user-manual: move howto/make-dist.txt into user manual
[git/git.git] / Documentation / user-manual.txt
CommitLineData
71f4b183
CW
1Git User's Manual (for version 1.5.1 or newer)
2______________________________________________
d19fbc3c
BF
3
4This manual is designed to be readable by someone with basic unix
79c96c57 5command-line skills, but no previous knowledge of git.
d19fbc3c 6
2624d9a5
BF
7<<repositories-and-branches>> and <<exploring-git-history>> explain how
8to fetch and study a project using git--read these chapters to learn how
9to build and test a particular version of a software project, search for
10regressions, and so on.
ef89f701 11
2624d9a5
BF
12People needing to do actual development will also want to read
13<<Developing-with-git>> and <<sharing-development>>.
6bd9b682
BF
14
15Further chapters cover more specialized topics.
16
d19fbc3c
BF
17Comprehensive reference documentation is available through the man
18pages. For a command such as "git clone", just use
19
20------------------------------------------------
21$ man git-clone
22------------------------------------------------
23
2624d9a5
BF
24See also <<git-quick-start>> for a brief overview of git commands,
25without any explanation.
ef89f701 26
2624d9a5
BF
27Also, see <<todo>> for ways that you can help make this manual more
28complete.
ef89f701 29
b181d57f 30
e34caace 31[[repositories-and-branches]]
d19fbc3c
BF
32Repositories and Branches
33=========================
34
e34caace 35[[how-to-get-a-git-repository]]
d19fbc3c
BF
36How to get a git repository
37---------------------------
38
39It will be useful to have a git repository to experiment with as you
40read this manual.
41
42The best way to get one is by using the gitlink:git-clone[1] command
43to download a copy of an existing repository for a project that you
44are interested in. If you don't already have a project in mind, here
45are some interesting examples:
46
47------------------------------------------------
48 # git itself (approx. 10MB download):
49$ git clone git://git.kernel.org/pub/scm/git/git.git
50 # the linux kernel (approx. 150MB download):
51$ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
52------------------------------------------------
53
54The initial clone may be time-consuming for a large project, but you
55will only need to clone once.
56
57The clone command creates a new directory named after the project
58("git" or "linux-2.6" in the examples above). After you cd into this
59directory, you will see that it contains a copy of the project files,
60together with a special top-level directory named ".git", which
61contains all the information about the history of the project.
62
d5cd5de4 63In most of the following, examples will be taken from one of the two
d19fbc3c
BF
64repositories above.
65
e34caace 66[[how-to-check-out]]
d19fbc3c
BF
67How to check out a different version of a project
68-------------------------------------------------
69
70Git is best thought of as a tool for storing the history of a
71collection of files. It stores the history as a compressed
72collection of interrelated snapshots (versions) of the project's
73contents.
74
81b6c950
BF
75A single git repository may contain multiple branches. It keeps track
76of them by keeping a list of <<def_head,heads>> which reference the
77latest version on each branch; the gitlink:git-branch[1] command shows
78you the list of branch heads:
d19fbc3c
BF
79
80------------------------------------------------
81$ git branch
82* master
83------------------------------------------------
84
4f752407
BF
85A freshly cloned repository contains a single branch head, by default
86named "master", with the working directory initialized to the state of
87the project referred to by that branch head.
d19fbc3c 88
81b6c950
BF
89Most projects also use <<def_tag,tags>>. Tags, like heads, are
90references into the project's history, and can be listed using the
d19fbc3c
BF
91gitlink:git-tag[1] command:
92
93------------------------------------------------
94$ git tag -l
95v2.6.11
96v2.6.11-tree
97v2.6.12
98v2.6.12-rc2
99v2.6.12-rc3
100v2.6.12-rc4
101v2.6.12-rc5
102v2.6.12-rc6
103v2.6.13
104...
105------------------------------------------------
106
fe4b3e59 107Tags are expected to always point at the same version of a project,
81b6c950 108while heads are expected to advance as development progresses.
fe4b3e59 109
81b6c950 110Create a new branch head pointing to one of these versions and check it
d19fbc3c
BF
111out using gitlink:git-checkout[1]:
112
113------------------------------------------------
114$ git checkout -b new v2.6.13
115------------------------------------------------
116
117The working directory then reflects the contents that the project had
118when it was tagged v2.6.13, and gitlink:git-branch[1] shows two
119branches, with an asterisk marking the currently checked-out branch:
120
121------------------------------------------------
122$ git branch
123 master
124* new
125------------------------------------------------
126
127If you decide that you'd rather see version 2.6.17, you can modify
128the current branch to point at v2.6.17 instead, with
129
130------------------------------------------------
131$ git reset --hard v2.6.17
132------------------------------------------------
133
81b6c950 134Note that if the current branch head was your only reference to a
d19fbc3c 135particular point in history, then resetting that branch may leave you
81b6c950
BF
136with no way to find the history it used to point to; so use this command
137carefully.
d19fbc3c 138
e34caace 139[[understanding-commits]]
d19fbc3c
BF
140Understanding History: Commits
141------------------------------
142
143Every change in the history of a project is represented by a commit.
