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 ------------------------------------------------
29 This is a quick summary of the major commands; the following chapters
30 will explain how these work in more detail.
32 Creating a new repository
33 -------------------------
37 -----------------------------------------------
38 $ tar xzf project.tar.gz
41 Initialized empty Git repository in .git/
44 -----------------------------------------------
46 From a remote repository:
48 -----------------------------------------------
49 $ git clone git://example.com/pub/project.git
51 -----------------------------------------------
56 -----------------------------------------------
57 $ git branch # list all branches in this repo
58 $ git checkout test # switch working directory to branch "test"
59 $ git branch new # create branch "new" starting at current HEAD
60 $ git branch -d new # delete branch "new"
61 -----------------------------------------------
63 Instead of basing new branch on current HEAD (the default), use:
65 -----------------------------------------------
66 $ git branch new test # branch named "test"
67 $ git branch new v2.6.15 # tag named v2.6.15
68 $ git branch new HEAD^ # commit before the most recent
69 $ git branch new HEAD^^ # commit before that
70 $ git branch new test~10 # ten commits before tip of branch "test"
71 -----------------------------------------------
73 Create and switch to a new branch at the same time:
75 -----------------------------------------------
76 $ git checkout -b new v2.6.15
77 -----------------------------------------------
79 Update and examine branches from the repository you cloned from:
81 -----------------------------------------------
83 $ git branch -r # list
87 $ git branch checkout -b masterwork origin/master
88 -----------------------------------------------
90 Fetch a branch from a different repository, and give it a new
91 name in your repository:
93 -----------------------------------------------
94 $ git fetch git://example.com/project.git theirbranch:mybranch
95 $ git fetch git://example.com/project.git v2.6.15:mybranch
96 -----------------------------------------------
98 Keep a list of repositories you work with regularly:
100 -----------------------------------------------
101 $ git remote add example git://example.com/project.git
102 $ git remote # list remote repositories
105 $ git remote show example # get details
107 URL: git://example.com/project.git
108 Tracked remote branches
110 $ git fetch example # update branches from example
111 $ git branch -r # list all remote branches
112 -----------------------------------------------
118 -----------------------------------------------
119 $ gitk # visualize and browse history
120 $ git log # list all commits
121 $ git log src/ # ...modifying src/
122 $ git log v2.6.15..v2.6.16 # ...in v2.6.16, not in v2.6.15
123 $ git log master..test # ...in branch test, not in branch master
124 $ git log test..master # ...in branch master, but not in test
125 $ git log test...master # ...in one branch, not in both
126 $ git log -S'foo()' # ...where difference contain "foo()"
127 $ git log --since="2 weeks ago"
128 $ git log -p # show patches as well
129 $ git show # most recent commit
130 $ git diff v2.6.15..v2.6.16 # diff between two tagged versions
131 $ git diff v2.6.15..HEAD # diff with current head
132 $ git grep "foo()" # search working directory for "foo()"
133 $ git grep v2.6.15 "foo()" # search old tree for "foo()"
134 $ git show v2.6.15:a.txt # look at old version of a.txt
135 -----------------------------------------------
137 Search for regressions:
139 -----------------------------------------------
141 $ git bisect bad # current version is bad
142 $ git bisect good v2.6.13-rc2 # last known good revision
143 Bisecting: 675 revisions left to test after this
145 $ git bisect good # if this revision is good, or
146 $ git bisect bad # if this revision is bad.
148 -----------------------------------------------
153 Make sure git knows who to blame:
155 ------------------------------------------------
156 $ cat >~/.gitconfig <<\EOF
158 name = Your Name Comes Here
159 email = you@yourdomain.example.com
161 ------------------------------------------------
163 Select file contents to include in the next commit, then make the
166 -----------------------------------------------
167 $ git add a.txt # updated file
168 $ git add b.txt # new file
169 $ git rm c.txt # old file
171 -----------------------------------------------
173 Or, prepare and create the commit in one step:
175 -----------------------------------------------
176 $ git commit d.txt # use latest content only of d.txt
177 $ git commit -a # use latest content of all tracked files
178 -----------------------------------------------
183 -----------------------------------------------
184 $ git merge test # merge branch "test" into the current branch
185 $ git pull git://example.com/project.git master
186 # fetch and merge in remote branch
187 $ git pull . test # equivalent to git merge test
188 -----------------------------------------------
193 Importing or exporting patches:
195 -----------------------------------------------
196 $ git format-patch origin..HEAD # format a patch for each commit
197 # in HEAD but not in origin
198 $ git-am mbox # import patches from the mailbox "mbox"
199 -----------------------------------------------
201 Fetch a branch in a different git repository, then merge into the
204 -----------------------------------------------
205 $ git pull git://example.com/project.git theirbranch
206 -----------------------------------------------
208 Store the fetched branch into a local branch before merging into the
211 -----------------------------------------------
212 $ git pull git://example.com/project.git theirbranch:mybranch
213 -----------------------------------------------
215 After creating commits on a local branch, update the remote
216 branch with your commits:
218 -----------------------------------------------
219 $ git push ssh://example.com/project.git mybranch:theirbranch
220 -----------------------------------------------
222 When remote and local branch are both named "test":
224 -----------------------------------------------
225 $ git push ssh://example.com/project.git test
226 -----------------------------------------------
228 Shortcut version for a frequently used remote repository:
230 -----------------------------------------------
231 $ git remote add example ssh://example.com/project.git
232 $ git push example test
233 -----------------------------------------------
235 Repository maintenance
236 ----------------------
238 Check for corruption:
240 -----------------------------------------------
242 -----------------------------------------------
244 Recompress, remove unused cruft:
246 -----------------------------------------------
248 -----------------------------------------------
250 Repositories and Branches
251 =========================
253 How to get a git repository
254 ---------------------------
256 It will be useful to have a git repository to experiment with as you
259 The best way to get one is by using the gitlink:git-clone[1] command
260 to download a copy of an existing repository for a project that you
261 are interested in. If you don't already have a project in mind, here
262 are some interesting examples:
264 ------------------------------------------------
265 # git itself (approx. 10MB download):
266 $ git clone git://git.kernel.org/pub/scm/git/git.git
267 # the linux kernel (approx. 150MB download):
268 $ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
269 ------------------------------------------------
271 The initial clone may be time-consuming for a large project, but you
272 will only need to clone once.
274 The clone command creates a new directory named after the project
275 ("git" or "linux-2.6" in the examples above). After you cd into this
276 directory, you will see that it contains a copy of the project files,
277 together with a special top-level directory named ".git", which
278 contains all the information about the history of the project.
280 In most of the following, examples will be taken from one of the two
283 How to check out a different version of a project
284 -------------------------------------------------
286 Git is best thought of as a tool for storing the history of a
287 collection of files. It stores the history as a compressed
288 collection of interrelated snapshots (versions) of the project's
291 A single git repository may contain multiple branches. Each branch
292 is a bookmark referencing a particular point in the project history.
293 The gitlink:git-branch[1] command shows you the list of branches:
295 ------------------------------------------------
298 ------------------------------------------------
300 A freshly cloned repository contains a single branch, named "master",
301 and the working directory contains the version of the project
302 referred to by the master branch.
304 Most projects also use tags. Tags, like branches, are references
305 into the project's history, and can be listed using the
306 gitlink:git-tag[1] command:
308 ------------------------------------------------
320 ------------------------------------------------
322 Tags are expected to always point at the same version of a project,
323 while branches are expected to advance as development progresses.
325 Create a new branch pointing to one of these versions and check it
326 out using gitlink:git-checkout[1]:
328 ------------------------------------------------
329 $ git checkout -b new v2.6.13
330 ------------------------------------------------
332 The working directory then reflects the contents that the project had
333 when it was tagged v2.6.13, and gitlink:git-branch[1] shows two
334 branches, with an asterisk marking the currently checked-out branch:
336 ------------------------------------------------
340 ------------------------------------------------
342 If you decide that you'd rather see version 2.6.17, you can modify
343 the current branch to point at v2.6.17 instead, with
345 ------------------------------------------------
346 $ git reset --hard v2.6.17
347 ------------------------------------------------
349 Note that if the current branch was your only reference to a
350 particular point in history, then resetting that branch may leave you
351 with no way to find the history it used to point to; so use this
354 Understanding History: Commits
355 ------------------------------
357 Every change in the history of a project is represented by a commit.
358 The gitlink:git-show[1] command shows the most recent commit on the
361 ------------------------------------------------
363 commit 2b5f6dcce5bf94b9b119e9ed8d537098ec61c3d2
364 Author: Jamal Hadi Salim <hadi@cyberus.ca>
365 Date: Sat Dec 2 22:22:25 2006 -0800
367 [XFRM]: Fix aevent structuring to be more complete.
369 aevents can not uniquely identify an SA. We break the ABI with this
370 patch, but consensus is that since it is not yet utilized by any
371 (known) application then it is fine (better do it now than later).
373 Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
374 Signed-off-by: David S. Miller <davem@davemloft.net>
376 diff --git a/Documentation/networking/xfrm_sync.txt b/Documentation/networking/xfrm_sync.txt
377 index 8be626f..d7aac9d 100644
378 --- a/Documentation/networking/xfrm_sync.txt
379 +++ b/Documentation/networking/xfrm_sync.txt
380 @@ -47,10 +47,13 @@ aevent_id structure looks like:
382 struct xfrm_aevent_id {
383 struct xfrm_usersa_id sa_id;
384 + xfrm_address_t saddr;
389 ------------------------------------------------
391 As you can see, a commit shows who made the latest change, what they
394 Every commit has a 40-hexdigit id, sometimes called the "object name"
395 or the "SHA1 id", shown on the first line of the "git show" output.
396 You can usually refer to a commit by a shorter name, such as a tag or a
397 branch name, but this longer name can also be useful. Most
398 importantly, it is a globally unique name for this commit: so if you
399 tell somebody else the object name (for example in email), then you are
400 guaranteed that name will refer to the same commit in their repository
401 that it does in yours (assuming their repository has that commit at
404 Understanding history: commits, parents, and reachability
405 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
407 Every commit (except the very first commit in a project) also has a
408 parent commit which shows what happened before this commit.
409 Following the chain of parents will eventually take you back to the
410 beginning of the project.
412 However, the commits do not form a simple list; git allows lines of
413 development to diverge and then reconverge, and the point where two
414 lines of development reconverge is called a "merge". The commit
415 representing a merge can therefore have more than one parent, with
416 each parent representing the most recent commit on one of the lines
417 of development leading to that point.
419 The best way to see how this works is using the gitlink:gitk[1]
420 command; running gitk now on a git repository and looking for merge
421 commits will help understand how the git organizes history.
423 In the following, we say that commit X is "reachable" from commit Y
424 if commit X is an ancestor of commit Y. Equivalently, you could say
425 that Y is a descendent of X, or that there is a chain of parents
426 leading from commit Y to commit X.
428 Understanding history: History diagrams
429 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
431 We will sometimes represent git history using diagrams like the one
432 below. Commits are shown as "o", and the links between them with
433 lines drawn with - / and \. Time goes left to right:
441 If we need to talk about a particular commit, the character "o" may
442 be replaced with another letter or number.
