------------------------
1454 -------------------------------------------------
1456 In this case git will attempt to undo the old change while leaving
1457 intact any changes made since then. If more recent changes overlap
1458 with the changes to be reverted, then you will be asked to fix
1459 conflicts manually, just as in the case of <<resolving-a-merge,
1460 resolving a merge>>.
1462 [[fixing-a-mistake-by-editing-history]]
1463 Fixing a mistake by editing history
1464 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1466 If the problematic commit is the most recent commit, and you have not
1467 yet made that commit public, then you may just
1468 <<undoing-a-merge,destroy it using git-reset>>.
1471 can edit the working directory and update the index to fix your
1472 mistake, just as if you were going to <<how-to-make-a-commit,create a
1473 new commit>>, then run
1475 -------------------------------------------------
1476 $ git commit --amend
1477 -------------------------------------------------
1479 which will replace the old commit by a new commit incorporating your
1480 changes, giving you a chance to edit the old commit message first.
1482 Again, you should never do this to a commit that may already have
1483 been merged into another branch; use gitlink:git-revert[1] instead in
1486 It is also possible to edit commits further back in the history, but
1487 this is an advanced topic to be left for
1488 <<cleaning-up-history,another chapter>>.
1490 [[checkout-of-path]]
1491 Checking out an old version of a file
1492 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1494 In the process of undoing a previous bad change, you may find it
1495 useful to check out an older version of a particular file using
1496 gitlink:git-checkout[1]. We've used git checkout before to switch
1497 branches, but it has quite different behavior if it is given a path
1500 -------------------------------------------------
1501 $ git checkout HEAD^ path/to/file
1502 -------------------------------------------------
1504 replaces path/to/file by the contents it had in the commit HEAD^, and
1505 also updates the index to match. It does not change branches.
1507 If you just want to look at an old version of the file, without
1508 modifying the working directory, you can do that with
1509 gitlink:git-show[1]:
1511 -------------------------------------------------
1512 $ git show HEAD^:path/to/file
1513 -------------------------------------------------
1515 which will display the given version of the file.
1517 [[ensuring-good-performance]]
1518 Ensuring good performance
1519 -------------------------
1521 On large repositories, git depends on compression to keep the history
1522 information from taking up to much space on disk or in memory.
1524 This compression is not performed automatically. Therefore you
1525 should occasionally run gitlink:git-gc[1]:
1527 -------------------------------------------------
1529 -------------------------------------------------
1531 to recompress the archive. This can be very time-consuming, so
1532 you may prefer to run git-gc when you are not doing other work.
1535 [[ensuring-reliability]]
1536 Ensuring reliability
1537 --------------------
1539 [[checking-for-corruption]]
1540 Checking the repository for corruption
1541 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1543 The gitlink:git-fsck[1] command runs a number of self-consistency checks
1544 on the repository, and reports on any problems. This may take some
1545 time. The most common warning by far is about "dangling" objects:
1547 -------------------------------------------------
1549 dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1550 dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1551 dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1552 dangling blob 218761f9d90712d37a9c5e36f406f92202db07eb
1553 dangling commit bf093535a34a4d35731aa2bd90fe6b176302f14f
1554 dangling commit 8e4bec7f2ddaa268bef999853c25755452100f8e
1555 dangling tree d50bb86186bf27b681d25af89d3b5b68382e4085
1556 dangling tree b24c2473f1fd3d91352a624795be026d64c8841f
1558 -------------------------------------------------
1560 Dangling objects are not a problem. At worst they may take up a little
1561 extra disk space. They can sometimes provide a last-resort method of
1562 recovery lost work--see <<dangling-objects>> for details. However, if
1563 you want, you may remove them with gitlink:git-prune[1] or the --prune
1564 option to gitlink:git-gc[1]:
1566 -------------------------------------------------
1568 -------------------------------------------------
1570 This may be time-consuming. Unlike most other git operations (including
1571 git-gc when run without any options), it is not safe to prune while
1572 other git operations are in progress in the same repository.
1574 [[recovering-lost-changes]]
1575 Recovering lost changes
1576 ~~~~~~~~~~~~~~~~~~~~~~~
1582 Say you modify a branch with gitlink:git-reset[1] --hard, and then
1583 realize that the branch was the only reference you had to that point in
1586 Fortunately, git also keeps a log, called a "reflog", of all the
1587 previous values of each branch. So in this case you can still find the
1588 old history using, for example,
1590 -------------------------------------------------
1591 $ git log master@{1}
1592 -------------------------------------------------
1594 This lists the commits reachable from the previous version of the head.
1595 This syntax can be used to with any git command that accepts a commit,
1596 not just with git log. Some other examples:
1598 -------------------------------------------------
1599 $ git show master@{2} # See where the branch pointed 2,
1600 $ git show master@{3} # 3, ... changes ago.
1601 $ gitk master@{yesterday} # See where it pointed yesterday,
1602 $ gitk master@{"1 week ago"} # ... or last week
1603 $ git log --walk-reflogs master # show reflog entries for master
1604 -------------------------------------------------
1606 A separate reflog is kept for the HEAD, so
1608 -------------------------------------------------
1609 $ git show HEAD@{"1 week ago"}
1610 -------------------------------------------------
1612 will show what HEAD pointed to one week ago, not what the current branch
1613 pointed to one week ago. This allows you to see the history of what
1616 The reflogs are kept by default for 30 days, after which they may be
1617 pruned. See gitlink:git-reflog[1] and gitlink:git-gc[1] to learn
1618 how to control this pruning, and see the "SPECIFYING REVISIONS"
1619 section of gitlink:git-rev-parse[1] for details.
1621 Note that the reflog history is very different from normal git history.
1622 While normal history is shared by every repository that works on the
1623 same project, the reflog history is not shared: it tells you only about
1624 how the branches in your local repository have changed over time.
1626 [[dangling-object-recovery]]
1627 Examining dangling objects
1628 ^^^^^^^^^^^^^^^^^^^^^^^^^^
1630 In some situations the reflog may not be able to save you. For example,
1631 suppose you delete a branch, then realize you need the history it
1632 contained. The reflog is also deleted; however, if you have not yet
1633 pruned the repository, then you may still be able to find the lost
1634 commits in the dangling objects that git-fsck reports. See
1635 <<dangling-objects>> for the details.
1637 -------------------------------------------------
1639 dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1640 dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1641 dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1643 -------------------------------------------------
1646 one of those dangling commits with, for example,
1648 ------------------------------------------------
1649 $ gitk 7281251ddd --not --all
1650 ------------------------------------------------
1652 which does what it sounds like: it says that you want to see the commit
1653 history that is described by the dangling commit(s), but not the
1654 history that is described by all your existing branches and tags. Thus
1655 you get exactly the history reachable from that commit that is lost.
1656 (And notice that it might not be just one commit: we only report the
1657 "tip of the line" as being dangling, but there might be a whole deep
1658 and complex commit history that was dropped.)
1660 If you decide you want the history back, you can always create a new
1661 reference pointing to it, for example, a new branch:
1663 ------------------------------------------------
1664 $ git branch recovered-branch 7281251ddd
1665 ------------------------------------------------
1667 Other types of dangling objects (blobs and trees) are also possible, and
1668 dangling objects can arise in other situations.