144The gitlink:git-show[1] command shows the most recent commit on the
145current branch:
146
147------------------------------------------------
148$ git show
149commit 2b5f6dcce5bf94b9b119e9ed8d537098ec61c3d2
150Author: Jamal Hadi Salim <hadi@cyberus.ca>
151Date: Sat Dec 2 22:22:25 2006 -0800
152
153 [XFRM]: Fix aevent structuring to be more complete.
154
155 aevents can not uniquely identify an SA. We break the ABI with this
156 patch, but consensus is that since it is not yet utilized by any
157 (known) application then it is fine (better do it now than later).
158
159 Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
160 Signed-off-by: David S. Miller <davem@davemloft.net>
161
162diff --git a/Documentation/networking/xfrm_sync.txt b/Documentation/networking/xfrm_sync.txt
163index 8be626f..d7aac9d 100644
164--- a/Documentation/networking/xfrm_sync.txt
165+++ b/Documentation/networking/xfrm_sync.txt
166@@ -47,10 +47,13 @@ aevent_id structure looks like:
167
168 struct xfrm_aevent_id {
169 struct xfrm_usersa_id sa_id;
170+ xfrm_address_t saddr;
171 __u32 flags;
172+ __u32 reqid;
173 };
174...
175------------------------------------------------
176
177As you can see, a commit shows who made the latest change, what they
178did, and why.
179
35121930
BF
180Every commit has a 40-hexdigit id, sometimes called the "object name" or the
181"SHA1 id", shown on the first line of the "git show" output. You can usually
182refer to a commit by a shorter name, such as a tag or a branch name, but this
183longer name can also be useful. Most importantly, it is a globally unique
184name for this commit: so if you tell somebody else the object name (for
185example in email), then you are guaranteed that name will refer to the same
186commit in their repository that it does in yours (assuming their repository
187has that commit at all). Since the object name is computed as a hash over the
188contents of the commit, you are guaranteed that the commit can never change
189without its name also changing.
190
191In fact, in <<git-internals>> we shall see that everything stored in git
192history, including file data and directory contents, is stored in an object
193with a name that is a hash of its contents.
d19fbc3c 194
e34caace 195[[understanding-reachability]]
d19fbc3c
BF
196Understanding history: commits, parents, and reachability
197~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
198
199Every commit (except the very first commit in a project) also has a
200parent commit which shows what happened before this commit.
201Following the chain of parents will eventually take you back to the
202beginning of the project.
203
204However, the commits do not form a simple list; git allows lines of
205development to diverge and then reconverge, and the point where two
206lines of development reconverge is called a "merge". The commit
207representing a merge can therefore have more than one parent, with
208each parent representing the most recent commit on one of the lines
209of development leading to that point.
210
211The best way to see how this works is using the gitlink:gitk[1]
212command; running gitk now on a git repository and looking for merge
213commits will help understand how the git organizes history.
214
215In the following, we say that commit X is "reachable" from commit Y
216if commit X is an ancestor of commit Y. Equivalently, you could say
217that Y is a descendent of X, or that there is a chain of parents
218leading from commit Y to commit X.
219
e34caace 220[[history-diagrams]]
3dff5379
PR
221Understanding history: History diagrams
222~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c
BF
223
224We will sometimes represent git history using diagrams like the one
225below. Commits are shown as "o", and the links between them with
226lines drawn with - / and \. Time goes left to right:
227
1dc71a91
BF
228
229................................................
d19fbc3c
BF
230 o--o--o <-- Branch A
231 /
232 o--o--o <-- master
233 \
234 o--o--o <-- Branch B
1dc71a91 235................................................
d19fbc3c
BF
236
237If we need to talk about a particular commit, the character "o" may
238be replaced with another letter or number.
239
e34caace 240[[what-is-a-branch]]
d19fbc3c
BF
241Understanding history: What is a branch?
242~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
243
81b6c950
BF
244When we need to be precise, we will use the word "branch" to mean a line
245of development, and "branch head" (or just "head") to mean a reference
246to the most recent commit on a branch. In the example above, the branch
247head named "A" is a pointer to one particular commit, but we refer to
248the line of three commits leading up to that point as all being part of
d19fbc3c
BF
249"branch A".
250
81b6c950
BF
251However, when no confusion will result, we often just use the term
252"branch" both for branches and for branch heads.
d19fbc3c 253
e34caace 254[[manipulating-branches]]
d19fbc3c
BF
255Manipulating branches
256---------------------
257
258Creating, deleting, and modifying branches is quick and easy; here's
259a summary of the commands:
260
261git branch::
262 list all branches
263git branch <branch>::
264 create a new branch named <branch>, referencing the same
265 point in history as the current branch
266git branch <branch> <start-point>::
267 create a new branch named <branch>, referencing
268 <start-point>, which may be specified any way you like,
269 including using a branch name or a tag name
270git branch -d <branch>::
271 delete the branch <branch>; if the branch you are deleting
c64415e2
BF
272 points to a commit which is not reachable from the current
273 branch, this command will fail with a warning.
d19fbc3c
BF
274git branch -D <branch>::
275 even if the branch points to a commit not reachable
276 from the current branch, you may know that that commit
277 is still reachable from some other branch or tag. In that
278 case it is safe to use this command to force git to delete
279 the branch.
280git checkout <branch>::
281 make the current branch <branch>, updating the working
282 directory to reflect the version referenced by <branch>
283git checkout -b <new> <start-point>::
284 create a new branch <new> referencing <start-point>, and
285 check it out.