444 Understanding history: What is a branch?
445 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
447 Though we've been using the word "branch" to mean a kind of reference
448 to a particular commit, the word branch is also commonly used to
449 refer to the line of commits leading up to that point. In the
450 example above, git may think of the branch named "A" as just a
451 pointer to one particular commit, but we may refer informally to the
452 line of three commits leading up to that point as all being part of
455 If we need to make it clear that we're just talking about the most
456 recent commit on the branch, we may refer to that commit as the
457 "head" of the branch.
459 Manipulating branches
460 ---------------------
462 Creating, deleting, and modifying branches is quick and easy; here's
463 a summary of the commands:
467 git branch <branch>::
468 create a new branch named <branch>, referencing the same
469 point in history as the current branch
470 git branch <branch> <start-point>::
471 create a new branch named <branch>, referencing
472 <start-point>, which may be specified any way you like,
473 including using a branch name or a tag name
474 git branch -d <branch>::
475 delete the branch <branch>; if the branch you are deleting
476 points to a commit which is not reachable from this branch,
477 this command will fail with a warning.
478 git branch -D <branch>::
479 even if the branch points to a commit not reachable
480 from the current branch, you may know that that commit
481 is still reachable from some other branch or tag. In that
482 case it is safe to use this command to force git to delete
484 git checkout <branch>::
485 make the current branch <branch>, updating the working
486 directory to reflect the version referenced by <branch>
487 git checkout -b <new> <start-point>::
488 create a new branch <new> referencing <start-point>, and
491 It is also useful to know that the special symbol "HEAD" can always
492 be used to refer to the current branch.
494 Examining branches from a remote repository
495 -------------------------------------------
497 The "master" branch that was created at the time you cloned is a copy
498 of the HEAD in the repository that you cloned from. That repository
499 may also have had other branches, though, and your local repository
500 keeps branches which track each of those remote branches, which you
501 can view using the "-r" option to gitlink:git-branch[1]:
503 ------------------------------------------------
513 ------------------------------------------------
515 You cannot check out these remote-tracking branches, but you can
516 examine them on a branch of your own, just as you would a tag:
518 ------------------------------------------------
519 $ git checkout -b my-todo-copy origin/todo
520 ------------------------------------------------
522 Note that the name "origin" is just the name that git uses by default
523 to refer to the repository that you cloned from.
525 [[how-git-stores-references]]
526 Naming branches, tags, and other references
527 -------------------------------------------
529 Branches, remote-tracking branches, and tags are all references to
530 commits. All references are named with a slash-separated path name
531 starting with "refs"; the names we've been using so far are actually
534 - The branch "test" is short for "refs/heads/test".
535 - The tag "v2.6.18" is short for "refs/tags/v2.6.18".
536 - "origin/master" is short for "refs/remotes/origin/master".
538 The full name is occasionally useful if, for example, there ever
539 exists a tag and a branch with the same name.
541 As another useful shortcut, if the repository "origin" posesses only
542 a single branch, you can refer to that branch as just "origin".
544 More generally, if you have defined a remote repository named
545 "example", you can refer to the branch in that repository as
546 "example". And for a repository with multiple branches, this will
547 refer to the branch designated as the "HEAD" branch.
549 For the complete list of paths which git checks for references, and
550 the order it uses to decide which to choose when there are multiple
551 references with the same shorthand name, see the "SPECIFYING
552 REVISIONS" section of gitlink:git-rev-parse[1].
554 [[Updating-a-repository-with-git-fetch]]
555 Updating a repository with git fetch
556 ------------------------------------
558 Eventually the developer cloned from will do additional work in her
559 repository, creating new commits and advancing the branches to point
562 The command "git fetch", with no arguments, will update all of the
563 remote-tracking branches to the latest version found in her
564 repository. It will not touch any of your own branches--not even the
565 "master" branch that was created for you on clone.
567 Fetching branches from other repositories
568 -----------------------------------------
570 You can also track branches from repositories other than the one you
571 cloned from, using gitlink:git-remote[1]:
573 -------------------------------------------------
574 $ git remote add linux-nfs git://linux-nfs.org/pub/nfs-2.6.git
576 * refs/remotes/linux-nfs/master: storing branch 'master' ...
578 -------------------------------------------------
580 New remote-tracking branches will be stored under the shorthand name
581 that you gave "git remote add", in this case linux-nfs:
583 -------------------------------------------------
587 -------------------------------------------------
589 If you run "git fetch <remote>" later, the tracking branches for the
590 named <remote> will be updated.
592 If you examine the file .git/config, you will see that git has added
595 -------------------------------------------------
599 url = git://linux-nfs.org/~bfields/git.git
600 fetch = +refs/heads/*:refs/remotes/linux-nfs-read/*
602 -------------------------------------------------
604 This is what causes git to track the remote's branches; you may modify
605 or delete these configuration options by editing .git/config with a
606 text editor. (See the "CONFIGURATION FILE" section of
607 gitlink:git-config[1] for details.)
609 Exploring git history
610 =====================
612 Git is best thought of as a tool for storing the history of a
613 collection of files. It does this by storing compressed snapshots of
614 the contents of a file heirarchy, together with "commits" which show
615 the relationships between these snapshots.
617 Git provides extremely flexible and fast tools for exploring the
618 history of a project.
620 We start with one specialized tool that is useful for finding the
621 commit that introduced a bug into a project.
623 How to use bisect to find a regression
624 --------------------------------------
626 Suppose version 2.6.18 of your project worked, but the version at
627 "master" crashes. Sometimes the best way to find the cause of such a
628 regression is to perform a brute-force search through the project's
629 history to find the particular commit that caused the problem. The
630 gitlink:git-bisect[1] command can help you do this:
632 -------------------------------------------------
634 $ git bisect good v2.6.18
635 $ git bisect bad master
636 Bisecting: 3537 revisions left to test after this
637 [65934a9a028b88e83e2b0f8b36618fe503349f8e] BLOCK: Make USB storage depend on SCSI rather than selecting it [try #6]
638 -------------------------------------------------
640 If you run "git branch" at this point, you'll see that git has
641 temporarily moved you to a new branch named "bisect". This branch
642 points to a commit (with commit id 65934...) that is reachable from
643 v2.6.19 but not from v2.6.18. Compile and test it, and see whether
644 it crashes. Assume it does crash. Then:
646 -------------------------------------------------
648 Bisecting: 1769 revisions left to test after this
649 [7eff82c8b1511017ae605f0c99ac275a7e21b867] i2c-core: Drop useless bitmaskings
650 -------------------------------------------------
652 checks out an older version. Continue like this, telling git at each
653 stage whether the version it gives you is good or bad, and notice
654 that the number of revisions left to test is cut approximately in
657 After about 13 tests (in this case), it will output the commit id of
658 the guilty commit. You can then examine the commit with
659 gitlink:git-show[1], find out who wrote it, and mail them your bug
660 report with the commit id. Finally, run
662 -------------------------------------------------
664 -------------------------------------------------
666 to return you to the branch you were on before and delete the
667 temporary "bisect" branch.
669 Note that the version which git-bisect checks out for you at each
670 point is just a suggestion, and you're free to try a different
671 version if you think it would be a good idea. For example,
672 occasionally you may land on a commit that broke something unrelated;
675 -------------------------------------------------
676 $ git bisect-visualize
677 -------------------------------------------------
679 which will run gitk and label the commit it chose with a marker that
680 says "bisect". Chose a safe-looking commit nearby, note its commit
681 id, and check it out with:
683 -------------------------------------------------
684 $ git reset --hard fb47ddb2db...
685 -------------------------------------------------
687 then test, run "bisect good" or "bisect bad" as appropriate, and
693 We have seen several ways of naming commits already:
695 - 40-hexdigit object name
696 - branch name: refers to the commit at the head of the given
698 - tag name: refers to the commit pointed to by the given tag
699 (we've seen branches and tags are special cases of
700 <<how-git-stores-references,references>>).
701 - HEAD: refers to the head of the current branch
703 There are many more; see the "SPECIFYING REVISIONS" section of the
704 gitlink:git-rev-parse[1] man page for the complete list of ways to
705 name revisions. Some examples:
707 -------------------------------------------------
708 $ git show fb47ddb2 # the first few characters of the object name
709 # are usually enough to specify it uniquely
710 $ git show HEAD^ # the parent of the HEAD commit
711 $ git show HEAD^^ # the grandparent
712 $ git show HEAD~4 # the great-great-grandparent
713 -------------------------------------------------
715 Recall that merge commits may have more than one parent; by default,
716 ^ and ~ follow the first parent listed in the commit, but you can
719 -------------------------------------------------
720 $ git show HEAD^1 # show the first parent of HEAD
721 $ git show HEAD^2 # show the second parent of HEAD
722 -------------------------------------------------
724 In addition to HEAD, there are several other special names for
727 Merges (to be discussed later), as well as operations such as
728 git-reset, which change the currently checked-out commit, generally
729 set ORIG_HEAD to the value HEAD had before the current operation.
731 The git-fetch operation always stores the head of the last fetched
732 branch in FETCH_HEAD. For example, if you run git fetch without
733 specifying a local branch as the target of the operation
735 -------------------------------------------------
736 $ git fetch git://example.com/proj.git theirbranch
737 -------------------------------------------------
739 the fetched commits will still be available from FETCH_HEAD.
741 When we discuss merges we'll also see the special name MERGE_HEAD,
742 which refers to the other branch that we're merging in to the current
745 The gitlink:git-rev-parse[1] command is a low-level command that is
746 occasionally useful for translating some name for a commit to the object
747 name for that commit:
749 -------------------------------------------------
750 $ git rev-parse origin
751 e05db0fd4f31dde7005f075a84f96b360d05984b
752 -------------------------------------------------
757 We can also create a tag to refer to a particular commit; after
760 -------------------------------------------------
761 $ git-tag stable-1 1b2e1d63ff
762 -------------------------------------------------
764 You can use stable-1 to refer to the commit 1b2e1d63ff.
766 This creates a "lightweight" tag. If the tag is a tag you wish to
767 share with others, and possibly sign cryptographically, then you
768 should create a tag object instead; see the gitlink:git-tag[1] man
774 The gitlink:git-log[1] command can show lists of commits. On its
775 own, it shows all commits reachable from the parent commit; but you
776 can also make more specific requests:
778 -------------------------------------------------
779 $ git log v2.5.. # commits since (not reachable from) v2.5
780 $ git log test..master # commits reachable from master but not test
781 $ git log master..test # ...reachable from test but not master
782 $ git log master...test # ...reachable from either test or master,
784 $ git log --since="2 weeks ago" # commits from the last 2 weeks
785 $ git log Makefile # commits which modify Makefile
786 $ git log fs/ # ... which modify any file under fs/
787 $ git log -S'foo()' # commits which add or remove any file data
788 # matching the string 'foo()'
789 -------------------------------------------------
791 And of course you can combine all of these; the following finds
792 commits since v2.5 which touch the Makefile or any file under fs:
794 -------------------------------------------------
795 $ git log v2.5.. Makefile fs/
796 -------------------------------------------------
798 You can also ask git log to show patches:
800 -------------------------------------------------
802 -------------------------------------------------
804 See the "--pretty" option in the gitlink:git-log[1] man page for more
807 Note that git log starts with the most recent commit and works
808 backwards through the parents; however, since git history can contain
809 multiple independent lines of development, the particular order that
810 commits are listed in may be somewhat arbitrary.