1671 [[sharing-development]]
1672 Sharing development with others
1673 ===============================
1675 [[getting-updates-with-git-pull]]
1676 Getting updates with git pull
1677 -----------------------------
1679 After you clone a repository and make a few changes of your own, you
1680 may wish to check the original repository for updates and merge them
1683 We have already seen <<Updating-a-repository-with-git-fetch,how to
1684 keep remote tracking branches up to date>> with gitlink:git-fetch[1],
1685 and how to merge two branches. So you can merge in changes from the
1686 original repository's master branch with:
1688 -------------------------------------------------
1690 $ git merge origin/master
1691 -------------------------------------------------
1693 However, the gitlink:git-pull[1] command provides a way to do this in
1696 -------------------------------------------------
1697 $ git pull origin master
1698 -------------------------------------------------
1700 In fact, "origin" is normally the default repository to pull from,
1701 and the default branch is normally the HEAD of the remote repository,
1702 so often you can accomplish the above with just
1704 -------------------------------------------------
1706 -------------------------------------------------
1708 See the descriptions of the branch.<name>.remote and
1709 branch.<name>.merge options in gitlink:git-config[1] to learn
1710 how to control these defaults depending on the current branch.
1712 In addition to saving you keystrokes, "git pull" also helps you by
1713 producing a default commit message documenting the branch and
1714 repository that you pulled from.
1716 (But note that no such commit will be created in the case of a
1717 <<fast-forwards,fast forward>>; instead, your branch will just be
1718 updated to point to the latest commit from the upstream branch.)
1720 The git-pull command can also be given "." as the "remote" repository,
1721 in which case it just merges in a branch from the current repository; so
1724 -------------------------------------------------
1727 -------------------------------------------------
1729 are roughly equivalent. The former is actually very commonly used.
1731 [[submitting-patches]]
1732 Submitting patches to a project
1733 -------------------------------
1735 If you just have a few changes, the simplest way to submit them may
1736 just be to send them as patches in email:
1738 First, use gitlink:git-format-patch[1]; for example:
1740 -------------------------------------------------
1741 $ git format-patch origin
1742 -------------------------------------------------
1744 will produce a numbered series of files in the current directory, one
1745 for each patch in the current branch but not in origin/HEAD.
1747 You can then import these into your mail client and send them by
1748 hand. However, if you have a lot to send at once, you may prefer to
1749 use the gitlink:git-send-email[1] script to automate the process.
1750 Consult the mailing list for your project first to determine how they
1751 prefer such patches be handled.
1753 [[importing-patches]]
1754 Importing patches to a project
1755 ------------------------------
1757 Git also provides a tool called gitlink:git-am[1] (am stands for
1758 "apply mailbox"), for importing such an emailed series of patches.
1759 Just save all of the patch-containing messages, in order, into a
1760 single mailbox file, say "patches.mbox", then run
1762 -------------------------------------------------
1763 $ git am -3 patches.mbox
1764 -------------------------------------------------
1766 Git will apply each patch in order; if any conflicts are found, it
1767 will stop, and you can fix the conflicts as described in
1768 "<<resolving-a-merge,Resolving a merge>>". (The "-3" option tells
1769 git to perform a merge; if you would prefer it just to abort and
1770 leave your tree and index untouched, you may omit that option.)
1772 Once the index is updated with the results of the conflict
1773 resolution, instead of creating a new commit, just run
1775 -------------------------------------------------
1777 -------------------------------------------------
1779 and git will create the commit for you and continue applying the
1780 remaining patches from the mailbox.
1782 The final result will be a series of commits, one for each patch in
1783 the original mailbox, with authorship and commit log message each
1784 taken from the message containing each patch.
1786 [[setting-up-a-public-repository]]
1787 Setting up a public repository
1788 ------------------------------
1790 Another way to submit changes to a project is to simply tell the
1791 maintainer of that project to pull from your repository, exactly as
1792 you did in the section "<<getting-updates-with-git-pull, Getting
1793 updates with git pull>>".
1795 If you and maintainer both have accounts on the same machine, then
1796 then you can just pull changes from each other's repositories
1797 directly; note that all of the commands (gitlink:git-clone[1],
1798 git-fetch[1], git-pull[1], etc.) that accept a URL as an argument
1799 will also accept a local directory name; so, for example, you can
1802 -------------------------------------------------
1803 $ git clone /path/to/repository
1804 $ git pull /path/to/other/repository
1805 -------------------------------------------------
1807 If this sort of setup is inconvenient or impossible, another (more
1808 common) option is to set up a public repository on a public server.
1809 This also allows you to cleanly separate private work in progress
1810 from publicly visible work.
1812 You will continue to do your day-to-day work in your personal
1813 repository, but periodically "push" changes from your personal
1814 repository into your public repository, allowing other developers to
1815 pull from that repository. So the flow of changes, in a situation
1816 where there is one other developer with a public repository, looks
1820 your personal repo ------------------> your public repo
1823 | you pull | they pull
1827 their public repo <------------------- their repo
1829 Now, assume your personal repository is in the directory ~/proj. We
1830 first create a new clone of the repository:
1832 -------------------------------------------------
1833 $ git clone --bare proj-clone.git
1834 -------------------------------------------------
1836 The resulting directory proj-clone.git will contains a "bare" git
1837 repository--it is just the contents of the ".git" directory, without
1838 a checked-out copy of a working directory.
1840 Next, copy proj-clone.git to the server where you plan to host the
1841 public repository. You can use scp, rsync, or whatever is most
1844 If somebody else maintains the public server, they may already have
1845 set up a git service for you, and you may skip to the section
1846 "<<pushing-changes-to-a-public-repository,Pushing changes to a public
1847 repository>>", below.
1849 Otherwise, the following sections explain how to export your newly
1850 created public repository:
1852 [[exporting-via-http]]
1853 Exporting a git repository via http
1854 -----------------------------------
1856 The git protocol gives better performance and reliability, but on a
1857 host with a web server set up, http exports may be simpler to set up.
1859 All you need to do is place the newly created bare git repository in
1860 a directory that is exported by the web server, and make some
1861 adjustments to give web clients some extra information they need:
1863 -------------------------------------------------
1864 $ mv proj.git /home/you/public_html/proj.git
1866 $ git update-server-info
1867 $ chmod a+x hooks/post-update
1868 -------------------------------------------------
1870 (For an explanation of the last two lines, see
1871 gitlink:git-update-server-info[1], and the documentation
1872 link:hooks.txt[Hooks used by git].)
1874 Advertise the url of proj.git. Anybody else should then be able to
1875 clone or pull from that url, for example with a commandline like:
1877 -------------------------------------------------
1878 $ git clone http://yourserver.com/~you/proj.git
1879 -------------------------------------------------
1882 link:howto/setup-git-server-over-http.txt[setup-git-server-over-http]
1883 for a slightly more sophisticated setup using WebDAV which also
1884 allows pushing over http.)
1886 [[exporting-via-git]]
1887 Exporting a git repository via the git protocol
1888 -----------------------------------------------
1890 This is the preferred method.
1892 For now, we refer you to the gitlink:git-daemon[1] man page for
1893 instructions. (See especially the examples section.)