286
72a76c95
BF
287The special symbol "HEAD" can always be used to refer to the current
288branch. In fact, git uses a file named "HEAD" in the .git directory to
289remember which branch is current:
290
291------------------------------------------------
292$ cat .git/HEAD
293ref: refs/heads/master
294------------------------------------------------
295
25d9f3fa 296[[detached-head]]
72a76c95
BF
297Examining an old version without creating a new branch
298------------------------------------------------------
299
300The git-checkout command normally expects a branch head, but will also
301accept an arbitrary commit; for example, you can check out the commit
302referenced by a tag:
303
304------------------------------------------------
305$ git checkout v2.6.17
306Note: moving to "v2.6.17" which isn't a local branch
307If you want to create a new branch from this checkout, you may do so
308(now or later) by using -b with the checkout command again. Example:
309 git checkout -b <new_branch_name>
310HEAD is now at 427abfa... Linux v2.6.17
311------------------------------------------------
312
313The HEAD then refers to the SHA1 of the commit instead of to a branch,
314and git branch shows that you are no longer on a branch:
315
316------------------------------------------------
317$ cat .git/HEAD
318427abfa28afedffadfca9dd8b067eb6d36bac53f
953f3d6f 319$ git branch
72a76c95
BF
320* (no branch)
321 master
322------------------------------------------------
323
324In this case we say that the HEAD is "detached".
325
953f3d6f
BF
326This is an easy way to check out a particular version without having to
327make up a name for the new branch. You can still create a new branch
328(or tag) for this version later if you decide to.
d19fbc3c 329
e34caace 330[[examining-remote-branches]]
d19fbc3c
BF
331Examining branches from a remote repository
332-------------------------------------------
333
334The "master" branch that was created at the time you cloned is a copy
335of the HEAD in the repository that you cloned from. That repository
336may also have had other branches, though, and your local repository
337keeps branches which track each of those remote branches, which you
338can view using the "-r" option to gitlink:git-branch[1]:
339
340------------------------------------------------
341$ git branch -r
342 origin/HEAD
343 origin/html
344 origin/maint
345 origin/man
346 origin/master
347 origin/next
348 origin/pu
349 origin/todo
350------------------------------------------------
351
352You cannot check out these remote-tracking branches, but you can
353examine them on a branch of your own, just as you would a tag:
354
355------------------------------------------------
356$ git checkout -b my-todo-copy origin/todo
357------------------------------------------------
358
359Note that the name "origin" is just the name that git uses by default
360to refer to the repository that you cloned from.
361
362[[how-git-stores-references]]
f60b9642
BF
363Naming branches, tags, and other references
364-------------------------------------------
d19fbc3c
BF
365
366Branches, remote-tracking branches, and tags are all references to
f60b9642
BF
367commits. All references are named with a slash-separated path name
368starting with "refs"; the names we've been using so far are actually
369shorthand:
d19fbc3c 370
f60b9642
BF
371 - The branch "test" is short for "refs/heads/test".
372 - The tag "v2.6.18" is short for "refs/tags/v2.6.18".
373 - "origin/master" is short for "refs/remotes/origin/master".
d19fbc3c 374
f60b9642
BF
375The full name is occasionally useful if, for example, there ever
376exists a tag and a branch with the same name.
d19fbc3c 377
c64415e2
BF
378As another useful shortcut, the "HEAD" of a repository can be referred
379to just using the name of that repository. So, for example, "origin"
380is usually a shortcut for the HEAD branch in the repository "origin".
d19fbc3c
BF
381
382For the complete list of paths which git checks for references, and
f60b9642
BF
383the order it uses to decide which to choose when there are multiple
384references with the same shorthand name, see the "SPECIFYING
385REVISIONS" section of gitlink:git-rev-parse[1].
d19fbc3c
BF
386
387[[Updating-a-repository-with-git-fetch]]
388Updating a repository with git fetch
389------------------------------------
390
391Eventually the developer cloned from will do additional work in her
392repository, creating new commits and advancing the branches to point
393at the new commits.
394
395The command "git fetch", with no arguments, will update all of the
396remote-tracking branches to the latest version found in her
397repository. It will not touch any of your own branches--not even the
398"master" branch that was created for you on clone.
399
e34caace 400[[fetching-branches]]
d5cd5de4
BF
401Fetching branches from other repositories
402-----------------------------------------
403
404You can also track branches from repositories other than the one you
405cloned from, using gitlink:git-remote[1]:
406
407-------------------------------------------------
408$ git remote add linux-nfs git://linux-nfs.org/pub/nfs-2.6.git
04483524 409$ git fetch linux-nfs
d5cd5de4
BF
410* refs/remotes/linux-nfs/master: storing branch 'master' ...
411 commit: bf81b46
412-------------------------------------------------
413
414New remote-tracking branches will be stored under the shorthand name
415that you gave "git remote add", in this case linux-nfs:
416
417-------------------------------------------------
418$ git branch -r
419linux-nfs/master
420origin/master
421-------------------------------------------------
422
423If you run "git fetch <remote>" later, the tracking branches for the
424named <remote> will be updated.