815 You can generate diffs between any two versions using
818 -------------------------------------------------
819 $ git diff master..test
820 -------------------------------------------------
822 Sometimes what you want instead is a set of patches:
824 -------------------------------------------------
825 $ git format-patch master..test
826 -------------------------------------------------
828 will generate a file with a patch for each commit reachable from test
829 but not from master. Note that if master also has commits which are
830 not reachable from test, then the combined result of these patches
831 will not be the same as the diff produced by the git-diff example.
833 Viewing old file versions
834 -------------------------
836 You can always view an old version of a file by just checking out the
837 correct revision first. But sometimes it is more convenient to be
838 able to view an old version of a single file without checking
839 anything out; this command does that:
841 -------------------------------------------------
842 $ git show v2.5:fs/locks.c
843 -------------------------------------------------
845 Before the colon may be anything that names a commit, and after it
846 may be any path to a file tracked by git.
851 Check whether two branches point at the same history
852 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
854 Suppose you want to check whether two branches point at the same point
857 -------------------------------------------------
858 $ git diff origin..master
859 -------------------------------------------------
861 will tell you whether the contents of the project are the same at the
862 two branches; in theory, however, it's possible that the same project
863 contents could have been arrived at by two different historical
864 routes. You could compare the object names:
866 -------------------------------------------------
867 $ git rev-list origin
868 e05db0fd4f31dde7005f075a84f96b360d05984b
869 $ git rev-list master
870 e05db0fd4f31dde7005f075a84f96b360d05984b
871 -------------------------------------------------
873 Or you could recall that the ... operator selects all commits
874 contained reachable from either one reference or the other but not
877 -------------------------------------------------
878 $ git log origin...master
879 -------------------------------------------------
881 will return no commits when the two branches are equal.
883 Find first tagged version including a given fix
884 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
886 Suppose you know that the commit e05db0fd fixed a certain problem.
887 You'd like to find the earliest tagged release that contains that
890 Of course, there may be more than one answer--if the history branched
891 after commit e05db0fd, then there could be multiple "earliest" tagged
894 You could just visually inspect the commits since e05db0fd:
896 -------------------------------------------------
898 -------------------------------------------------
900 Or you can use gitlink:git-name-rev[1], which will give the commit a
901 name based on any tag it finds pointing to one of the commit's
904 -------------------------------------------------
905 $ git name-rev e05db0fd
906 e05db0fd tags/v1.5.0-rc1^0~23
907 -------------------------------------------------
909 The gitlink:git-describe[1] command does the opposite, naming the
910 revision using a tag on which the given commit is based:
912 -------------------------------------------------
913 $ git describe e05db0fd
915 -------------------------------------------------
917 but that may sometimes help you guess which tags might come after the
920 If you just want to verify whether a given tagged version contains a
921 given commit, you could use gitlink:git-merge-base[1]:
923 -------------------------------------------------
924 $ git merge-base e05db0fd v1.5.0-rc1
925 e05db0fd4f31dde7005f075a84f96b360d05984b
926 -------------------------------------------------
928 The merge-base command finds a common ancestor of the given commits,
929 and always returns one or the other in the case where one is a
930 descendant of the other; so the above output shows that e05db0fd
931 actually is an ancestor of v1.5.0-rc1.
933 Alternatively, note that
935 -------------------------------------------------
936 $ git log v1.5.0-rc1..e05db0fd
937 -------------------------------------------------
939 will produce empty output if and only if v1.5.0-rc1 includes e05db0fd,
940 because it outputs only commits that are not reachable from v1.5.0-rc1.
942 As yet another alternative, the gitlink:git-show-branch[1] command lists
943 the commits reachable from its arguments with a display on the left-hand
944 side that indicates which arguments that commit is reachable from. So,
945 you can run something like
947 -------------------------------------------------
948 $ git show-branch e05db0fd v1.5.0-rc0 v1.5.0-rc1 v1.5.0-rc2
949 ! [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
951 ! [v1.5.0-rc0] GIT v1.5.0 preview
952 ! [v1.5.0-rc1] GIT v1.5.0-rc1
953 ! [v1.5.0-rc2] GIT v1.5.0-rc2
955 -------------------------------------------------
957 then search for a line that looks like
959 -------------------------------------------------
960 + ++ [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
962 -------------------------------------------------
964 Which shows that e05db0fd is reachable from itself, from v1.5.0-rc1, and
965 from v1.5.0-rc2, but not from v1.5.0-rc0.
971 Telling git your name
972 ---------------------
974 Before creating any commits, you should introduce yourself to git. The
975 easiest way to do so is:
977 ------------------------------------------------
978 $ cat >~/.gitconfig <<\EOF
980 name = Your Name Comes Here
981 email = you@yourdomain.example.com
983 ------------------------------------------------
985 (See the "CONFIGURATION FILE" section of gitlink:git-config[1] for
986 details on the configuration file.)
989 Creating a new repository
990 -------------------------
992 Creating a new repository from scratch is very easy:
994 -------------------------------------------------
998 -------------------------------------------------
1000 If you have some initial content (say, a tarball):
1002 -------------------------------------------------
1003 $ tar -xzvf project.tar.gz
1006 $ git add . # include everything below ./ in the first commit:
1008 -------------------------------------------------
1010 [[how-to-make-a-commit]]
1011 how to make a commit
1012 --------------------
1014 Creating a new commit takes three steps:
1016 1. Making some changes to the working directory using your
1018 2. Telling git about your changes.
1019 3. Creating the commit using the content you told git about
1022 In practice, you can interleave and repeat steps 1 and 2 as many
1023 times as you want: in order to keep track of what you want committed
1024 at step 3, git maintains a snapshot of the tree's contents in a
1025 special staging area called "the index."
1027 At the beginning, the content of the index will be identical to
1028 that of the HEAD. The command "git diff --cached", which shows
1029 the difference between the HEAD and the index, should therefore
1030 produce no output at that point.
1032 Modifying the index is easy:
1034 To update the index with the new contents of a modified file, use
1036 -------------------------------------------------
1037 $ git add path/to/file
1038 -------------------------------------------------
1040 To add the contents of a new file to the index, use
1042 -------------------------------------------------
1043 $ git add path/to/file
1044 -------------------------------------------------
1046 To remove a file from the index and from the working tree,
1048 -------------------------------------------------
1049 $ git rm path/to/file
1050 -------------------------------------------------
1052 After each step you can verify that
1054 -------------------------------------------------
1056 -------------------------------------------------
1058 always shows the difference between the HEAD and the index file--this
1059 is what you'd commit if you created the commit now--and that
1061 -------------------------------------------------
1063 -------------------------------------------------
1065 shows the difference between the working tree and the index file.
1067 Note that "git add" always adds just the current contents of a file
1068 to the index; further changes to the same file will be ignored unless
1069 you run git-add on the file again.
1071 When you're ready, just run
1073 -------------------------------------------------
1075 -------------------------------------------------
1077 and git will prompt you for a commit message and then create the new
1078 commit. Check to make sure it looks like what you expected with
1080 -------------------------------------------------
1082 -------------------------------------------------
1084 As a special shortcut,
1086 -------------------------------------------------
1088 -------------------------------------------------
1090 will update the index with any files that you've modified or removed
1091 and create a commit, all in one step.
1093 A number of commands are useful for keeping track of what you're
1096 -------------------------------------------------
1097 $ git diff --cached # difference between HEAD and the index; what
1098 # would be commited if you ran "commit" now.
1099 $ git diff # difference between the index file and your
1100 # working directory; changes that would not
1101 # be included if you ran "commit" now.
1102 $ git status # a brief per-file summary of the above.
1103 -------------------------------------------------
1105 creating good commit messages
1106 -----------------------------
1108 Though not required, it's a good idea to begin the commit message
1109 with a single short (less than 50 character) line summarizing the
1110 change, followed by a blank line and then a more thorough
1111 description. Tools that turn commits into email, for example, use
1112 the first line on the Subject line and the rest of the commit in the
1118 You can rejoin two diverging branches of development using
1119 gitlink:git-merge[1]:
1121 -------------------------------------------------
1122 $ git merge branchname
1123 -------------------------------------------------
1125 merges the development in the branch "branchname" into the current
1126 branch. If there are conflicts--for example, if the same file is
1127 modified in two different ways in the remote branch and the local
1128 branch--then you are warned; the output may look something like this:
1130 -------------------------------------------------
1132 Trying really trivial in-index merge...
1133 fatal: Merge requires file-level merging
1135 Merging HEAD with 77976da35a11db4580b80ae27e8d65caf5208086
1139 found 1 common ancestor(s):
1141 Auto-merging file.txt
1142 CONFLICT (content): Merge conflict in file.txt
1143 Automatic merge failed; fix conflicts and then commit the result.
1144 -------------------------------------------------
1146 Conflict markers are left in the problematic files, and after
1147 you resolve the conflicts manually, you can update the index
1148 with the contents and run git commit, as you normally would when
1149 creating a new file.
1151 If you examine the resulting commit using gitk, you will see that it
1152 has two parents, one pointing to the top of the current branch, and
1153 one to the top of the other branch.
1157 [[resolving-a-merge]]
1161 When a merge isn't resolved automatically, git leaves the index and
1162 the working tree in a special state that gives you all the
1163 information you need to help resolve the merge.
1165 Files with conflicts are marked specially in the index, so until you
1166 resolve the problem and update the index, git commit will fail:
1168 -------------------------------------------------
1170 file.txt: needs merge
1171 -------------------------------------------------
1173 Also, git status will list those files as "unmerged".
1175 All of the changes that git was able to merge automatically are
1176 already added to the index file, so gitlink:git-diff[1] shows only
1177 the conflicts. Also, it uses a somewhat unusual syntax:
1179 -------------------------------------------------
1182 index 802992c,2b60207..0000000
1185 @@@ -1,1 -1,1 +1,5 @@@
1186 ++<<<<<<< HEAD:file.txt
1190 ++>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1191 -------------------------------------------------
1193 Recall that the commit which will be commited after we resolve this
1194 conflict will have two parents instead of the usual one: one parent
1195 will be HEAD, the tip of the current branch; the other will be the
1196 tip of the other branch, which is stored temporarily in MERGE_HEAD.
1198 The diff above shows the differences between the working-tree version
1199 of file.txt and two previous versions: one version from HEAD, and one
1200 from MERGE_HEAD. So instead of preceding each line by a single "+"
1201 or "-", it now uses two columns: the first column is used for
1202 differences between the first parent and the working directory copy,
1203 and the second for differences between the second parent and the
1204 working directory copy. Thus after resolving the conflict in the
1205 obvious way, the diff will look like:
1207 -------------------------------------------------
1210 index 802992c,2b60207..0000000
1213 @@@ -1,1 -1,1 +1,1 @@@
1217 -------------------------------------------------
1219 This shows that our resolved version deleted "Hello world" from the
1220 first parent, deleted "Goodbye" from the second parent, and added
1221 "Goodbye world", which was previously absent from both.
1223 The gitlink:git-log[1] command also provides special help for merges:
1225 -------------------------------------------------
1227 -------------------------------------------------
1229 This will list all commits which exist only on HEAD or on MERGE_HEAD,
1230 and which touch an unmerged file.