1895 [[pushing-changes-to-a-public-repository]]
1896 Pushing changes to a public repository
1897 --------------------------------------
1899 Note that the two techniques outline above (exporting via
1900 <<exporting-via-http,http>> or <<exporting-via-git,git>>) allow other
1901 maintainers to fetch your latest changes, but they do not allow write
1902 access, which you will need to update the public repository with the
1903 latest changes created in your private repository.
1905 The simplest way to do this is using gitlink:git-push[1] and ssh; to
1906 update the remote branch named "master" with the latest state of your
1907 branch named "master", run
1909 -------------------------------------------------
1910 $ git push ssh://yourserver.com/~you/proj.git master:master
1911 -------------------------------------------------
1915 -------------------------------------------------
1916 $ git push ssh://yourserver.com/~you/proj.git master
1917 -------------------------------------------------
1919 As with git-fetch, git-push will complain if this does not result in
1920 a <<fast-forwards,fast forward>>. Normally this is a sign of
1921 something wrong. However, if you are sure you know what you're
1922 doing, you may force git-push to perform the update anyway by
1923 proceeding the branch name by a plus sign:
1925 -------------------------------------------------
1926 $ git push ssh://yourserver.com/~you/proj.git +master
1927 -------------------------------------------------
1929 As with git-fetch, you may also set up configuration options to
1930 save typing; so, for example, after
1932 -------------------------------------------------
1933 $ cat >.git/config <<EOF
1934 [remote "public-repo"]
1935 url = ssh://yourserver.com/~you/proj.git
1937 -------------------------------------------------
1939 you should be able to perform the above push with just
1941 -------------------------------------------------
1942 $ git push public-repo master
1943 -------------------------------------------------
1945 See the explanations of the remote.<name>.url, branch.<name>.remote,
1946 and remote.<name>.push options in gitlink:git-config[1] for
1949 [[setting-up-a-shared-repository]]
1950 Setting up a shared repository
1951 ------------------------------
1953 Another way to collaborate is by using a model similar to that
1954 commonly used in CVS, where several developers with special rights
1955 all push to and pull from a single shared repository. See
1956 link:cvs-migration.txt[git for CVS users] for instructions on how to
1959 [[setting-up-gitweb]]
1960 Allow web browsing of a repository
1961 ----------------------------------
1963 The gitweb cgi script provides users an easy way to browse your
1964 project's files and history without having to install git; see the file
1965 gitweb/INSTALL in the git source tree for instructions on setting it up.
1967 [[sharing-development-examples]]
1971 TODO: topic branches, typical roles as in everyday.txt, ?
1974 [[cleaning-up-history]]
1975 Rewriting history and maintaining patch series
1976 ==============================================
1978 Normally commits are only added to a project, never taken away or
1979 replaced. Git is designed with this assumption, and violating it will
1980 cause git's merge machinery (for example) to do the wrong thing.
1982 However, there is a situation in which it can be useful to violate this
1986 Creating the perfect patch series
1987 ---------------------------------
1989 Suppose you are a contributor to a large project, and you want to add a
1990 complicated feature, and to present it to the other developers in a way
1991 that makes it easy for them to read your changes, verify that they are
1992 correct, and understand why you made each change.
1994 If you present all of your changes as a single patch (or commit), they
1995 may find that it is too much to digest all at once.
1997 If you present them with the entire history of your work, complete with
1998 mistakes, corrections, and dead ends, they may be overwhelmed.
2000 So the ideal is usually to produce a series of patches such that:
2002 1. Each patch can be applied in order.
2004 2. Each patch includes a single logical change, together with a
2005 message explaining the change.
2007 3. No patch introduces a regression: after applying any initial
2008 part of the series, the resulting project still compiles and
2009 works, and has no bugs that it didn't have before.
2011 4. The complete series produces the same end result as your own
2012 (probably much messier!) development process did.
2014 We will introduce some tools that can help you do this, explain how to
2015 use them, and then explain some of the problems that can arise because
2016 you are rewriting history.
2018 [[using-git-rebase]]
2019 Keeping a patch series up to date using git-rebase
2020 --------------------------------------------------
2022 Suppose that you create a branch "mywork" on a remote-tracking branch
2023 "origin", and create some commits on top of it:
2025 -------------------------------------------------
2026 $ git checkout -b mywork origin
2032 -------------------------------------------------
2034 You have performed no merges into mywork, so it is just a simple linear
2035 sequence of patches on top of "origin":
2037 ................................................
2041 ................................................
2043 Some more interesting work has been done in the upstream project, and
2044 "origin" has advanced:
2046 ................................................
2047 o--o--O--o--o--o <-- origin
2050 ................................................
2052 At this point, you could use "pull" to merge your changes back in;
2053 the result would create a new merge commit, like this:
2055 ................................................
2056 o--o--O--o--o--o <-- origin
2058 a--b--c--m <-- mywork
2059 ................................................
2061 However, if you prefer to keep the history in mywork a simple series of
2062 commits without any merges, you may instead choose to use
2063 gitlink:git-rebase[1]:
2065 -------------------------------------------------
2066 $ git checkout mywork
2068 -------------------------------------------------
2070 This will remove each of your commits from mywork, temporarily saving
2071 them as patches (in a directory named ".dotest"), update mywork to
2072 point at the latest version of origin, then apply each of the saved
2073 patches to the new mywork. The result will look like:
2076 ................................................
2077 o--o--O--o--o--o <-- origin
2079 a'--b'--c' <-- mywork
2080 ................................................
2082 In the process, it may discover conflicts. In that case it will stop
2083 and allow you to fix the conflicts; after fixing conflicts, use "git
2084 add" to update the index with those contents, and then, instead of
2085 running git-commit, just run
2087 -------------------------------------------------
2088 $ git rebase --continue
2089 -------------------------------------------------
2091 and git will continue applying the rest of the patches.
2093 At any point you may use the --abort option to abort this process and
2094 return mywork to the state it had before you started the rebase:
2096 -------------------------------------------------
2097 $ git rebase --abort
2098 -------------------------------------------------
2100 [[modifying-one-commit]]
2101 Modifying a single commit
2102 -------------------------
2104 We saw in <<fixing-a-mistake-by-editing-history>> that you can replace the
2105 most recent commit using
2107 -------------------------------------------------
2108 $ git commit --amend
2109 -------------------------------------------------
2111 which will replace the old commit by a new commit incorporating your
2112 changes, giving you a chance to edit the old commit message first.
2114 You can also use a combination of this and gitlink:git-rebase[1] to edit
2115 commits further back in your history. First, tag the problematic commit with
2117 -------------------------------------------------
2118 $ git tag bad mywork~5
2119 -------------------------------------------------
2121 (Either gitk or git-log may be useful for finding the commit.)
2123 Then check out that commit, edit it, and rebase the rest of the series
2124 on top of it (note that we could check out the commit on a temporary
2125 branch, but instead we're using a <<detached-head,detached head>>):
2127 -------------------------------------------------
2129 $ # make changes here and update the index
2130 $ git commit --amend
2131 $ git rebase --onto HEAD bad mywork
2132 -------------------------------------------------
2134 When you're done, you'll be left with mywork checked out, with the top
2135 patches on mywork reapplied on top of your modified commit. You can
2138 -------------------------------------------------
2140 -------------------------------------------------
2142 Note that the immutable nature of git history means that you haven't really
2143 "modified" existing commits; instead, you have replaced the old commits with
2144 new commits having new object names.