425
426If you examine the file .git/config, you will see that git has added
427a new stanza:
428
429-------------------------------------------------
430$ cat .git/config
431...
432[remote "linux-nfs"]
923642fe
BF
433 url = git://linux-nfs.org/pub/nfs-2.6.git
434 fetch = +refs/heads/*:refs/remotes/linux-nfs/*
d5cd5de4
BF
435...
436-------------------------------------------------
437
fc90c536
BF
438This is what causes git to track the remote's branches; you may modify
439or delete these configuration options by editing .git/config with a
440text editor. (See the "CONFIGURATION FILE" section of
441gitlink:git-config[1] for details.)
d5cd5de4 442
e34caace 443[[exploring-git-history]]
d19fbc3c
BF
444Exploring git history
445=====================
446
447Git is best thought of as a tool for storing the history of a
448collection of files. It does this by storing compressed snapshots of
449the contents of a file heirarchy, together with "commits" which show
450the relationships between these snapshots.
451
452Git provides extremely flexible and fast tools for exploring the
453history of a project.
454
aacd404e 455We start with one specialized tool that is useful for finding the
d19fbc3c
BF
456commit that introduced a bug into a project.
457
e34caace 458[[using-bisect]]
d19fbc3c
BF
459How to use bisect to find a regression
460--------------------------------------
461
462Suppose version 2.6.18 of your project worked, but the version at
463"master" crashes. Sometimes the best way to find the cause of such a
464regression is to perform a brute-force search through the project's
465history to find the particular commit that caused the problem. The
466gitlink:git-bisect[1] command can help you do this:
467
468-------------------------------------------------
469$ git bisect start
470$ git bisect good v2.6.18
471$ git bisect bad master
472Bisecting: 3537 revisions left to test after this
473[65934a9a028b88e83e2b0f8b36618fe503349f8e] BLOCK: Make USB storage depend on SCSI rather than selecting it [try #6]
474-------------------------------------------------
475
476If you run "git branch" at this point, you'll see that git has
477temporarily moved you to a new branch named "bisect". This branch
478points to a commit (with commit id 65934...) that is reachable from
479v2.6.19 but not from v2.6.18. Compile and test it, and see whether
480it crashes. Assume it does crash. Then:
481
482-------------------------------------------------
483$ git bisect bad
484Bisecting: 1769 revisions left to test after this
485[7eff82c8b1511017ae605f0c99ac275a7e21b867] i2c-core: Drop useless bitmaskings
486-------------------------------------------------
487
488checks out an older version. Continue like this, telling git at each
489stage whether the version it gives you is good or bad, and notice
490that the number of revisions left to test is cut approximately in
491half each time.
492
493After about 13 tests (in this case), it will output the commit id of
494the guilty commit. You can then examine the commit with
495gitlink:git-show[1], find out who wrote it, and mail them your bug
496report with the commit id. Finally, run
497
498-------------------------------------------------
499$ git bisect reset
500-------------------------------------------------
501
502to return you to the branch you were on before and delete the
503temporary "bisect" branch.
504
505Note that the version which git-bisect checks out for you at each
506point is just a suggestion, and you're free to try a different
507version if you think it would be a good idea. For example,
508occasionally you may land on a commit that broke something unrelated;
509run
510
511-------------------------------------------------
04483524 512$ git bisect visualize
d19fbc3c
BF
513-------------------------------------------------
514
515which will run gitk and label the commit it chose with a marker that
516says "bisect". Chose a safe-looking commit nearby, note its commit
517id, and check it out with:
518
519-------------------------------------------------
520$ git reset --hard fb47ddb2db...
521-------------------------------------------------
522
523then test, run "bisect good" or "bisect bad" as appropriate, and
524continue.
525
e34caace 526[[naming-commits]]
d19fbc3c
BF
527Naming commits
528--------------
529
530We have seen several ways of naming commits already:
531
d55ae921 532 - 40-hexdigit object name
d19fbc3c
BF
533 - branch name: refers to the commit at the head of the given
534 branch
535 - tag name: refers to the commit pointed to by the given tag
536 (we've seen branches and tags are special cases of
537 <<how-git-stores-references,references>>).
538 - HEAD: refers to the head of the current branch
539
eb6ae7f4 540There are many more; see the "SPECIFYING REVISIONS" section of the
aec053bb 541gitlink:git-rev-parse[1] man page for the complete list of ways to
d19fbc3c
BF
542name revisions. Some examples:
543
544-------------------------------------------------
d55ae921 545$ git show fb47ddb2 # the first few characters of the object name
d19fbc3c
BF
546 # are usually enough to specify it uniquely
547$ git show HEAD^ # the parent of the HEAD commit
548$ git show HEAD^^ # the grandparent
549$ git show HEAD~4 # the great-great-grandparent
550-------------------------------------------------
551
552Recall that merge commits may have more than one parent; by default,
553^ and ~ follow the first parent listed in the commit, but you can
554also choose:
555
556-------------------------------------------------
557$ git show HEAD^1 # show the first parent of HEAD
558$ git show HEAD^2 # show the second parent of HEAD
559-------------------------------------------------
560
561In addition to HEAD, there are several other special names for
562commits:
563
564Merges (to be discussed later), as well as operations such as
565git-reset, which change the currently checked-out commit, generally
566set ORIG_HEAD to the value HEAD had before the current operation.