1232 We can now add the resolved version to the index and commit:
1234 -------------------------------------------------
1237 -------------------------------------------------
1239 Note that the commit message will already be filled in for you with
1240 some information about the merge. Normally you can just use this
1241 default message unchanged, but you may add additional commentary of
1242 your own if desired.
1248 If you get stuck and decide to just give up and throw the whole mess
1249 away, you can always return to the pre-merge state with
1251 -------------------------------------------------
1252 $ git reset --hard HEAD
1253 -------------------------------------------------
1255 Or, if you've already commited the merge that you want to throw away,
1257 -------------------------------------------------
1258 $ git reset --hard HEAD^
1259 -------------------------------------------------
1261 However, this last command can be dangerous in some cases--never
1262 throw away a commit you have already committed if that commit may
1263 itself have been merged into another branch, as doing so may confuse
1269 There is one special case not mentioned above, which is treated
1270 differently. Normally, a merge results in a merge commit, with two
1271 parents, one pointing at each of the two lines of development that
1274 However, if one of the two lines of development is completely
1275 contained within the other--so every commit present in the one is
1276 already contained in the other--then git just performs a
1277 <<fast-forwards,fast forward>>; the head of the current branch is
1278 moved forward to point at the head of the merged-in branch, without
1279 any new commits being created.
1284 If you've messed up the working tree, but haven't yet committed your
1285 mistake, you can return the entire working tree to the last committed
1288 -------------------------------------------------
1289 $ git reset --hard HEAD
1290 -------------------------------------------------
1292 If you make a commit that you later wish you hadn't, there are two
1293 fundamentally different ways to fix the problem:
1295 1. You can create a new commit that undoes whatever was done
1296 by the previous commit. This is the correct thing if your
1297 mistake has already been made public.
1299 2. You can go back and modify the old commit. You should
1300 never do this if you have already made the history public;
1301 git does not normally expect the "history" of a project to
1302 change, and cannot correctly perform repeated merges from
1303 a branch that has had its history changed.
1305 Fixing a mistake with a new commit
1306 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1308 Creating a new commit that reverts an earlier change is very easy;
1309 just pass the gitlink:git-revert[1] command a reference to the bad
1310 commit; for example, to revert the most recent commit:
1312 -------------------------------------------------
1314 -------------------------------------------------
1316 This will create a new commit which undoes the change in HEAD. You
1317 will be given a chance to edit the commit message for the new commit.
1319 You can also revert an earlier change, for example, the next-to-last:
1321 -------------------------------------------------
1323 -------------------------------------------------
1325 In this case git will attempt to undo the old change while leaving
1326 intact any changes made since then. If more recent changes overlap
1327 with the changes to be reverted, then you will be asked to fix
1328 conflicts manually, just as in the case of <<resolving-a-merge,
1329 resolving a merge>>.
1331 Fixing a mistake by editing history
1332 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1334 If the problematic commit is the most recent commit, and you have not
1335 yet made that commit public, then you may just
1336 <<undoing-a-merge,destroy it using git-reset>>.
1339 can edit the working directory and update the index to fix your
1340 mistake, just as if you were going to <<how-to-make-a-commit,create a
1341 new commit>>, then run
1343 -------------------------------------------------
1344 $ git commit --amend
1345 -------------------------------------------------
1347 which will replace the old commit by a new commit incorporating your
1348 changes, giving you a chance to edit the old commit message first.
1350 Again, you should never do this to a commit that may already have
1351 been merged into another branch; use gitlink:git-revert[1] instead in
1354 It is also possible to edit commits further back in the history, but
1355 this is an advanced topic to be left for
1356 <<cleaning-up-history,another chapter>>.
1358 Checking out an old version of a file
1359 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1361 In the process of undoing a previous bad change, you may find it
1362 useful to check out an older version of a particular file using
1363 gitlink:git-checkout[1]. We've used git checkout before to switch
1364 branches, but it has quite different behavior if it is given a path
1367 -------------------------------------------------
1368 $ git checkout HEAD^ path/to/file
1369 -------------------------------------------------
1371 replaces path/to/file by the contents it had in the commit HEAD^, and
1372 also updates the index to match. It does not change branches.
1374 If you just want to look at an old version of the file, without
1375 modifying the working directory, you can do that with
1376 gitlink:git-show[1]:
1378 -------------------------------------------------
1379 $ git show HEAD^ path/to/file
1380 -------------------------------------------------
1382 which will display the given version of the file.
1384 Ensuring good performance
1385 -------------------------
1387 On large repositories, git depends on compression to keep the history
1388 information from taking up to much space on disk or in memory.
1390 This compression is not performed automatically. Therefore you
1391 should occasionally run gitlink:git-gc[1]:
1393 -------------------------------------------------
1395 -------------------------------------------------
1397 to recompress the archive. This can be very time-consuming, so
1398 you may prefer to run git-gc when you are not doing other work.
1400 Ensuring reliability
1401 --------------------
1403 Checking the repository for corruption
1404 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1406 The gitlink:git-fsck[1] command runs a number of self-consistency checks
1407 on the repository, and reports on any problems. This may take some
1408 time. The most common warning by far is about "dangling" objects:
1410 -------------------------------------------------
1412 dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1413 dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1414 dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1415 dangling blob 218761f9d90712d37a9c5e36f406f92202db07eb
1416 dangling commit bf093535a34a4d35731aa2bd90fe6b176302f14f
1417 dangling commit 8e4bec7f2ddaa268bef999853c25755452100f8e
1418 dangling tree d50bb86186bf27b681d25af89d3b5b68382e4085
1419 dangling tree b24c2473f1fd3d91352a624795be026d64c8841f
1421 -------------------------------------------------
1423 Dangling objects are objects that are harmless, but also unnecessary;
1424 you can remove them at any time with gitlink:git-prune[1] or the --prune
1425 option to gitlink:git-gc[1]:
1427 -------------------------------------------------
1429 -------------------------------------------------
1431 This may be time-consuming. Unlike most other git operations (including
1432 git-gc when run without any options), it is not safe to prune while
1433 other git operations are in progress in the same repository.
1435 For more about dangling objects, see <<dangling-objects>>.
1438 Recovering lost changes
1439 ~~~~~~~~~~~~~~~~~~~~~~~
1444 Say you modify a branch with gitlink:git-reset[1] --hard, and then
1445 realize that the branch was the only reference you had to that point in
1448 Fortunately, git also keeps a log, called a "reflog", of all the
1449 previous values of each branch. So in this case you can still find the
1450 old history using, for example,
1452 -------------------------------------------------
1453 $ git log master@{1}
1454 -------------------------------------------------
1456 This lists the commits reachable from the previous version of the head.
1457 This syntax can be used to with any git command that accepts a commit,
1458 not just with git log. Some other examples:
1460 -------------------------------------------------
1461 $ git show master@{2} # See where the branch pointed 2,
1462 $ git show master@{3} # 3, ... changes ago.
1463 $ gitk master@{yesterday} # See where it pointed yesterday,
1464 $ gitk master@{"1 week ago"} # ... or last week
1465 -------------------------------------------------
1467 The reflogs are kept by default for 30 days, after which they may be
1468 pruned. See gitlink:git-reflog[1] and gitlink:git-gc[1] to learn
1469 how to control this pruning, and see the "SPECIFYING REVISIONS"
1470 section of gitlink:git-rev-parse[1] for details.
1472 Note that the reflog history is very different from normal git history.
1473 While normal history is shared by every repository that works on the
1474 same project, the reflog history is not shared: it tells you only about
1475 how the branches in your local repository have changed over time.
1477 Examining dangling objects
1478 ^^^^^^^^^^^^^^^^^^^^^^^^^^
1480 In some situations the reflog may not be able to save you. For
1481 example, suppose you delete a branch, then realize you need the history
1482 it contained. The reflog is also deleted; however, if you have not
1483 yet pruned the repository, then you may still be able to find
1484 the lost commits; run git-fsck and watch for output that mentions
1487 -------------------------------------------------
1489 dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1490 dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1491 dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1493 -------------------------------------------------
1496 one of those dangling commits with, for example,
1498 ------------------------------------------------
1499 $ gitk 7281251ddd --not --all
1500 ------------------------------------------------
1502 which does what it sounds like: it says that you want to see the commit
1503 history that is described by the dangling commit(s), but not the
1504 history that is described by all your existing branches and tags. Thus
1505 you get exactly the history reachable from that commit that is lost.
1506 (And notice that it might not be just one commit: we only report the
1507 "tip of the line" as being dangling, but there might be a whole deep
1508 and complex commit history that was dropped.)
1510 If you decide you want the history back, you can always create a new
1511 reference pointing to it, for example, a new branch:
1513 ------------------------------------------------
1514 $ git branch recovered-branch 7281251ddd
1515 ------------------------------------------------
1518 Sharing development with others
1519 ===============================
1521 [[getting-updates-with-git-pull]]
1522 Getting updates with git pull
1523 -----------------------------
1525 After you clone a repository and make a few changes of your own, you
1526 may wish to check the original repository for updates and merge them
1529 We have already seen <<Updating-a-repository-with-git-fetch,how to
1530 keep remote tracking branches up to date>> with gitlink:git-fetch[1],
1531 and how to merge two branches. So you can merge in changes from the
1532 original repository's master branch with:
1534 -------------------------------------------------
1536 $ git merge origin/master
1537 -------------------------------------------------
1539 However, the gitlink:git-pull[1] command provides a way to do this in
1542 -------------------------------------------------
1543 $ git pull origin master
1544 -------------------------------------------------
1546 In fact, "origin" is normally the default repository to pull from,
1547 and the default branch is normally the HEAD of the remote repository,
1548 so often you can accomplish the above with just
1550 -------------------------------------------------
1552 -------------------------------------------------
1554 See the descriptions of the branch.<name>.remote and
1555 branch.<name>.merge options in gitlink:git-config[1] to learn
1556 how to control these defaults depending on the current branch.
1558 In addition to saving you keystrokes, "git pull" also helps you by
1559 producing a default commit message documenting the branch and
1560 repository that you pulled from.
1562 (But note that no such commit will be created in the case of a
1563 <<fast-forwards,fast forward>>; instead, your branch will just be
1564 updated to point to the latest commit from the upstream branch.)
1566 The git-pull command can also be given "." as the "remote" repository,
1567 in which case it just merges in a branch from the current repository; so
1570 -------------------------------------------------
1573 -------------------------------------------------
1575 are roughly equivalent. The former is actually very commonly used.
1577 Submitting patches to a project
1578 -------------------------------
1580 If you just have a few changes, the simplest way to submit them may
1581 just be to send them as patches in email:
1583 First, use gitlink:git-format-patch[1]; for example:
1585 -------------------------------------------------
1586 $ git format-patch origin
1587 -------------------------------------------------
1589 will produce a numbered series of files in the current directory, one
1590 for each patch in the current branch but not in origin/HEAD.
1592 You can then import these into your mail client and send them by
1593 hand. However, if you have a lot to send at once, you may prefer to
1594 use the gitlink:git-send-email[1] script to automate the process.
1595 Consult the mailing list for your project first to determine how they
1596 prefer such patches be handled.