2146 [[reordering-patch-series]]
2147 Reordering or selecting from a patch series
2148 -------------------------------------------
2150 Given one existing commit, the gitlink:git-cherry-pick[1] command
2151 allows you to apply the change introduced by that commit and create a
2152 new commit that records it. So, for example, if "mywork" points to a
2153 series of patches on top of "origin", you might do something like:
2155 -------------------------------------------------
2156 $ git checkout -b mywork-new origin
2157 $ gitk origin..mywork &
2158 -------------------------------------------------
2160 And browse through the list of patches in the mywork branch using gitk,
2161 applying them (possibly in a different order) to mywork-new using
2162 cherry-pick, and possibly modifying them as you go using commit
2165 Another technique is to use git-format-patch to create a series of
2166 patches, then reset the state to before the patches:
2168 -------------------------------------------------
2169 $ git format-patch origin
2170 $ git reset --hard origin
2171 -------------------------------------------------
2173 Then modify, reorder, or eliminate patches as preferred before applying
2174 them again with gitlink:git-am[1].
2176 [[patch-series-tools]]
2180 There are numerous other tools, such as stgit, which exist for the
2181 purpose of maintaining a patch series. These are outside of the scope of
2184 [[problems-with-rewriting-history]]
2185 Problems with rewriting history
2186 -------------------------------
2188 The primary problem with rewriting the history of a branch has to do
2189 with merging. Suppose somebody fetches your branch and merges it into
2190 their branch, with a result something like this:
2192 ................................................
2193 o--o--O--o--o--o <-- origin
2195 t--t--t--m <-- their branch:
2196 ................................................
2198 Then suppose you modify the last three commits:
2200 ................................................
2201 o--o--o <-- new head of origin
2203 o--o--O--o--o--o <-- old head of origin
2204 ................................................
2206 If we examined all this history together in one repository, it will
2209 ................................................
2210 o--o--o <-- new head of origin
2212 o--o--O--o--o--o <-- old head of origin
2214 t--t--t--m <-- their branch:
2215 ................................................
2217 Git has no way of knowing that the new head is an updated version of
2218 the old head; it treats this situation exactly the same as it would if
2219 two developers had independently done the work on the old and new heads
2220 in parallel. At this point, if someone attempts to merge the new head
2221 in to their branch, git will attempt to merge together the two (old and
2222 new) lines of development, instead of trying to replace the old by the
2223 new. The results are likely to be unexpected.
2225 You may still choose to publish branches whose history is rewritten,
2226 and it may be useful for others to be able to fetch those branches in
2227 order to examine or test them, but they should not attempt to pull such
2228 branches into their own work.
2230 For true distributed development that supports proper merging,
2231 published branches should never be rewritten.
2233 [[advanced-branch-management]]
2234 Advanced branch management
2235 ==========================
2237 [[fetching-individual-branches]]
2238 Fetching individual branches
2239 ----------------------------
2241 Instead of using gitlink:git-remote[1], you can also choose just
2242 to update one branch at a time, and to store it locally under an
2245 -------------------------------------------------
2246 $ git fetch origin todo:my-todo-work
2247 -------------------------------------------------
2249 The first argument, "origin", just tells git to fetch from the
2250 repository you originally cloned from. The second argument tells git
2251 to fetch the branch named "todo" from the remote repository, and to
2252 store it locally under the name refs/heads/my-todo-work.
2254 You can also fetch branches from other repositories; so
2256 -------------------------------------------------
2257 $ git fetch git://example.com/proj.git master:example-master
2258 -------------------------------------------------
2260 will create a new branch named "example-master" and store in it the
2261 branch named "master" from the repository at the given URL. If you
2262 already have a branch named example-master, it will attempt to
2263 <<fast-forwards,fast-forward>> to the commit given by example.com's
2264 master branch. In more detail:
2266 [[fetch-fast-forwards]]
2267 git fetch and fast-forwards
2268 ---------------------------
2270 In the previous example, when updating an existing branch, "git
2271 fetch" checks to make sure that the most recent commit on the remote
2272 branch is a descendant of the most recent commit on your copy of the
2273 branch before updating your copy of the branch to point at the new
2274 commit. Git calls this process a <<fast-forwards,fast forward>>.
2276 A fast forward looks something like this:
2278 ................................................
2279 o--o--o--o <-- old head of the branch
2281 o--o--o <-- new head of the branch
2282 ................................................
2285 In some cases it is possible that the new head will *not* actually be
2286 a descendant of the old head. For example, the developer may have
2287 realized she made a serious mistake, and decided to backtrack,
2288 resulting in a situation like:
2290 ................................................
2291 o--o--o--o--a--b <-- old head of the branch
2293 o--o--o <-- new head of the branch
2294 ................................................
2296 In this case, "git fetch" will fail, and print out a warning.
2298 In that case, you can still force git to update to the new head, as
2299 described in the following section. However, note that in the
2300 situation above this may mean losing the commits labeled "a" and "b",
2301 unless you've already created a reference of your own pointing to
2305 Forcing git fetch to do non-fast-forward updates
2306 ------------------------------------------------
2308 If git fetch fails because the new head of a branch is not a
2309 descendant of the old head, you may force the update with:
2311 -------------------------------------------------
2312 $ git fetch git://example.com/proj.git +master:refs/remotes/example/master
2313 -------------------------------------------------
2315 Note the addition of the "+" sign. Be aware that commits that the
2316 old version of example/master pointed at may be lost, as we saw in
2317 the previous section.
2319 [[remote-branch-configuration]]
2320 Configuring remote branches
2321 ---------------------------
2323 We saw above that "origin" is just a shortcut to refer to the
2324 repository that you originally cloned from. This information is
2325 stored in git configuration variables, which you can see using
2326 gitlink:git-config[1]:
2328 -------------------------------------------------
2330 core.repositoryformatversion=0
2332 core.logallrefupdates=true
2333 remote.origin.url=git://git.kernel.org/pub/scm/git/git.git
2334 remote.origin.fetch=+refs/heads/*:refs/remotes/origin/*
2335 branch.master.remote=origin
2336 branch.master.merge=refs/heads/master
2337 -------------------------------------------------
2339 If there are other repositories that you also use frequently, you can
2340 create similar configuration options to save typing; for example,
2343 -------------------------------------------------
2344 $ git config remote.example.url git://example.com/proj.git
2345 -------------------------------------------------
2347 then the following two commands will do the same thing:
2349 -------------------------------------------------
2350 $ git fetch git://example.com/proj.git master:refs/remotes/example/master
2351 $ git fetch example master:refs/remotes/example/master
2352 -------------------------------------------------
2354 Even better, if you add one more option:
2356 -------------------------------------------------
2357 $ git config remote.example.fetch master:refs/remotes/example/master
2358 -------------------------------------------------
2360 then the following commands will all do the same thing:
2362 -------------------------------------------------
2363 $ git fetch git://example.com/proj.git master:ref/remotes/example/master
2364 $ git fetch example master:ref/remotes/example/master
2365 $ git fetch example example/master
2367 -------------------------------------------------
2369 You can also add a "+" to force the update each time:
2371 -------------------------------------------------
2372 $ git config remote.example.fetch +master:ref/remotes/example/master
2373 -------------------------------------------------
2375 Don't do this unless you're sure you won't mind "git fetch" possibly
2376 throwing away commits on mybranch.