567
568The git-fetch operation always stores the head of the last fetched
569branch in FETCH_HEAD. For example, if you run git fetch without
570specifying a local branch as the target of the operation
571
572-------------------------------------------------
573$ git fetch git://example.com/proj.git theirbranch
574-------------------------------------------------
575
576the fetched commits will still be available from FETCH_HEAD.
577
578When we discuss merges we'll also see the special name MERGE_HEAD,
579which refers to the other branch that we're merging in to the current
580branch.
581
aec053bb 582The gitlink:git-rev-parse[1] command is a low-level command that is
d55ae921
BF
583occasionally useful for translating some name for a commit to the object
584name for that commit:
aec053bb
BF
585
586-------------------------------------------------
587$ git rev-parse origin
588e05db0fd4f31dde7005f075a84f96b360d05984b
589-------------------------------------------------
590
e34caace 591[[creating-tags]]
d19fbc3c
BF
592Creating tags
593-------------
594
595We can also create a tag to refer to a particular commit; after
596running
597
598-------------------------------------------------
04483524 599$ git tag stable-1 1b2e1d63ff
d19fbc3c
BF
600-------------------------------------------------
601
602You can use stable-1 to refer to the commit 1b2e1d63ff.
603
c64415e2
BF
604This creates a "lightweight" tag. If you would also like to include a
605comment with the tag, and possibly sign it cryptographically, then you
606should create a tag object instead; see the gitlink:git-tag[1] man page
607for details.
d19fbc3c 608
e34caace 609[[browsing-revisions]]
d19fbc3c
BF
610Browsing revisions
611------------------
612
613The gitlink:git-log[1] command can show lists of commits. On its
614own, it shows all commits reachable from the parent commit; but you
615can also make more specific requests:
616
617-------------------------------------------------
618$ git log v2.5.. # commits since (not reachable from) v2.5
619$ git log test..master # commits reachable from master but not test
620$ git log master..test # ...reachable from test but not master
621$ git log master...test # ...reachable from either test or master,
622 # but not both
623$ git log --since="2 weeks ago" # commits from the last 2 weeks
624$ git log Makefile # commits which modify Makefile
625$ git log fs/ # ... which modify any file under fs/
626$ git log -S'foo()' # commits which add or remove any file data
627 # matching the string 'foo()'
628-------------------------------------------------
629
630And of course you can combine all of these; the following finds
631commits since v2.5 which touch the Makefile or any file under fs:
632
633-------------------------------------------------
634$ git log v2.5.. Makefile fs/
635-------------------------------------------------
636
637You can also ask git log to show patches:
638
639-------------------------------------------------
640$ git log -p
641-------------------------------------------------
642
643See the "--pretty" option in the gitlink:git-log[1] man page for more
644display options.
645
646Note that git log starts with the most recent commit and works
647backwards through the parents; however, since git history can contain
3dff5379 648multiple independent lines of development, the particular order that
d19fbc3c
BF
649commits are listed in may be somewhat arbitrary.
650
e34caace 651[[generating-diffs]]
d19fbc3c
BF
652Generating diffs
653----------------
654
655You can generate diffs between any two versions using
656gitlink:git-diff[1]:
657
658-------------------------------------------------
659$ git diff master..test
660-------------------------------------------------
661
662Sometimes what you want instead is a set of patches:
663
664-------------------------------------------------
665$ git format-patch master..test
666-------------------------------------------------
667
668will generate a file with a patch for each commit reachable from test
669but not from master. Note that if master also has commits which are
670not reachable from test, then the combined result of these patches
671will not be the same as the diff produced by the git-diff example.
672
e34caace 673[[viewing-old-file-versions]]
d19fbc3c
BF
674Viewing old file versions
675-------------------------
676
677You can always view an old version of a file by just checking out the
678correct revision first. But sometimes it is more convenient to be
679able to view an old version of a single file without checking
680anything out; this command does that:
681
682-------------------------------------------------
683$ git show v2.5:fs/locks.c
684-------------------------------------------------
685
686Before the colon may be anything that names a commit, and after it
687may be any path to a file tracked by git.
688
e34caace 689[[history-examples]]
aec053bb
BF
690Examples
691--------
692
e34caace 693[[checking-for-equal-branches]]
aec053bb 694Check whether two branches point at the same history
2f99710c 695~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
aec053bb
BF
696
697Suppose you want to check whether two branches point at the same point
698in history.
699
700-------------------------------------------------
701$ git diff origin..master
702-------------------------------------------------
703
69f7ad73
BF
704will tell you whether the contents of the project are the same at the
705two branches; in theory, however, it's possible that the same project
706contents could have been arrived at by two different historical
d55ae921 707routes. You could compare the object names:
aec053bb
BF
708
709-------------------------------------------------
710$ git rev-list origin
711e05db0fd4f31dde7005f075a84f96b360d05984b
712$ git rev-list master
713e05db0fd4f31dde7005f075a84f96b360d05984b
714-------------------------------------------------
715
69f7ad73
BF
716Or you could recall that the ... operator selects all commits
717contained reachable from either one reference or the other but not
718both: so
aec053bb
BF
719
720-------------------------------------------------
721$ git log origin...master
722-------------------------------------------------
723
724will return no commits when the two branches are equal.