1598 Importing patches to a project
1599 ------------------------------
1601 Git also provides a tool called gitlink:git-am[1] (am stands for
1602 "apply mailbox"), for importing such an emailed series of patches.
1603 Just save all of the patch-containing messages, in order, into a
1604 single mailbox file, say "patches.mbox", then run
1606 -------------------------------------------------
1607 $ git am -3 patches.mbox
1608 -------------------------------------------------
1610 Git will apply each patch in order; if any conflicts are found, it
1611 will stop, and you can fix the conflicts as described in
1612 "<<resolving-a-merge,Resolving a merge>>". (The "-3" option tells
1613 git to perform a merge; if you would prefer it just to abort and
1614 leave your tree and index untouched, you may omit that option.)
1616 Once the index is updated with the results of the conflict
1617 resolution, instead of creating a new commit, just run
1619 -------------------------------------------------
1621 -------------------------------------------------
1623 and git will create the commit for you and continue applying the
1624 remaining patches from the mailbox.
1626 The final result will be a series of commits, one for each patch in
1627 the original mailbox, with authorship and commit log message each
1628 taken from the message containing each patch.
1630 [[setting-up-a-public-repository]]
1631 Setting up a public repository
1632 ------------------------------
1634 Another way to submit changes to a project is to simply tell the
1635 maintainer of that project to pull from your repository, exactly as
1636 you did in the section "<<getting-updates-with-git-pull, Getting
1637 updates with git pull>>".
1639 If you and maintainer both have accounts on the same machine, then
1640 then you can just pull changes from each other's repositories
1641 directly; note that all of the commands (gitlink:git-clone[1],
1642 git-fetch[1], git-pull[1], etc.) that accept a URL as an argument
1643 will also accept a local file patch; so, for example, you can
1646 -------------------------------------------------
1647 $ git clone /path/to/repository
1648 $ git pull /path/to/other/repository
1649 -------------------------------------------------
1651 If this sort of setup is inconvenient or impossible, another (more
1652 common) option is to set up a public repository on a public server.
1653 This also allows you to cleanly separate private work in progress
1654 from publicly visible work.
1656 You will continue to do your day-to-day work in your personal
1657 repository, but periodically "push" changes from your personal
1658 repository into your public repository, allowing other developers to
1659 pull from that repository. So the flow of changes, in a situation
1660 where there is one other developer with a public repository, looks
1664 your personal repo ------------------> your public repo
1667 | you pull | they pull
1671 their public repo <------------------- their repo
1673 Now, assume your personal repository is in the directory ~/proj. We
1674 first create a new clone of the repository:
1676 -------------------------------------------------
1677 $ git clone --bare proj-clone.git
1678 -------------------------------------------------
1680 The resulting directory proj-clone.git will contains a "bare" git
1681 repository--it is just the contents of the ".git" directory, without
1682 a checked-out copy of a working directory.
1684 Next, copy proj-clone.git to the server where you plan to host the
1685 public repository. You can use scp, rsync, or whatever is most
1688 If somebody else maintains the public server, they may already have
1689 set up a git service for you, and you may skip to the section
1690 "<<pushing-changes-to-a-public-repository,Pushing changes to a public
1691 repository>>", below.
1693 Otherwise, the following sections explain how to export your newly
1694 created public repository:
1696 [[exporting-via-http]]
1697 Exporting a git repository via http
1698 -----------------------------------
1700 The git protocol gives better performance and reliability, but on a
1701 host with a web server set up, http exports may be simpler to set up.
1703 All you need to do is place the newly created bare git repository in
1704 a directory that is exported by the web server, and make some
1705 adjustments to give web clients some extra information they need:
1707 -------------------------------------------------
1708 $ mv proj.git /home/you/public_html/proj.git
1710 $ git update-server-info
1711 $ chmod a+x hooks/post-update
1712 -------------------------------------------------
1714 (For an explanation of the last two lines, see
1715 gitlink:git-update-server-info[1], and the documentation
1716 link:hooks.txt[Hooks used by git].)
1718 Advertise the url of proj.git. Anybody else should then be able to
1719 clone or pull from that url, for example with a commandline like:
1721 -------------------------------------------------
1722 $ git clone http://yourserver.com/~you/proj.git
1723 -------------------------------------------------
1726 link:howto/setup-git-server-over-http.txt[setup-git-server-over-http]
1727 for a slightly more sophisticated setup using WebDAV which also
1728 allows pushing over http.)
1730 [[exporting-via-git]]
1731 Exporting a git repository via the git protocol
1732 -----------------------------------------------
1734 This is the preferred method.
1736 For now, we refer you to the gitlink:git-daemon[1] man page for
1737 instructions. (See especially the examples section.)
1739 [[pushing-changes-to-a-public-repository]]
1740 Pushing changes to a public repository
1741 --------------------------------------
1743 Note that the two techniques outline above (exporting via
1744 <<exporting-via-http,http>> or <<exporting-via-git,git>>) allow other
1745 maintainers to fetch your latest changes, but they do not allow write
1746 access, which you will need to update the public repository with the
1747 latest changes created in your private repository.
1749 The simplest way to do this is using gitlink:git-push[1] and ssh; to
1750 update the remote branch named "master" with the latest state of your
1751 branch named "master", run
1753 -------------------------------------------------
1754 $ git push ssh://yourserver.com/~you/proj.git master:master
1755 -------------------------------------------------
1759 -------------------------------------------------
1760 $ git push ssh://yourserver.com/~you/proj.git master
1761 -------------------------------------------------
1763 As with git-fetch, git-push will complain if this does not result in
1764 a <<fast-forwards,fast forward>>. Normally this is a sign of
1765 something wrong. However, if you are sure you know what you're
1766 doing, you may force git-push to perform the update anyway by
1767 proceeding the branch name by a plus sign:
1769 -------------------------------------------------
1770 $ git push ssh://yourserver.com/~you/proj.git +master
1771 -------------------------------------------------
1773 As with git-fetch, you may also set up configuration options to
1774 save typing; so, for example, after
1776 -------------------------------------------------
1777 $ cat >.git/config <<EOF
1778 [remote "public-repo"]
1779 url = ssh://yourserver.com/~you/proj.git
1781 -------------------------------------------------
1783 you should be able to perform the above push with just
1785 -------------------------------------------------
1786 $ git push public-repo master
1787 -------------------------------------------------
1789 See the explanations of the remote.<name>.url, branch.<name>.remote,
1790 and remote.<name>.push options in gitlink:git-config[1] for
1793 Setting up a shared repository
1794 ------------------------------
1796 Another way to collaborate is by using a model similar to that
1797 commonly used in CVS, where several developers with special rights
1798 all push to and pull from a single shared repository. See
1799 link:cvs-migration.txt[git for CVS users] for instructions on how to
1802 Allow web browsing of a repository
1803 ----------------------------------
1805 The gitweb cgi script provides users an easy way to browse your
1806 project's files and history without having to install git; see the file
1807 gitweb/README in the git source tree for instructions on setting it up.
1812 TODO: topic branches, typical roles as in everyday.txt, ?
1815 [[cleaning-up-history]]
1816 Rewriting history and maintaining patch series
1817 ==============================================
1819 Normally commits are only added to a project, never taken away or
1820 replaced. Git is designed with this assumption, and violating it will
1821 cause git's merge machinery (for example) to do the wrong thing.
1823 However, there is a situation in which it can be useful to violate this
1826 Creating the perfect patch series
1827 ---------------------------------
1829 Suppose you are a contributor to a large project, and you want to add a
1830 complicated feature, and to present it to the other developers in a way
1831 that makes it easy for them to read your changes, verify that they are
1832 correct, and understand why you made each change.
1834 If you present all of your changes as a single patch (or commit), they
1835 may find that it is too much to digest all at once.
1837 If you present them with the entire history of your work, complete with
1838 mistakes, corrections, and dead ends, they may be overwhelmed.
1840 So the ideal is usually to produce a series of patches such that:
1842 1. Each patch can be applied in order.
1844 2. Each patch includes a single logical change, together with a
1845 message explaining the change.
1847 3. No patch introduces a regression: after applying any initial
1848 part of the series, the resulting project still compiles and
1849 works, and has no bugs that it didn't have before.
1851 4. The complete series produces the same end result as your own
1852 (probably much messier!) development process did.
1854 We will introduce some tools that can help you do this, explain how to
1855 use them, and then explain some of the problems that can arise because
1856 you are rewriting history.
1858 Keeping a patch series up to date using git-rebase
1859 --------------------------------------------------
1861 Suppose that you create a branch "mywork" on a remote-tracking branch
1862 "origin", and create some commits on top of it:
1864 -------------------------------------------------
1865 $ git checkout -b mywork origin
1871 -------------------------------------------------
1873 You have performed no merges into mywork, so it is just a simple linear
1874 sequence of patches on top of "origin":
1881 Some more interesting work has been done in the upstream project, and
1882 "origin" has advanced:
1884 o--o--O--o--o--o <-- origin
1888 At this point, you could use "pull" to merge your changes back in;
1889 the result would create a new merge commit, like this:
1892 o--o--O--o--o--o <-- origin
1894 a--b--c--m <-- mywork
1896 However, if you prefer to keep the history in mywork a simple series of
1897 commits without any merges, you may instead choose to use
1898 gitlink:git-rebase[1]:
1900 -------------------------------------------------
1901 $ git checkout mywork
1903 -------------------------------------------------
1905 This will remove each of your commits from mywork, temporarily saving
1906 them as patches (in a directory named ".dotest"), update mywork to
1907 point at the latest version of origin, then apply each of the saved
1908 patches to the new mywork. The result will look like:
1911 o--o--O--o--o--o <-- origin
1913 a'--b'--c' <-- mywork
1915 In the process, it may discover conflicts. In that case it will stop
1916 and allow you to fix the conflicts; after fixing conflicts, use "git
1917 add" to update the index with those contents, and then, instead of
1918 running git-commit, just run
1920 -------------------------------------------------
1921 $ git rebase --continue
1922 -------------------------------------------------
1924 and git will continue applying the rest of the patches.
1926 At any point you may use the --abort option to abort this process and
1927 return mywork to the state it had before you started the rebase:
1929 -------------------------------------------------
1930 $ git rebase --abort
1931 -------------------------------------------------
1933 Reordering or selecting from a patch series
1934 -------------------------------------------
1936 Given one existing commit, the gitlink:git-cherry-pick[1] command
1937 allows you to apply the change introduced by that commit and create a
1938 new commit that records it. So, for example, if "mywork" points to a
1939 series of patches on top of "origin", you might do something like:
1941 -------------------------------------------------
1942 $ git checkout -b mywork-new origin
1943 $ gitk origin..mywork &
1944 -------------------------------------------------
1946 And browse through the list of patches in the mywork branch using gitk,
1947 applying them (possibly in a different order) to mywork-new using
1948 cherry-pick, and possibly modifying them as you go using commit
1951 Another technique is to use git-format-patch to create a series of
1952 patches, then reset the state to before the patches:
1954 -------------------------------------------------
1955 $ git format-patch origin
1956 $ git reset --hard origin
1957 -------------------------------------------------
1959 Then modify, reorder, or eliminate patches as preferred before applying
1960 them again with gitlink:git-am[1].