2378 Also note that all of the above configuration can be performed by
2379 directly editing the file .git/config instead of using
2380 gitlink:git-config[1].
2382 See gitlink:git-config[1] for more details on the configuration
2383 options mentioned above.
2390 Git depends on two fundamental abstractions: the "object database", and
2391 the "current directory cache" aka "index".
2393 [[the-object-database]]
2397 The object database is literally just a content-addressable collection
2398 of objects. All objects are named by their content, which is
2399 approximated by the SHA1 hash of the object itself. Objects may refer
2400 to other objects (by referencing their SHA1 hash), and so you can
2401 build up a hierarchy of objects.
2403 All objects have a statically determined "type" aka "tag", which is
2404 determined at object creation time, and which identifies the format of
2405 the object (i.e. how it is used, and how it can refer to other
2406 objects). There are currently four different object types: "blob",
2407 "tree", "commit", and "tag".
2409 A <<def_blob_object,"blob" object>> cannot refer to any other object,
2410 and is, as the name implies, a pure storage object containing some
2411 user data. It is used to actually store the file data, i.e. a blob
2412 object is associated with some particular version of some file.
2414 A <<def_tree_object,"tree" object>> is an object that ties one or more
2415 "blob" objects into a directory structure. In addition, a tree object
2416 can refer to other tree objects, thus creating a directory hierarchy.
2418 A <<def_commit_object,"commit" object>> ties such directory hierarchies
2419 together into a <<def_DAG,directed acyclic graph>> of revisions - each
2420 "commit" is associated with exactly one tree (the directory hierarchy at
2421 the time of the commit). In addition, a "commit" refers to one or more
2422 "parent" commit objects that describe the history of how we arrived at
2423 that directory hierarchy.
2425 As a special case, a commit object with no parents is called the "root"
2426 object, and is the point of an initial project commit. Each project
2427 must have at least one root, and while you can tie several different
2428 root objects together into one project by creating a commit object which
2429 has two or more separate roots as its ultimate parents, that's probably
2430 just going to confuse people. So aim for the notion of "one root object
2431 per project", even if git itself does not enforce that.
2433 A <<def_tag_object,"tag" object>> symbolically identifies and can be
2434 used to sign other objects. It contains the identifier and type of
2435 another object, a symbolic name (of course!) and, optionally, a
2438 Regardless of object type, all objects share the following
2439 characteristics: they are all deflated with zlib, and have a header
2440 that not only specifies their type, but also provides size information
2441 about the data in the object. It's worth noting that the SHA1 hash
2442 that is used to name the object is the hash of the original data
2443 plus this header, so `sha1sum` 'file' does not match the object name
2445 (Historical note: in the dawn of the age of git the hash
2446 was the sha1 of the 'compressed' object.)
2448 As a result, the general consistency of an object can always be tested
2449 independently of the contents or the type of the object: all objects can
2450 be validated by verifying that (a) their hashes match the content of the
2451 file and (b) the object successfully inflates to a stream of bytes that
2452 forms a sequence of <ascii type without space> + <space> + <ascii decimal
2453 size> + <byte\0> + <binary object data>.
2455 The structured objects can further have their structure and
2456 connectivity to other objects verified. This is generally done with
2457 the `git-fsck` program, which generates a full dependency graph
2458 of all objects, and verifies their internal consistency (in addition
2459 to just verifying their superficial consistency through the hash).
2461 The object types in some more detail:
2467 A "blob" object is nothing but a binary blob of data, and doesn't
2468 refer to anything else. There is no signature or any other
2469 verification of the data, so while the object is consistent (it 'is'
2470 indexed by its sha1 hash, so the data itself is certainly correct), it
2471 has absolutely no other attributes. No name associations, no
2472 permissions. It is purely a blob of data (i.e. normally "file
2475 In particular, since the blob is entirely defined by its data, if two
2476 files in a directory tree (or in multiple different versions of the
2477 repository) have the same contents, they will share the same blob
2478 object. The object is totally independent of its location in the
2479 directory tree, and renaming a file does not change the object that
2480 file is associated with in any way.
2482 A blob is typically created when gitlink:git-update-index[1]
2483 is run, and its data can be accessed by gitlink:git-cat-file[1].
2489 The next hierarchical object type is the "tree" object. A tree object
2490 is a list of mode/name/blob data, sorted by name. Alternatively, the
2491 mode data may specify a directory mode, in which case instead of
2492 naming a blob, that name is associated with another TREE object.
2494 Like the "blob" object, a tree object is uniquely determined by the
2495 set contents, and so two separate but identical trees will always
2496 share the exact same object. This is true at all levels, i.e. it's
2497 true for a "leaf" tree (which does not refer to any other trees, only
2498 blobs) as well as for a whole subdirectory.
2500 For that reason a "tree" object is just a pure data abstraction: it
2501 has no history, no signatures, no verification of validity, except
2502 that since the contents are again protected by the hash itself, we can
2503 trust that the tree is immutable and its contents never change.
2505 So you can trust the contents of a tree to be valid, the same way you
2506 can trust the contents of a blob, but you don't know where those
2507 contents 'came' from.
2509 Side note on trees: since a "tree" object is a sorted list of
2510 "filename+content", you can create a diff between two trees without
2511 actually having to unpack two trees. Just ignore all common parts,
2512 and your diff will look right. In other words, you can effectively
2513 (and efficiently) tell the difference between any two random trees by
2514 O(n) where "n" is the size of the difference, rather than the size of
2517 Side note 2 on trees: since the name of a "blob" depends entirely and
2518 exclusively on its contents (i.e. there are no names or permissions
2519 involved), you can see trivial renames or permission changes by
2520 noticing that the blob stayed the same. However, renames with data
2521 changes need a smarter "diff" implementation.
2523 A tree is created with gitlink:git-write-tree[1] and
2524 its data can be accessed by gitlink:git-ls-tree[1].
2525 Two trees can be compared with gitlink:git-diff-tree[1].
2531 The "commit" object is an object that introduces the notion of
2532 history into the picture. In contrast to the other objects, it
2533 doesn't just describe the physical state of a tree, it describes how
2534 we got there, and why.
2536 A "commit" is defined by the tree-object that it results in, the
2537 parent commits (zero, one or more) that led up to that point, and a
2538 comment on what happened. Again, a commit is not trusted per se:
2539 the contents are well-defined and "safe" due to the cryptographically
2540 strong signatures at all levels, but there is no reason to believe
2541 that the tree is "good" or that the merge information makes sense.
2542 The parents do not have to actually have any relationship with the
2543 result, for example.
2545 Note on commits: unlike real SCM's, commits do not contain
2546 rename information or file mode change information. All of that is
2547 implicit in the trees involved (the result tree, and the result trees
2548 of the parents), and describing that makes no sense in this idiotic
2551 A commit is created with gitlink:git-commit-tree[1] and
2552 its data can be accessed by gitlink:git-cat-file[1].