725
e34caace 726[[finding-tagged-descendants]]
b181d57f
BF
727Find first tagged version including a given fix
728~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
aec053bb 729
69f7ad73
BF
730Suppose you know that the commit e05db0fd fixed a certain problem.
731You'd like to find the earliest tagged release that contains that
732fix.
733
734Of course, there may be more than one answer--if the history branched
735after commit e05db0fd, then there could be multiple "earliest" tagged
736releases.
737
738You could just visually inspect the commits since e05db0fd:
739
740-------------------------------------------------
741$ gitk e05db0fd..
742-------------------------------------------------
743
b181d57f
BF
744Or you can use gitlink:git-name-rev[1], which will give the commit a
745name based on any tag it finds pointing to one of the commit's
746descendants:
747
748-------------------------------------------------
04483524 749$ git name-rev --tags e05db0fd
b181d57f
BF
750e05db0fd tags/v1.5.0-rc1^0~23
751-------------------------------------------------
752
753The gitlink:git-describe[1] command does the opposite, naming the
754revision using a tag on which the given commit is based:
755
756-------------------------------------------------
757$ git describe e05db0fd
04483524 758v1.5.0-rc0-260-ge05db0f
b181d57f
BF
759-------------------------------------------------
760
761but that may sometimes help you guess which tags might come after the
762given commit.
763
764If you just want to verify whether a given tagged version contains a
765given commit, you could use gitlink:git-merge-base[1]:
766
767-------------------------------------------------
768$ git merge-base e05db0fd v1.5.0-rc1
769e05db0fd4f31dde7005f075a84f96b360d05984b
770-------------------------------------------------
771
772The merge-base command finds a common ancestor of the given commits,
773and always returns one or the other in the case where one is a
774descendant of the other; so the above output shows that e05db0fd
775actually is an ancestor of v1.5.0-rc1.
776
777Alternatively, note that
778
779-------------------------------------------------
4a7979ca 780$ git log v1.5.0-rc1..e05db0fd
b181d57f
BF
781-------------------------------------------------
782
4a7979ca 783will produce empty output if and only if v1.5.0-rc1 includes e05db0fd,
b181d57f 784because it outputs only commits that are not reachable from v1.5.0-rc1.
aec053bb 785
4a7979ca
BF
786As yet another alternative, the gitlink:git-show-branch[1] command lists
787the commits reachable from its arguments with a display on the left-hand
788side that indicates which arguments that commit is reachable from. So,
789you can run something like
790
791-------------------------------------------------
792$ git show-branch e05db0fd v1.5.0-rc0 v1.5.0-rc1 v1.5.0-rc2
793! [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
794available
795 ! [v1.5.0-rc0] GIT v1.5.0 preview
796 ! [v1.5.0-rc1] GIT v1.5.0-rc1
797 ! [v1.5.0-rc2] GIT v1.5.0-rc2
798...
799-------------------------------------------------
800
801then search for a line that looks like
802
803-------------------------------------------------
804+ ++ [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
805available
806-------------------------------------------------
807
808Which shows that e05db0fd is reachable from itself, from v1.5.0-rc1, and
809from v1.5.0-rc2, but not from v1.5.0-rc0.
810
82c8bf28
BF
811[[making-a-release]]
812Creating a changelog and tarball for a software release
813~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
814
815The gitlink:git-archive[1] command can create a tar or zip archive from
816any version of a project; for example:
817
818-------------------------------------------------
819$ git archive --format=tar --prefix=project/ HEAD | gzip >latest.tar.gz
820-------------------------------------------------
821
822will use HEAD to produce a tar archive in which each filename is
823preceded by "prefix/".
824
825If you're releasing a new version of a software project, you may want
826to simultaneously make a changelog to include in the release
827announcement.
828
829Linus Torvalds, for example, makes new kernel releases by tagging them,
830then running:
831
832-------------------------------------------------
833$ release-script 2.6.12 2.6.13-rc6 2.6.13-rc7
834-------------------------------------------------
835
836where release-script is a shell script that looks like:
837
838-------------------------------------------------
839#!/bin/sh
840stable="$1"
841last="$2"
842new="$3"
843echo "# git tag v$new"
844echo "git archive --prefix=linux-$new/ v$new | gzip -9 > ../linux-$new.tar.gz"
845echo "git diff v$stable v$new | gzip -9 > ../patch-$new.gz"
846echo "git log --no-merges v$new ^v$last > ../ChangeLog-$new"
847echo "git shortlog --no-merges v$new ^v$last > ../ShortLog"
848echo "git diff --stat --summary -M v$last v$new > ../diffstat-$new"
849-------------------------------------------------
850
851and then he just cut-and-pastes the output commands after verifying that
852they look OK.
4a7979ca 853
e34caace 854[[Developing-with-git]]
d19fbc3c
BF
855Developing with git
856===================
857
e34caace 858[[telling-git-your-name]]
d19fbc3c
BF
859Telling git your name
860---------------------
861
862Before creating any commits, you should introduce yourself to git. The
58c19d1f
BF
863easiest way to do so is to make sure the following lines appear in a
864file named .gitconfig in your home directory:
d19fbc3c
BF
865
866------------------------------------------------
d19fbc3c
BF
867[user]
868 name = Your Name Comes Here
869 email = you@yourdomain.example.com
d19fbc3c
BF
870------------------------------------------------
871
fc90c536
BF
872(See the "CONFIGURATION FILE" section of gitlink:git-config[1] for
873details on the configuration file.)