1965 There are numerous other tools, such as stgit, which exist for the
1966 purpose of maintaining a patch series. These are outside of the scope of
1969 Problems with rewriting history
1970 -------------------------------
1972 The primary problem with rewriting the history of a branch has to do
1973 with merging. Suppose somebody fetches your branch and merges it into
1974 their branch, with a result something like this:
1976 o--o--O--o--o--o <-- origin
1978 t--t--t--m <-- their branch:
1980 Then suppose you modify the last three commits:
1982 o--o--o <-- new head of origin
1984 o--o--O--o--o--o <-- old head of origin
1986 If we examined all this history together in one repository, it will
1989 o--o--o <-- new head of origin
1991 o--o--O--o--o--o <-- old head of origin
1993 t--t--t--m <-- their branch:
1995 Git has no way of knowing that the new head is an updated version of
1996 the old head; it treats this situation exactly the same as it would if
1997 two developers had independently done the work on the old and new heads
1998 in parallel. At this point, if someone attempts to merge the new head
1999 in to their branch, git will attempt to merge together the two (old and
2000 new) lines of development, instead of trying to replace the old by the
2001 new. The results are likely to be unexpected.
2003 You may still choose to publish branches whose history is rewritten,
2004 and it may be useful for others to be able to fetch those branches in
2005 order to examine or test them, but they should not attempt to pull such
2006 branches into their own work.
2008 For true distributed development that supports proper merging,
2009 published branches should never be rewritten.
2011 Advanced branch management
2012 ==========================
2014 Fetching individual branches
2015 ----------------------------
2017 Instead of using gitlink:git-remote[1], you can also choose just
2018 to update one branch at a time, and to store it locally under an
2021 -------------------------------------------------
2022 $ git fetch origin todo:my-todo-work
2023 -------------------------------------------------
2025 The first argument, "origin", just tells git to fetch from the
2026 repository you originally cloned from. The second argument tells git
2027 to fetch the branch named "todo" from the remote repository, and to
2028 store it locally under the name refs/heads/my-todo-work.
2030 You can also fetch branches from other repositories; so
2032 -------------------------------------------------
2033 $ git fetch git://example.com/proj.git master:example-master
2034 -------------------------------------------------
2036 will create a new branch named "example-master" and store in it the
2037 branch named "master" from the repository at the given URL. If you
2038 already have a branch named example-master, it will attempt to
2039 "fast-forward" to the commit given by example.com's master branch. So
2040 next we explain what a fast-forward is:
2043 Understanding git history: fast-forwards
2044 ----------------------------------------
2046 In the previous example, when updating an existing branch, "git
2047 fetch" checks to make sure that the most recent commit on the remote
2048 branch is a descendant of the most recent commit on your copy of the
2049 branch before updating your copy of the branch to point at the new
2050 commit. Git calls this process a "fast forward".
2052 A fast forward looks something like this:
2054 o--o--o--o <-- old head of the branch
2056 o--o--o <-- new head of the branch
2059 In some cases it is possible that the new head will *not* actually be
2060 a descendant of the old head. For example, the developer may have
2061 realized she made a serious mistake, and decided to backtrack,
2062 resulting in a situation like:
2064 o--o--o--o--a--b <-- old head of the branch
2066 o--o--o <-- new head of the branch
2070 In this case, "git fetch" will fail, and print out a warning.
2072 In that case, you can still force git to update to the new head, as
2073 described in the following section. However, note that in the
2074 situation above this may mean losing the commits labeled "a" and "b",
2075 unless you've already created a reference of your own pointing to
2078 Forcing git fetch to do non-fast-forward updates
2079 ------------------------------------------------
2081 If git fetch fails because the new head of a branch is not a
2082 descendant of the old head, you may force the update with:
2084 -------------------------------------------------
2085 $ git fetch git://example.com/proj.git +master:refs/remotes/example/master
2086 -------------------------------------------------
2088 Note the addition of the "+" sign. Be aware that commits that the
2089 old version of example/master pointed at may be lost, as we saw in
2090 the previous section.
2092 Configuring remote branches
2093 ---------------------------
2095 We saw above that "origin" is just a shortcut to refer to the
2096 repository that you originally cloned from. This information is
2097 stored in git configuration variables, which you can see using
2098 gitlink:git-config[1]:
2100 -------------------------------------------------
2102 core.repositoryformatversion=0
2104 core.logallrefupdates=true
2105 remote.origin.url=git://git.kernel.org/pub/scm/git/git.git
2106 remote.origin.fetch=+refs/heads/*:refs/remotes/origin/*
2107 branch.master.remote=origin
2108 branch.master.merge=refs/heads/master
2109 -------------------------------------------------
2111 If there are other repositories that you also use frequently, you can
2112 create similar configuration options to save typing; for example,
2115 -------------------------------------------------
2116 $ git config remote.example.url git://example.com/proj.git
2117 -------------------------------------------------
2119 then the following two commands will do the same thing:
2121 -------------------------------------------------
2122 $ git fetch git://example.com/proj.git master:refs/remotes/example/master
2123 $ git fetch example master:refs/remotes/example/master
2124 -------------------------------------------------
2126 Even better, if you add one more option:
2128 -------------------------------------------------
2129 $ git config remote.example.fetch master:refs/remotes/example/master
2130 -------------------------------------------------
2132 then the following commands will all do the same thing:
2134 -------------------------------------------------
2135 $ git fetch git://example.com/proj.git master:ref/remotes/example/master
2136 $ git fetch example master:ref/remotes/example/master
2137 $ git fetch example example/master
2139 -------------------------------------------------
2141 You can also add a "+" to force the update each time:
2143 -------------------------------------------------
2144 $ git config remote.example.fetch +master:ref/remotes/example/master
2145 -------------------------------------------------
2147 Don't do this unless you're sure you won't mind "git fetch" possibly
2148 throwing away commits on mybranch.
2150 Also note that all of the above configuration can be performed by
2151 directly editing the file .git/config instead of using
2152 gitlink:git-config[1].
2154 See gitlink:git-config[1] for more details on the configuration
2155 options mentioned above.
2161 There are two object abstractions: the "object database", and the
2162 "current directory cache" aka "index".
2167 The object database is literally just a content-addressable collection
2168 of objects. All objects are named by their content, which is
2169 approximated by the SHA1 hash of the object itself. Objects may refer
2170 to other objects (by referencing their SHA1 hash), and so you can
2171 build up a hierarchy of objects.
2173 All objects have a statically determined "type" aka "tag", which is
2174 determined at object creation time, and which identifies the format of
2175 the object (i.e. how it is used, and how it can refer to other
2176 objects). There are currently four different object types: "blob",
2177 "tree", "commit" and "tag".
2179 A "blob" object cannot refer to any other object, and is, like the type
2180 implies, a pure storage object containing some user data. It is used to
2181 actually store the file data, i.e. a blob object is associated with some
2182 particular version of some file.
2184 A "tree" object is an object that ties one or more "blob" objects into a
2185 directory structure. In addition, a tree object can refer to other tree
2186 objects, thus creating a directory hierarchy.
2188 A "commit" object ties such directory hierarchies together into
2189 a DAG of revisions - each "commit" is associated with exactly one tree
2190 (the directory hierarchy at the time of the commit). In addition, a
2191 "commit" refers to one or more "parent" commit objects that describe the
2192 history of how we arrived at that directory hierarchy.
2194 As a special case, a commit object with no parents is called the "root"
2195 object, and is the point of an initial project commit. Each project
2196 must have at least one root, and while you can tie several different
2197 root objects together into one project by creating a commit object which
2198 has two or more separate roots as its ultimate parents, that's probably
2199 just going to confuse people. So aim for the notion of "one root object
2200 per project", even if git itself does not enforce that.
2202 A "tag" object symbolically identifies and can be used to sign other
2203 objects. It contains the identifier and type of another object, a
2204 symbolic name (of course!) and, optionally, a signature.
2206 Regardless of object type, all objects share the following
2207 characteristics: they are all deflated with zlib, and have a header
2208 that not only specifies their type, but also provides size information
2209 about the data in the object. It's worth noting that the SHA1 hash
2210 that is used to name the object is the hash of the original data
2211 plus this header, so `sha1sum` 'file' does not match the object name
2213 (Historical note: in the dawn of the age of git the hash
2214 was the sha1 of the 'compressed' object.)
2216 As a result, the general consistency of an object can always be tested
2217 independently of the contents or the type of the object: all objects can
2218 be validated by verifying that (a) their hashes match the content of the
2219 file and (b) the object successfully inflates to a stream of bytes that
2220 forms a sequence of <ascii type without space> + <space> + <ascii decimal
2221 size> + <byte\0> + <binary object data>.
2223 The structured objects can further have their structure and
2224 connectivity to other objects verified. This is generally done with
2225 the `git-fsck` program, which generates a full dependency graph
2226 of all objects, and verifies their internal consistency (in addition
2227 to just verifying their superficial consistency through the hash).
2229 The object types in some more detail:
2234 A "blob" object is nothing but a binary blob of data, and doesn't
2235 refer to anything else. There is no signature or any other
2236 verification of the data, so while the object is consistent (it 'is'
2237 indexed by its sha1 hash, so the data itself is certainly correct), it
2238 has absolutely no other attributes. No name associations, no
2239 permissions. It is purely a blob of data (i.e. normally "file
2242 In particular, since the blob is entirely defined by its data, if two
2243 files in a directory tree (or in multiple different versions of the
2244 repository) have the same contents, they will share the same blob
2245 object. The object is totally independent of its location in the
2246 directory tree, and renaming a file does not change the object that
2247 file is associated with in any way.
2249 A blob is typically created when gitlink:git-update-index[1]
2250 is run, and its data can be accessed by gitlink:git-cat-file[1].
2255 The next hierarchical object type is the "tree" object. A tree object
2256 is a list of mode/name/blob data, sorted by name. Alternatively, the
2257 mode data may specify a directory mode, in which case instead of
2258 naming a blob, that name is associated with another TREE object.
2260 Like the "blob" object, a tree object is uniquely determined by the
2261 set contents, and so two separate but identical trees will always
2262 share the exact same object. This is true at all levels, i.e. it's
2263 true for a "leaf" tree (which does not refer to any other trees, only
2264 blobs) as well as for a whole subdirectory.
2266 For that reason a "tree" object is just a pure data abstraction: it
2267 has no history, no signatures, no verification of validity, except
2268 that since the contents are again protected by the hash itself, we can
2269 trust that the tree is immutable and its contents never change.
2271 So you can trust the contents of a tree to be valid, the same way you
2272 can trust the contents of a blob, but you don't know where those
2273 contents 'came' from.
2275 Side note on trees: since a "tree" object is a sorted list of
2276 "filename+content", you can create a diff between two trees without
2277 actually having to unpack two trees. Just ignore all common parts,
2278 and your diff will look right. In other words, you can effectively
2279 (and efficiently) tell the difference between any two random trees by
2280 O(n) where "n" is the size of the difference, rather than the size of
2283 Side note 2 on trees: since the name of a "blob" depends entirely and
2284 exclusively on its contents (i.e. there are no names or permissions
2285 involved), you can see trivial renames or permission changes by
2286 noticing that the blob stayed the same. However, renames with data
2287 changes need a smarter "diff" implementation.
2289 A tree is created with gitlink:git-write-tree[1] and
2290 its data can be accessed by gitlink:git-ls-tree[1].