2558 An aside on the notion of "trust". Trust is really outside the scope
2559 of "git", but it's worth noting a few things. First off, since
2560 everything is hashed with SHA1, you 'can' trust that an object is
2561 intact and has not been messed with by external sources. So the name
2562 of an object uniquely identifies a known state - just not a state that
2563 you may want to trust.
2565 Furthermore, since the SHA1 signature of a commit refers to the
2566 SHA1 signatures of the tree it is associated with and the signatures
2567 of the parent, a single named commit specifies uniquely a whole set
2568 of history, with full contents. You can't later fake any step of the
2569 way once you have the name of a commit.
2571 So to introduce some real trust in the system, the only thing you need
2572 to do is to digitally sign just 'one' special note, which includes the
2573 name of a top-level commit. Your digital signature shows others
2574 that you trust that commit, and the immutability of the history of
2575 commits tells others that they can trust the whole history.
2577 In other words, you can easily validate a whole archive by just
2578 sending out a single email that tells the people the name (SHA1 hash)
2579 of the top commit, and digitally sign that email using something
2582 To assist in this, git also provides the tag object...
2588 Git provides the "tag" object to simplify creating, managing and
2589 exchanging symbolic and signed tokens. The "tag" object at its
2590 simplest simply symbolically identifies another object by containing
2591 the sha1, type and symbolic name.
2593 However it can optionally contain additional signature information
2594 (which git doesn't care about as long as there's less than 8k of
2595 it). This can then be verified externally to git.
2597 Note that despite the tag features, "git" itself only handles content
2598 integrity; the trust framework (and signature provision and
2599 verification) has to come from outside.
2601 A tag is created with gitlink:git-mktag[1],
2602 its data can be accessed by gitlink:git-cat-file[1],
2603 and the signature can be verified by
2604 gitlink:git-verify-tag[1].
2608 The "index" aka "Current Directory Cache"
2609 -----------------------------------------
2611 The index is a simple binary file, which contains an efficient
2612 representation of a virtual directory content at some random time. It
2613 does so by a simple array that associates a set of names, dates,
2614 permissions and content (aka "blob") objects together. The cache is
2615 always kept ordered by name, and names are unique (with a few very
2616 specific rules) at any point in time, but the cache has no long-term
2617 meaning, and can be partially updated at any time.
2619 In particular, the index certainly does not need to be consistent with
2620 the current directory contents (in fact, most operations will depend on
2621 different ways to make the index 'not' be consistent with the directory
2622 hierarchy), but it has three very important attributes:
2624 '(a) it can re-generate the full state it caches (not just the
2625 directory structure: it contains pointers to the "blob" objects so
2626 that it can regenerate the data too)'
2628 As a special case, there is a clear and unambiguous one-way mapping
2629 from a current directory cache to a "tree object", which can be
2630 efficiently created from just the current directory cache without
2631 actually looking at any other data. So a directory cache at any one
2632 time uniquely specifies one and only one "tree" object (but has
2633 additional data to make it easy to match up that tree object with what
2634 has happened in the directory)
2636 '(b) it has efficient methods for finding inconsistencies between that
2637 cached state ("tree object waiting to be instantiated") and the
2640 '(c) it can additionally efficiently represent information about merge
2641 conflicts between different tree objects, allowing each pathname to be
2642 associated with sufficient information about the trees involved that
2643 you can create a three-way merge between them.'
2645 Those are the ONLY three things that the directory cache does. It's a
2646 cache, and the normal operation is to re-generate it completely from a
2647 known tree object, or update/compare it with a live tree that is being
2648 developed. If you blow the directory cache away entirely, you generally
2649 haven't lost any information as long as you have the name of the tree
2652 At the same time, the index is at the same time also the
2653 staging area for creating new trees, and creating a new tree always
2654 involves a controlled modification of the index file. In particular,
2655 the index file can have the representation of an intermediate tree that
2656 has not yet been instantiated. So the index can be thought of as a
2657 write-back cache, which can contain dirty information that has not yet
2658 been written back to the backing store.
2666 Generally, all "git" operations work on the index file. Some operations
2667 work *purely* on the index file (showing the current state of the
2668 index), but most operations move data to and from the index file. Either
2669 from the database or from the working directory. Thus there are four
2672 [[working-directory-to-index]]
2673 working directory -> index
2674 ~~~~~~~~~~~~~~~~~~~~~~~~~~
2676 You update the index with information from the working directory with
2677 the gitlink:git-update-index[1] command. You
2678 generally update the index information by just specifying the filename
2679 you want to update, like so:
2681 -------------------------------------------------
2682 $ git-update-index filename
2683 -------------------------------------------------
2685 but to avoid common mistakes with filename globbing etc, the command
2686 will not normally add totally new entries or remove old entries,
2687 i.e. it will normally just update existing cache entries.
2689 To tell git that yes, you really do realize that certain files no
2690 longer exist, or that new files should be added, you
2691 should use the `--remove` and `--add` flags respectively.
2693 NOTE! A `--remove` flag does 'not' mean that subsequent filenames will
2694 necessarily be removed: if the files still exist in your directory
2695 structure, the index will be updated with their new status, not
2696 removed. The only thing `--remove` means is that update-cache will be
2697 considering a removed file to be a valid thing, and if the file really
2698 does not exist any more, it will update the index accordingly.
2700 As a special case, you can also do `git-update-index --refresh`, which
2701 will refresh the "stat" information of each index to match the current
2702 stat information. It will 'not' update the object status itself, and
2703 it will only update the fields that are used to quickly test whether
2704 an object still matches its old backing store object.
2706 [[index-to-object-database]]
2707 index -> object database
2708 ~~~~~~~~~~~~~~~~~~~~~~~~
2710 You write your current index file to a "tree" object with the program
2712 -------------------------------------------------
2714 -------------------------------------------------
2716 that doesn't come with any options - it will just write out the
2717 current index into the set of tree objects that describe that state,
2718 and it will return the name of the resulting top-level tree. You can
2719 use that tree to re-generate the index at any time by going in the
2722 [[object-database-to-index]]
2723 object database -> index
2724 ~~~~~~~~~~~~~~~~~~~~~~~~
2726 You read a "tree" file from the object database, and use that to
2727 populate (and overwrite - don't do this if your index contains any
2728 unsaved state that you might want to restore later!) your current
2729 index. Normal operation is just
2731 -------------------------------------------------
2732 $ git-read-tree <sha1 of tree>
2733 -------------------------------------------------
2735 and your index file will now be equivalent to the tree that you saved
2736 earlier. However, that is only your 'index' file: your working
2737 directory contents have not been modified.
2739 [[index-to-working-directory]]
2740 index -> working directory
2741 ~~~~~~~~~~~~~~~~~~~~~~~~~~
2743 You update your working directory from the index by "checking out"
2744 files. This is not a very common operation, since normally you'd just
2745 keep your files updated, and rather than write to your working
2746 directory, you'd tell the index files about the changes in your
2747 working directory (i.e. `git-update-index`).
2749 However, if you decide to jump to a new version, or check out somebody
2750 else's version, or just restore a previous tree, you'd populate your
2751 index file with read-tree, and then you need to check out the result
2754 -------------------------------------------------
2755 $ git-checkout-index filename
2756 -------------------------------------------------
2758 or, if you want to check out all of the index, use `-a`.