874
d19fbc3c 875
e34caace 876[[creating-a-new-repository]]
d19fbc3c
BF
877Creating a new repository
878-------------------------
879
880Creating a new repository from scratch is very easy:
881
882-------------------------------------------------
883$ mkdir project
884$ cd project
f1d2b477 885$ git init
d19fbc3c
BF
886-------------------------------------------------
887
888If you have some initial content (say, a tarball):
889
890-------------------------------------------------
891$ tar -xzvf project.tar.gz
892$ cd project
f1d2b477 893$ git init
d19fbc3c
BF
894$ git add . # include everything below ./ in the first commit:
895$ git commit
896-------------------------------------------------
897
898[[how-to-make-a-commit]]
ae25c67a 899How to make a commit
d19fbc3c
BF
900--------------------
901
902Creating a new commit takes three steps:
903
904 1. Making some changes to the working directory using your
905 favorite editor.
906 2. Telling git about your changes.
907 3. Creating the commit using the content you told git about
908 in step 2.
909
910In practice, you can interleave and repeat steps 1 and 2 as many
911times as you want: in order to keep track of what you want committed
912at step 3, git maintains a snapshot of the tree's contents in a
913special staging area called "the index."
914
01997b4a
BF
915At the beginning, the content of the index will be identical to
916that of the HEAD. The command "git diff --cached", which shows
917the difference between the HEAD and the index, should therefore
918produce no output at that point.
eb6ae7f4 919
d19fbc3c
BF
920Modifying the index is easy:
921
922To update the index with the new contents of a modified file, use
923
924-------------------------------------------------
925$ git add path/to/file
926-------------------------------------------------
927
928To add the contents of a new file to the index, use
929
930-------------------------------------------------
931$ git add path/to/file
932-------------------------------------------------
933
eb6ae7f4 934To remove a file from the index and from the working tree,
d19fbc3c
BF
935
936-------------------------------------------------
937$ git rm path/to/file
938-------------------------------------------------
939
940After each step you can verify that
941
942-------------------------------------------------
943$ git diff --cached
944-------------------------------------------------
945
946always shows the difference between the HEAD and the index file--this
947is what you'd commit if you created the commit now--and that
948
949-------------------------------------------------
950$ git diff
951-------------------------------------------------
952
953shows the difference between the working tree and the index file.
954
955Note that "git add" always adds just the current contents of a file
956to the index; further changes to the same file will be ignored unless
957you run git-add on the file again.
958
959When you're ready, just run
960
961-------------------------------------------------
962$ git commit
963-------------------------------------------------
964
965and git will prompt you for a commit message and then create the new
3dff5379 966commit. Check to make sure it looks like what you expected with
d19fbc3c
BF
967
968-------------------------------------------------
969$ git show
970-------------------------------------------------
971
972As a special shortcut,
973
974-------------------------------------------------
975$ git commit -a
976-------------------------------------------------
977
978will update the index with any files that you've modified or removed
979and create a commit, all in one step.
980
981A number of commands are useful for keeping track of what you're
982about to commit:
983
984-------------------------------------------------
985$ git diff --cached # difference between HEAD and the index; what
986 # would be commited if you ran "commit" now.
987$ git diff # difference between the index file and your
988 # working directory; changes that would not
989 # be included if you ran "commit" now.
c64415e2
BF
990$ git diff HEAD # difference between HEAD and working tree; what
991 # would be committed if you ran "commit -a" now.
d19fbc3c
BF
992$ git status # a brief per-file summary of the above.
993-------------------------------------------------
994
e34caace 995[[creating-good-commit-messages]]
ae25c67a 996Creating good commit messages
d19fbc3c
BF
997-----------------------------
998
999Though not required, it's a good idea to begin the commit message
1000with a single short (less than 50 character) line summarizing the
1001change, followed by a blank line and then a more thorough
1002description. Tools that turn commits into email, for example, use
1003the first line on the Subject line and the rest of the commit in the
1004body.
1005
e34caace 1006[[how-to-merge]]
ae25c67a 1007How to merge
d19fbc3c
BF
1008------------
1009
1010You can rejoin two diverging branches of development using
1011gitlink:git-merge[1]:
1012
1013-------------------------------------------------
1014$ git merge branchname
1015-------------------------------------------------
1016
1017merges the development in the branch "branchname" into the current
1018branch. If there are conflicts--for example, if the same file is
1019modified in two different ways in the remote branch and the local
1020branch--then you are warned; the output may look something like this:
1021
1022-------------------------------------------------
fabbd8f6
BF
1023$ git merge next
1024 100% (4/4) done
1025Auto-merged file.txt
d19fbc3c
BF
1026CONFLICT (content): Merge conflict in file.txt
1027Automatic merge failed; fix conflicts and then commit the result.
1028-------------------------------------------------
1029
1030Conflict markers are left in the problematic files, and after
1031you resolve the conflicts manually, you can update the index
1032with the contents and run git commit, as you normally would when
1033creating a new file.