2291 Two trees can be compared with gitlink:git-diff-tree[1].
2296 The "commit" object is an object that introduces the notion of
2297 history into the picture. In contrast to the other objects, it
2298 doesn't just describe the physical state of a tree, it describes how
2299 we got there, and why.
2301 A "commit" is defined by the tree-object that it results in, the
2302 parent commits (zero, one or more) that led up to that point, and a
2303 comment on what happened. Again, a commit is not trusted per se:
2304 the contents are well-defined and "safe" due to the cryptographically
2305 strong signatures at all levels, but there is no reason to believe
2306 that the tree is "good" or that the merge information makes sense.
2307 The parents do not have to actually have any relationship with the
2308 result, for example.
2310 Note on commits: unlike real SCM's, commits do not contain
2311 rename information or file mode change information. All of that is
2312 implicit in the trees involved (the result tree, and the result trees
2313 of the parents), and describing that makes no sense in this idiotic
2316 A commit is created with gitlink:git-commit-tree[1] and
2317 its data can be accessed by gitlink:git-cat-file[1].
2322 An aside on the notion of "trust". Trust is really outside the scope
2323 of "git", but it's worth noting a few things. First off, since
2324 everything is hashed with SHA1, you 'can' trust that an object is
2325 intact and has not been messed with by external sources. So the name
2326 of an object uniquely identifies a known state - just not a state that
2327 you may want to trust.
2329 Furthermore, since the SHA1 signature of a commit refers to the
2330 SHA1 signatures of the tree it is associated with and the signatures
2331 of the parent, a single named commit specifies uniquely a whole set
2332 of history, with full contents. You can't later fake any step of the
2333 way once you have the name of a commit.
2335 So to introduce some real trust in the system, the only thing you need
2336 to do is to digitally sign just 'one' special note, which includes the
2337 name of a top-level commit. Your digital signature shows others
2338 that you trust that commit, and the immutability of the history of
2339 commits tells others that they can trust the whole history.
2341 In other words, you can easily validate a whole archive by just
2342 sending out a single email that tells the people the name (SHA1 hash)
2343 of the top commit, and digitally sign that email using something
2346 To assist in this, git also provides the tag object...
2351 Git provides the "tag" object to simplify creating, managing and
2352 exchanging symbolic and signed tokens. The "tag" object at its
2353 simplest simply symbolically identifies another object by containing
2354 the sha1, type and symbolic name.
2356 However it can optionally contain additional signature information
2357 (which git doesn't care about as long as there's less than 8k of
2358 it). This can then be verified externally to git.
2360 Note that despite the tag features, "git" itself only handles content
2361 integrity; the trust framework (and signature provision and
2362 verification) has to come from outside.
2364 A tag is created with gitlink:git-mktag[1],
2365 its data can be accessed by gitlink:git-cat-file[1],
2366 and the signature can be verified by
2367 gitlink:git-verify-tag[1].
2370 The "index" aka "Current Directory Cache"
2371 -----------------------------------------
2373 The index is a simple binary file, which contains an efficient
2374 representation of a virtual directory content at some random time. It
2375 does so by a simple array that associates a set of names, dates,
2376 permissions and content (aka "blob") objects together. The cache is
2377 always kept ordered by name, and names are unique (with a few very
2378 specific rules) at any point in time, but the cache has no long-term
2379 meaning, and can be partially updated at any time.
2381 In particular, the index certainly does not need to be consistent with
2382 the current directory contents (in fact, most operations will depend on
2383 different ways to make the index 'not' be consistent with the directory
2384 hierarchy), but it has three very important attributes:
2386 '(a) it can re-generate the full state it caches (not just the
2387 directory structure: it contains pointers to the "blob" objects so
2388 that it can regenerate the data too)'
2390 As a special case, there is a clear and unambiguous one-way mapping
2391 from a current directory cache to a "tree object", which can be
2392 efficiently created from just the current directory cache without
2393 actually looking at any other data. So a directory cache at any one
2394 time uniquely specifies one and only one "tree" object (but has
2395 additional data to make it easy to match up that tree object with what
2396 has happened in the directory)
2398 '(b) it has efficient methods for finding inconsistencies between that
2399 cached state ("tree object waiting to be instantiated") and the
2402 '(c) it can additionally efficiently represent information about merge
2403 conflicts between different tree objects, allowing each pathname to be
2404 associated with sufficient information about the trees involved that
2405 you can create a three-way merge between them.'
2407 Those are the ONLY three things that the directory cache does. It's a
2408 cache, and the normal operation is to re-generate it completely from a
2409 known tree object, or update/compare it with a live tree that is being
2410 developed. If you blow the directory cache away entirely, you generally
2411 haven't lost any information as long as you have the name of the tree
2414 At the same time, the index is at the same time also the
2415 staging area for creating new trees, and creating a new tree always
2416 involves a controlled modification of the index file. In particular,
2417 the index file can have the representation of an intermediate tree that
2418 has not yet been instantiated. So the index can be thought of as a
2419 write-back cache, which can contain dirty information that has not yet
2420 been written back to the backing store.
2427 Generally, all "git" operations work on the index file. Some operations
2428 work *purely* on the index file (showing the current state of the
2429 index), but most operations move data to and from the index file. Either
2430 from the database or from the working directory. Thus there are four
2433 working directory -> index
2434 ~~~~~~~~~~~~~~~~~~~~~~~~~~
2436 You update the index with information from the working directory with
2437 the gitlink:git-update-index[1] command. You
2438 generally update the index information by just specifying the filename
2439 you want to update, like so:
2441 -------------------------------------------------
2442 $ git-update-index filename
2443 -------------------------------------------------
2445 but to avoid common mistakes with filename globbing etc, the command
2446 will not normally add totally new entries or remove old entries,
2447 i.e. it will normally just update existing cache entries.
2449 To tell git that yes, you really do realize that certain files no
2450 longer exist, or that new files should be added, you
2451 should use the `--remove` and `--add` flags respectively.
2453 NOTE! A `--remove` flag does 'not' mean that subsequent filenames will
2454 necessarily be removed: if the files still exist in your directory
2455 structure, the index will be updated with their new status, not
2456 removed. The only thing `--remove` means is that update-cache will be
2457 considering a removed file to be a valid thing, and if the file really
2458 does not exist any more, it will update the index accordingly.
2460 As a special case, you can also do `git-update-index --refresh`, which
2461 will refresh the "stat" information of each index to match the current
2462 stat information. It will 'not' update the object status itself, and
2463 it will only update the fields that are used to quickly test whether
2464 an object still matches its old backing store object.
2466 index -> object database
2467 ~~~~~~~~~~~~~~~~~~~~~~~~
2469 You write your current index file to a "tree" object with the program
2471 -------------------------------------------------
2473 -------------------------------------------------
2475 that doesn't come with any options - it will just write out the
2476 current index into the set of tree objects that describe that state,
2477 and it will return the name of the resulting top-level tree. You can
2478 use that tree to re-generate the index at any time by going in the
2481 object database -> index
2482 ~~~~~~~~~~~~~~~~~~~~~~~~
2484 You read a "tree" file from the object database, and use that to
2485 populate (and overwrite - don't do this if your index contains any
2486 unsaved state that you might want to restore later!) your current
2487 index. Normal operation is just
2489 -------------------------------------------------
2490 $ git-read-tree <sha1 of tree>
2491 -------------------------------------------------
2493 and your index file will now be equivalent to the tree that you saved
2494 earlier. However, that is only your 'index' file: your working
2495 directory contents have not been modified.
2497 index -> working directory
2498 ~~~~~~~~~~~~~~~~~~~~~~~~~~
2500 You update your working directory from the index by "checking out"
2501 files. This is not a very common operation, since normally you'd just
2502 keep your files updated, and rather than write to your working
2503 directory, you'd tell the index files about the changes in your
2504 working directory (i.e. `git-update-index`).
2506 However, if you decide to jump to a new version, or check out somebody
2507 else's version, or just restore a previous tree, you'd populate your
2508 index file with read-tree, and then you need to check out the result
2511 -------------------------------------------------
2512 $ git-checkout-index filename
2513 -------------------------------------------------
2515 or, if you want to check out all of the index, use `-a`.
2517 NOTE! git-checkout-index normally refuses to overwrite old files, so
2518 if you have an old version of the tree already checked out, you will
2519 need to use the "-f" flag ('before' the "-a" flag or the filename) to
2520 'force' the checkout.
2523 Finally, there are a few odds and ends which are not purely moving
2524 from one representation to the other:
2526 Tying it all together
2527 ~~~~~~~~~~~~~~~~~~~~~
2529 To commit a tree you have instantiated with "git-write-tree", you'd
2530 create a "commit" object that refers to that tree and the history
2531 behind it - most notably the "parent" commits that preceded it in
2534 Normally a "commit" has one parent: the previous state of the tree
2535 before a certain change was made. However, sometimes it can have two
2536 or more parent commits, in which case we call it a "merge", due to the
2537 fact that such a commit brings together ("merges") two or more
2538 previous states represented by other commits.
2540 In other words, while a "tree" represents a particular directory state
2541 of a working directory, a "commit" represents that state in "time",
2542 and explains how we got there.
2544 You create a commit object by giving it the tree that describes the
2545 state at the time of the commit, and a list of parents:
2547 -------------------------------------------------
2548 $ git-commit-tree <tree> -p <parent> [-p <parent2> ..]
2549 -------------------------------------------------
2551 and then giving the reason for the commit on stdin (either through
2552 redirection from a pipe or file, or by just typing it at the tty).
2554 git-commit-tree will return the name of the object that represents
2555 that commit, and you should save it away for later use. Normally,
2556 you'd commit a new `HEAD` state, and while git doesn't care where you
2557 save the note about that state, in practice we tend to just write the
2558 result to the file pointed at by `.git/HEAD`, so that we can always see
2559 what the last committed state was.
2561 Here is an ASCII art by Jon Loeliger that illustrates how
2562 various pieces fit together.
2590 checkout-index -u | | checkout-index
2604 You can examine the data represented in the object database and the
2605 index with various helper tools. For every object, you can use
2606 gitlink:git-cat-file[1] to examine details about the
2609 -------------------------------------------------
2610 $ git-cat-file -t <objectname>
2611 -------------------------------------------------
2613 shows the type of the object, and once you have the type (which is
2614 usually implicit in where you find the object), you can use
2616 -------------------------------------------------
2617 $ git-cat-file blob|tree|commit|tag <objectname>
2618 -------------------------------------------------
2620 to show its contents. NOTE! Trees have binary content, and as a result
2621 there is a special helper for showing that content, called
2622 `git-ls-tree`, which turns the binary content into a more easily
2625 It's especially instructive to look at "commit" objects, since those
2626 tend to be small and fairly self-explanatory. In particular, if you
2627 follow the convention of having the top commit name in `.git/HEAD`,
2630 -------------------------------------------------
2631 $ git-cat-file commit HEAD
2632 -------------------------------------------------
2634 to see what the top commit was.
2636 Merging multiple trees
2637 ----------------------
2639 Git helps you do a three-way merge, which you can expand to n-way by
2640 repeating the merge procedure arbitrary times until you finally
2641 "commit" the state. The normal situation is that you'd only do one
2642 three-way merge (two parents), and commit it, but if you like to, you
2643 can do multiple parents in one go.