2760 NOTE! git-checkout-index normally refuses to overwrite old files, so
2761 if you have an old version of the tree already checked out, you will
2762 need to use the "-f" flag ('before' the "-a" flag or the filename) to
2763 'force' the checkout.
2766 Finally, there are a few odds and ends which are not purely moving
2767 from one representation to the other:
2769 [[tying-it-all-together]]
2770 Tying it all together
2771 ~~~~~~~~~~~~~~~~~~~~~
2773 To commit a tree you have instantiated with "git-write-tree", you'd
2774 create a "commit" object that refers to that tree and the history
2775 behind it - most notably the "parent" commits that preceded it in
2778 Normally a "commit" has one parent: the previous state of the tree
2779 before a certain change was made. However, sometimes it can have two
2780 or more parent commits, in which case we call it a "merge", due to the
2781 fact that such a commit brings together ("merges") two or more
2782 previous states represented by other commits.
2784 In other words, while a "tree" represents a particular directory state
2785 of a working directory, a "commit" represents that state in "time",
2786 and explains how we got there.
2788 You create a commit object by giving it the tree that describes the
2789 state at the time of the commit, and a list of parents:
2791 -------------------------------------------------
2792 $ git-commit-tree <tree> -p <parent> [-p <parent2> ..]
2793 -------------------------------------------------
2795 and then giving the reason for the commit on stdin (either through
2796 redirection from a pipe or file, or by just typing it at the tty).
2798 git-commit-tree will return the name of the object that represents
2799 that commit, and you should save it away for later use. Normally,
2800 you'd commit a new `HEAD` state, and while git doesn't care where you
2801 save the note about that state, in practice we tend to just write the
2802 result to the file pointed at by `.git/HEAD`, so that we can always see
2803 what the last committed state was.
2805 Here is an ASCII art by Jon Loeliger that illustrates how
2806 various pieces fit together.
2834 checkout-index -u | | checkout-index
2845 [[examining-the-data]]
2849 You can examine the data represented in the object database and the
2850 index with various helper tools. For every object, you can use
2851 gitlink:git-cat-file[1] to examine details about the
2854 -------------------------------------------------
2855 $ git-cat-file -t <objectname>
2856 -------------------------------------------------
2858 shows the type of the object, and once you have the type (which is
2859 usually implicit in where you find the object), you can use
2861 -------------------------------------------------
2862 $ git-cat-file blob|tree|commit|tag <objectname>
2863 -------------------------------------------------
2865 to show its contents. NOTE! Trees have binary content, and as a result
2866 there is a special helper for showing that content, called
2867 `git-ls-tree`, which turns the binary content into a more easily
2870 It's especially instructive to look at "commit" objects, since those
2871 tend to be small and fairly self-explanatory. In particular, if you
2872 follow the convention of having the top commit name in `.git/HEAD`,
2875 -------------------------------------------------
2876 $ git-cat-file commit HEAD
2877 -------------------------------------------------
2879 to see what the top commit was.
2881 [[merging-multiple-trees]]
2882 Merging multiple trees
2883 ----------------------
2885 Git helps you do a three-way merge, which you can expand to n-way by
2886 repeating the merge procedure arbitrary times until you finally
2887 "commit" the state. The normal situation is that you'd only do one
2888 three-way merge (two parents), and commit it, but if you like to, you
2889 can do multiple parents in one go.
2891 To do a three-way merge, you need the two sets of "commit" objects
2892 that you want to merge, use those to find the closest common parent (a
2893 third "commit" object), and then use those commit objects to find the
2894 state of the directory ("tree" object) at these points.
2896 To get the "base" for the merge, you first look up the common parent
2899 -------------------------------------------------
2900 $ git-merge-base <commit1> <commit2>
2901 -------------------------------------------------
2903 which will return you the commit they are both based on. You should
2904 now look up the "tree" objects of those commits, which you can easily
2905 do with (for example)
2907 -------------------------------------------------
2908 $ git-cat-file commit <commitname> | head -1
2909 -------------------------------------------------
2911 since the tree object information is always the first line in a commit
2914 Once you know the three trees you are going to merge (the one "original"
2915 tree, aka the common case, and the two "result" trees, aka the branches
2916 you want to merge), you do a "merge" read into the index. This will
2917 complain if it has to throw away your old index contents, so you should
2918 make sure that you've committed those - in fact you would normally
2919 always do a merge against your last commit (which should thus match what
2920 you have in your current index anyway).
2924 -------------------------------------------------
2925 $ git-read-tree -m -u <origtree> <yourtree> <targettree>
2926 -------------------------------------------------
2928 which will do all trivial merge operations for you directly in the
2929 index file, and you can just write the result out with
2933 [[merging-multiple-trees-2]]
2934 Merging multiple trees, continued
2935 ---------------------------------
2937 Sadly, many merges aren't trivial. If there are files that have
2938 been added.moved or removed, or if both branches have modified the
2939 same file, you will be left with an index tree that contains "merge
2940 entries" in it. Such an index tree can 'NOT' be written out to a tree
2941 object, and you will have to resolve any such merge clashes using
2942 other tools before you can write out the result.
2944 You can examine such index state with `git-ls-files --unmerged`
2945 command. An example:
2947 ------------------------------------------------
2948 $ git-read-tree -m $orig HEAD $target
2949 $ git-ls-files --unmerged
2950 100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1 hello.c
2951 100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2 hello.c
2952 100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello.c
2953 ------------------------------------------------
2955 Each line of the `git-ls-files --unmerged` output begins with
2956 the blob mode bits, blob SHA1, 'stage number', and the
2957 filename. The 'stage number' is git's way to say which tree it
2958 came from: stage 1 corresponds to `$orig` tree, stage 2 `HEAD`
2959 tree, and stage3 `$target` tree.
2961 Earlier we said that trivial merges are done inside
2962 `git-read-tree -m`. For example, if the file did not change
2963 from `$orig` to `HEAD` nor `$target`, or if the file changed
2964 from `$orig` to `HEAD` and `$orig` to `$target` the same way,
2965 obviously the final outcome is what is in `HEAD`. What the
2966 above example shows is that file `hello.c` was changed from
2967 `$orig` to `HEAD` and `$orig` to `$target` in a different way.
2968 You could resolve this by running your favorite 3-way merge
2969 program, e.g. `diff3` or `merge`, on the blob objects from
2970 these three stages yourself, like this:
2972 ------------------------------------------------
2973 $ git-cat-file blob 263414f... >hello.c~1
2974 $ git-cat-file blob 06fa6a2... >hello.c~2
2975 $ git-cat-file blob cc44c73... >hello.c~3
2976 $ merge hello.c~2 hello.c~1 hello.c~3
2977 ------------------------------------------------
2979 This would leave the merge result in `hello.c~2` file, along
2980 with conflict markers if there are conflicts. After verifying
2981 the merge result makes sense, you can tell git what the final
2982 merge result for this file is by:
2984 -------------------------------------------------
2985 $ mv -f hello.c~2 hello.c
2986 $ git-update-index hello.c
2987 -------------------------------------------------
2989 When a path is in unmerged state, running `git-update-index` for
2990 that path tells git to mark the path resolved.
2992 The above is the description of a git merge at the lowest level,
2993 to help you understand what conceptually happens under the hood.