2645 To do a three-way merge, you need the two sets of "commit" objects
2646 that you want to merge, use those to find the closest common parent (a
2647 third "commit" object), and then use those commit objects to find the
2648 state of the directory ("tree" object) at these points.
2650 To get the "base" for the merge, you first look up the common parent
2653 -------------------------------------------------
2654 $ git-merge-base <commit1> <commit2>
2655 -------------------------------------------------
2657 which will return you the commit they are both based on. You should
2658 now look up the "tree" objects of those commits, which you can easily
2659 do with (for example)
2661 -------------------------------------------------
2662 $ git-cat-file commit <commitname> | head -1
2663 -------------------------------------------------
2665 since the tree object information is always the first line in a commit
2668 Once you know the three trees you are going to merge (the one "original"
2669 tree, aka the common case, and the two "result" trees, aka the branches
2670 you want to merge), you do a "merge" read into the index. This will
2671 complain if it has to throw away your old index contents, so you should
2672 make sure that you've committed those - in fact you would normally
2673 always do a merge against your last commit (which should thus match what
2674 you have in your current index anyway).
2678 -------------------------------------------------
2679 $ git-read-tree -m -u <origtree> <yourtree> <targettree>
2680 -------------------------------------------------
2682 which will do all trivial merge operations for you directly in the
2683 index file, and you can just write the result out with
2687 Merging multiple trees, continued
2688 ---------------------------------
2690 Sadly, many merges aren't trivial. If there are files that have
2691 been added.moved or removed, or if both branches have modified the
2692 same file, you will be left with an index tree that contains "merge
2693 entries" in it. Such an index tree can 'NOT' be written out to a tree
2694 object, and you will have to resolve any such merge clashes using
2695 other tools before you can write out the result.
2697 You can examine such index state with `git-ls-files --unmerged`
2698 command. An example:
2700 ------------------------------------------------
2701 $ git-read-tree -m $orig HEAD $target
2702 $ git-ls-files --unmerged
2703 100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1 hello.c
2704 100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2 hello.c
2705 100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello.c
2706 ------------------------------------------------
2708 Each line of the `git-ls-files --unmerged` output begins with
2709 the blob mode bits, blob SHA1, 'stage number', and the
2710 filename. The 'stage number' is git's way to say which tree it
2711 came from: stage 1 corresponds to `$orig` tree, stage 2 `HEAD`
2712 tree, and stage3 `$target` tree.
2714 Earlier we said that trivial merges are done inside
2715 `git-read-tree -m`. For example, if the file did not change
2716 from `$orig` to `HEAD` nor `$target`, or if the file changed
2717 from `$orig` to `HEAD` and `$orig` to `$target` the same way,
2718 obviously the final outcome is what is in `HEAD`. What the
2719 above example shows is that file `hello.c` was changed from
2720 `$orig` to `HEAD` and `$orig` to `$target` in a different way.
2721 You could resolve this by running your favorite 3-way merge
2722 program, e.g. `diff3` or `merge`, on the blob objects from
2723 these three stages yourself, like this:
2725 ------------------------------------------------
2726 $ git-cat-file blob 263414f... >hello.c~1
2727 $ git-cat-file blob 06fa6a2... >hello.c~2
2728 $ git-cat-file blob cc44c73... >hello.c~3
2729 $ merge hello.c~2 hello.c~1 hello.c~3
2730 ------------------------------------------------
2732 This would leave the merge result in `hello.c~2` file, along
2733 with conflict markers if there are conflicts. After verifying
2734 the merge result makes sense, you can tell git what the final
2735 merge result for this file is by:
2737 -------------------------------------------------
2738 $ mv -f hello.c~2 hello.c
2739 $ git-update-index hello.c
2740 -------------------------------------------------
2742 When a path is in unmerged state, running `git-update-index` for
2743 that path tells git to mark the path resolved.
2745 The above is the description of a git merge at the lowest level,
2746 to help you understand what conceptually happens under the hood.
2747 In practice, nobody, not even git itself, uses three `git-cat-file`
2748 for this. There is `git-merge-index` program that extracts the
2749 stages to temporary files and calls a "merge" script on it:
2751 -------------------------------------------------
2752 $ git-merge-index git-merge-one-file hello.c
2753 -------------------------------------------------
2755 and that is what higher level `git merge -s resolve` is implemented with.
2757 How git stores objects efficiently: pack files
2758 ----------------------------------------------
2760 We've seen how git stores each object in a file named after the
2763 Unfortunately this system becomes inefficient once a project has a
2764 lot of objects. Try this on an old project:
2766 ------------------------------------------------
2768 6930 objects, 47620 kilobytes
2769 ------------------------------------------------
2771 The first number is the number of objects which are kept in
2772 individual files. The second is the amount of space taken up by
2773 those "loose" objects.
2775 You can save space and make git faster by moving these loose objects in
2776 to a "pack file", which stores a group of objects in an efficient
2777 compressed format; the details of how pack files are formatted can be
2778 found in link:technical/pack-format.txt[technical/pack-format.txt].
2780 To put the loose objects into a pack, just run git repack:
2782 ------------------------------------------------
2785 Done counting 6020 objects.
2786 Deltifying 6020 objects.
2787 100% (6020/6020) done
2788 Writing 6020 objects.
2789 100% (6020/6020) done
2790 Total 6020, written 6020 (delta 4070), reused 0 (delta 0)
2791 Pack pack-3e54ad29d5b2e05838c75df582c65257b8d08e1c created.
2792 ------------------------------------------------
2796 ------------------------------------------------
2798 ------------------------------------------------
2800 to remove any of the "loose" objects that are now contained in the
2801 pack. This will also remove any unreferenced objects (which may be
2802 created when, for example, you use "git reset" to remove a commit).
2803 You can verify that the loose objects are gone by looking at the
2804 .git/objects directory or by running
2806 ------------------------------------------------
2808 0 objects, 0 kilobytes
2809 ------------------------------------------------
2811 Although the object files are gone, any commands that refer to those
2812 objects will work exactly as they did before.
2814 The gitlink:git-gc[1] command performs packing, pruning, and more for
2815 you, so is normally the only high-level command you need.
2817 [[dangling-objects]]
2821 The gitlink:git-fsck[1] command will sometimes complain about dangling
2822 objects. They are not a problem.
2824 The most common cause of dangling objects is that you've rebased a
2825 branch, or you have pulled from somebody else who rebased a branch--see
2826 <<cleaning-up-history>>. In that case, the old head of the original
2827 branch still exists, as does obviously everything it pointed to. The
2828 branch pointer itself just doesn't, since you replaced it with another
2831 There are also other situations too that cause dangling objects. For
2832 example, a "dangling blob" may arise because you did a "git add" of a
2833 file, but then, before you actually committed it and made it part of the
2834 bigger picture, you changed something else in that file and committed
2835 that *updated* thing - the old state that you added originally ends up
2836 not being pointed to by any commit or tree, so it's now a dangling blob
2839 Similarly, when the "recursive" merge strategy runs, and finds that
2840 there are criss-cross merges and thus more than one merge base (which is
2841 fairly unusual, but it does happen), it will generate one temporary
2842 midway tree (or possibly even more, if you had lots of criss-crossing
2843 merges and more than two merge bases) as a temporary internal merge
2844 base, and again, those are real objects, but the end result will not end
2845 up pointing to them, so they end up "dangling" in your repository.
2847 Generally, dangling objects aren't anything to worry about. They can
2848 even be very useful: if you screw something up, the dangling objects can
2849 be how you recover your old tree (say, you did a rebase, and realized
2850 that you really didn't want to - you can look at what dangling objects
2851 you have, and decide to reset your head to some old dangling state).
2853 For commits, the most useful thing to do with dangling objects tends to
2856 ------------------------------------------------
2857 $ gitk <dangling-commit-sha-goes-here> --not --all
2858 ------------------------------------------------
2860 For blobs and trees, you can't do the same, but you can examine them.
2863 ------------------------------------------------
2864 $ git show <dangling-blob/tree-sha-goes-here>
2865 ------------------------------------------------
2867 to show what the contents of the blob were (or, for a tree, basically
2868 what the "ls" for that directory was), and that may give you some idea
2869 of what the operation was that left that dangling object.
2871 Usually, dangling blobs and trees aren't very interesting. They're
2872 almost always the result of either being a half-way mergebase (the blob
2873 will often even have the conflict markers from a merge in it, if you
2874 have had conflicting merges that you fixed up by hand), or simply
2875 because you interrupted a "git fetch" with ^C or something like that,
2876 leaving _some_ of the new objects in the object database, but just
2877 dangling and useless.
2879 Anyway, once you are sure that you're not interested in any dangling
2880 state, you can just prune all unreachable objects:
2882 ------------------------------------------------
2884 ------------------------------------------------
2886 and they'll be gone. But you should only run "git prune" on a quiescent
2887 repository - it's kind of like doing a filesystem fsck recovery: you
2888 don't want to do that while the filesystem is mounted.
2890 (The same is true of "git-fsck" itself, btw - but since
2891 git-fsck never actually *changes* the repository, it just reports
2892 on what it found, git-fsck itself is never "dangerous" to run.
2893 Running it while somebody is actually changing the repository can cause
2894 confusing and scary messages, but it won't actually do anything bad. In
2895 contrast, running "git prune" while somebody is actively changing the
2896 repository is a *BAD* idea).
2898 Glossary of git terms
2899 =====================
2901 include::glossary.txt[]
2903 Notes and todo list for this manual
2904 ===================================
2906 This is a work in progress.
2908 The basic requirements:
2909 - It must be readable in order, from beginning to end, by
2910 someone intelligent with a basic grasp of the unix
2911 commandline, but without any special knowledge of git. If
2912 necessary, any other prerequisites should be specifically
2913 mentioned as they arise.
2914 - Whenever possible, section headings should clearly describe
2915 the task they explain how to do, in language that requires
2916 no more knowledge than necessary: for example, "importing
2917 patches into a project" rather than "the git-am command"
2919 Think about how to create a clear chapter dependency graph that will
2920 allow people to get to important topics without necessarily reading
2921 everything in between.
2923 Say something about .gitignore.
2925 Scan Documentation/ for other stuff left out; in particular:
2929 list of commands in gitlink:git[1]
2931 Scan email archives for other stuff left out
2933 Scan man pages to see if any assume more background than this manual
2936 Simplify beginning by suggesting disconnected head instead of
2937 temporary branch creation?
2939 Explain how to refer to file stages in the "how to resolve a merge"
2940 section: diff -1, -2, -3, --ours, --theirs :1:/path notation. The
2941 "git ls-files --unmerged --stage" thing is sorta useful too,
2942 actually. And note gitk --merge.
2944 Add more good examples. Entire sections of just cookbook examples
2945 might be a good idea; maybe make an "advanced examples" section a
2946 standard end-of-chapter section?
2948 Include cross-references to the glossary, where appropriate.
2950 Document shallow clones? See draft 1.5.0 release notes for some
2953 Add a section on working with other version control systems, including
2954 CVS, Subversion, and just imports of series of release tarballs.
2956 More details on gitweb?
2958 Write a chapter on using plumbing and writing scripts.