2994 In practice, nobody, not even git itself, uses three `git-cat-file`
2995 for this. There is `git-merge-index` program that extracts the
2996 stages to temporary files and calls a "merge" script on it:
2998 -------------------------------------------------
2999 $ git-merge-index git-merge-one-file hello.c
3000 -------------------------------------------------
3002 and that is what higher level `git merge -s resolve` is implemented with.
3005 How git stores objects efficiently: pack files
3006 ----------------------------------------------
3008 We've seen how git stores each object in a file named after the
3011 Unfortunately this system becomes inefficient once a project has a
3012 lot of objects. Try this on an old project:
3014 ------------------------------------------------
3016 6930 objects, 47620 kilobytes
3017 ------------------------------------------------
3019 The first number is the number of objects which are kept in
3020 individual files. The second is the amount of space taken up by
3021 those "loose" objects.
3023 You can save space and make git faster by moving these loose objects in
3024 to a "pack file", which stores a group of objects in an efficient
3025 compressed format; the details of how pack files are formatted can be
3026 found in link:technical/pack-format.txt[technical/pack-format.txt].
3028 To put the loose objects into a pack, just run git repack:
3030 ------------------------------------------------
3033 Done counting 6020 objects.
3034 Deltifying 6020 objects.
3035 100% (6020/6020) done
3036 Writing 6020 objects.
3037 100% (6020/6020) done
3038 Total 6020, written 6020 (delta 4070), reused 0 (delta 0)
3039 Pack pack-3e54ad29d5b2e05838c75df582c65257b8d08e1c created.
3040 ------------------------------------------------
3044 ------------------------------------------------
3046 ------------------------------------------------
3048 to remove any of the "loose" objects that are now contained in the
3049 pack. This will also remove any unreferenced objects (which may be
3050 created when, for example, you use "git reset" to remove a commit).
3051 You can verify that the loose objects are gone by looking at the
3052 .git/objects directory or by running
3054 ------------------------------------------------
3056 0 objects, 0 kilobytes
3057 ------------------------------------------------
3059 Although the object files are gone, any commands that refer to those
3060 objects will work exactly as they did before.
3062 The gitlink:git-gc[1] command performs packing, pruning, and more for
3063 you, so is normally the only high-level command you need.
3065 [[dangling-objects]]
3069 The gitlink:git-fsck[1] command will sometimes complain about dangling
3070 objects. They are not a problem.
3072 The most common cause of dangling objects is that you've rebased a
3073 branch, or you have pulled from somebody else who rebased a branch--see
3074 <<cleaning-up-history>>. In that case, the old head of the original
3075 branch still exists, as does everything it pointed to. The branch
3076 pointer itself just doesn't, since you replaced it with another one.
3078 There are also other situations that cause dangling objects. For
3079 example, a "dangling blob" may arise because you did a "git add" of a
3080 file, but then, before you actually committed it and made it part of the
3081 bigger picture, you changed something else in that file and committed
3082 that *updated* thing - the old state that you added originally ends up
3083 not being pointed to by any commit or tree, so it's now a dangling blob
3086 Similarly, when the "recursive" merge strategy runs, and finds that
3087 there are criss-cross merges and thus more than one merge base (which is
3088 fairly unusual, but it does happen), it will generate one temporary
3089 midway tree (or possibly even more, if you had lots of criss-crossing
3090 merges and more than two merge bases) as a temporary internal merge
3091 base, and again, those are real objects, but the end result will not end
3092 up pointing to them, so they end up "dangling" in your repository.
3094 Generally, dangling objects aren't anything to worry about. They can
3095 even be very useful: if you screw something up, the dangling objects can
3096 be how you recover your old tree (say, you did a rebase, and realized
3097 that you really didn't want to - you can look at what dangling objects
3098 you have, and decide to reset your head to some old dangling state).
3100 For commits, you can just use:
3102 ------------------------------------------------
3103 $ gitk <dangling-commit-sha-goes-here> --not --all
3104 ------------------------------------------------
3106 This asks for all the history reachable from the given commit but not
3107 from any branch, tag, or other reference. If you decide it's something
3108 you want, you can always create a new reference to it, e.g.,
3110 ------------------------------------------------
3111 $ git branch recovered-branch <dangling-commit-sha-goes-here>
3112 ------------------------------------------------
3114 For blobs and trees, you can't do the same, but you can still examine
3115 them. You can just do
3117 ------------------------------------------------
3118 $ git show <dangling-blob/tree-sha-goes-here>
3119 ------------------------------------------------
3121 to show what the contents of the blob were (or, for a tree, basically
3122 what the "ls" for that directory was), and that may give you some idea
3123 of what the operation was that left that dangling object.
3125 Usually, dangling blobs and trees aren't very interesting. They're
3126 almost always the result of either being a half-way mergebase (the blob
3127 will often even have the conflict markers from a merge in it, if you
3128 have had conflicting merges that you fixed up by hand), or simply
3129 because you interrupted a "git fetch" with ^C or something like that,
3130 leaving _some_ of the new objects in the object database, but just
3131 dangling and useless.
3133 Anyway, once you are sure that you're not interested in any dangling
3134 state, you can just prune all unreachable objects:
3136 ------------------------------------------------
3138 ------------------------------------------------
3140 and they'll be gone. But you should only run "git prune" on a quiescent
3141 repository - it's kind of like doing a filesystem fsck recovery: you
3142 don't want to do that while the filesystem is mounted.
3144 (The same is true of "git-fsck" itself, btw - but since
3145 git-fsck never actually *changes* the repository, it just reports
3146 on what it found, git-fsck itself is never "dangerous" to run.
3147 Running it while somebody is actually changing the repository can cause
3148 confusing and scary messages, but it won't actually do anything bad. In
3149 contrast, running "git prune" while somebody is actively changing the
3150 repository is a *BAD* idea).
3153 include::glossary.txt[]
3156 Notes and todo list for this manual
3157 ===================================
3159 This is a work in progress.
3161 The basic requirements:
3162 - It must be readable in order, from beginning to end, by
3163 someone intelligent with a basic grasp of the unix
3164 commandline, but without any special knowledge of git. If
3165 necessary, any other prerequisites should be specifically
3166 mentioned as they arise.
3167 - Whenever possible, section headings should clearly describe
3168 the task they explain how to do, in language that requires
3169 no more knowledge than necessary: for example, "importing
3170 patches into a project" rather than "the git-am command"
3172 Think about how to create a clear chapter dependency graph that will
3173 allow people to get to important topics without necessarily reading
3174 everything in between.
3176 Say something about .gitignore.
3178 Scan Documentation/ for other stuff left out; in particular:
3182 list of commands in gitlink:git[1]
3184 Scan email archives for other stuff left out
3186 Scan man pages to see if any assume more background than this manual
3189 Simplify beginning by suggesting disconnected head instead of
3190 temporary branch creation?
3192 Add more good examples. Entire sections of just cookbook examples
3193 might be a good idea; maybe make an "advanced examples" section a
3194 standard end-of-chapter section?
3196 Include cross-references to the glossary, where appropriate.
3198 Document shallow clones? See draft 1.5.0 release notes for some
3201 Add a section on working with other version control systems, including
3202 CVS, Subversion, and just imports of series of release tarballs.
3204 More details on gitweb?
3206 Write a chapter on using plumbing and writing scripts.