1034
1035If you examine the resulting commit using gitk, you will see that it
1036has two parents, one pointing to the top of the current branch, and
1037one to the top of the other branch.
1038
d19fbc3c
BF
1039[[resolving-a-merge]]
1040Resolving a merge
1041-----------------
1042
1043When a merge isn't resolved automatically, git leaves the index and
1044the working tree in a special state that gives you all the
1045information you need to help resolve the merge.
1046
1047Files with conflicts are marked specially in the index, so until you
ef561ac7
BF
1048resolve the problem and update the index, gitlink:git-commit[1] will
1049fail:
d19fbc3c
BF
1050
1051-------------------------------------------------
1052$ git commit
1053file.txt: needs merge
1054-------------------------------------------------
1055
ef561ac7
BF
1056Also, gitlink:git-status[1] will list those files as "unmerged", and the
1057files with conflicts will have conflict markers added, like this:
1058
1059-------------------------------------------------
1060<<<<<<< HEAD:file.txt
1061Hello world
1062=======
1063Goodbye
1064>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1065-------------------------------------------------
1066
1067All you need to do is edit the files to resolve the conflicts, and then
1068
1069-------------------------------------------------
1070$ git add file.txt
1071$ git commit
1072-------------------------------------------------
1073
1074Note that the commit message will already be filled in for you with
1075some information about the merge. Normally you can just use this
1076default message unchanged, but you may add additional commentary of
1077your own if desired.
1078
1079The above is all you need to know to resolve a simple merge. But git
1080also provides more information to help resolve conflicts:
1081
e34caace 1082[[conflict-resolution]]
ef561ac7
BF
1083Getting conflict-resolution help during a merge
1084~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c
BF
1085
1086All of the changes that git was able to merge automatically are
1087already added to the index file, so gitlink:git-diff[1] shows only
ef561ac7 1088the conflicts. It uses an unusual syntax:
d19fbc3c
BF
1089
1090-------------------------------------------------
1091$ git diff
1092diff --cc file.txt
1093index 802992c,2b60207..0000000
1094--- a/file.txt
1095+++ b/file.txt
1096@@@ -1,1 -1,1 +1,5 @@@
1097++<<<<<<< HEAD:file.txt
1098 +Hello world
1099++=======
1100+ Goodbye
1101++>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1102-------------------------------------------------
1103
1104Recall that the commit which will be commited after we resolve this
1105conflict will have two parents instead of the usual one: one parent
1106will be HEAD, the tip of the current branch; the other will be the
1107tip of the other branch, which is stored temporarily in MERGE_HEAD.
1108
ef561ac7
BF
1109During the merge, the index holds three versions of each file. Each of
1110these three "file stages" represents a different version of the file:
1111
1112-------------------------------------------------
1113$ git show :1:file.txt # the file in a common ancestor of both branches
1114$ git show :2:file.txt # the version from HEAD, but including any
1115 # nonconflicting changes from MERGE_HEAD
1116$ git show :3:file.txt # the version from MERGE_HEAD, but including any
1117 # nonconflicting changes from HEAD.
1118-------------------------------------------------
1119
1120Since the stage 2 and stage 3 versions have already been updated with
1121nonconflicting changes, the only remaining differences between them are
1122the important ones; thus gitlink:git-diff[1] can use the information in
1123the index to show only those conflicts.
1124
1125The diff above shows the differences between the working-tree version of
1126file.txt and the stage 2 and stage 3 versions. So instead of preceding
1127each line by a single "+" or "-", it now uses two columns: the first
1128column is used for differences between the first parent and the working
1129directory copy, and the second for differences between the second parent
1130and the working directory copy. (See the "COMBINED DIFF FORMAT" section
1131of gitlink:git-diff-files[1] for a details of the format.)
1132
1133After resolving the conflict in the obvious way (but before updating the
1134index), the diff will look like:
d19fbc3c
BF
1135
1136-------------------------------------------------
1137$ git diff
1138diff --cc file.txt
1139index 802992c,2b60207..0000000
1140--- a/file.txt
1141+++ b/file.txt
1142@@@ -1,1 -1,1 +1,1 @@@
1143- Hello world
1144 -Goodbye
1145++Goodbye world
1146-------------------------------------------------
1147
1148This shows that our resolved version deleted "Hello world" from the
1149first parent, deleted "Goodbye" from the second parent, and added
1150"Goodbye world", which was previously absent from both.
1151
ef561ac7
BF
1152Some special diff options allow diffing the working directory against
1153any of these stages:
1154
1155-------------------------------------------------
1156$ git diff -1 file.txt # diff against stage 1
1157$ git diff --base file.txt # same as the above
1158$ git diff -2 file.txt # diff against stage 2
1159$ git diff --ours file.txt # same as the above
1160$ git diff -3 file.txt # diff against stage 3
1161$ git diff --theirs file.txt # same as the above.
1162-------------------------------------------------
1163
1164The gitlink:git-log[1] and gitk[1] commands also provide special help
1165for merges:
d19fbc3c
BF
1166
1167-------------------------------------------------
1168$ git log --merge
ef561ac7 1169$ gitk --merge
d19fbc3c
BF
1170-------------------------------------------------
1171
ef561ac7
BF
1172These will display all commits which exist only on HEAD or on
1173MERGE_HEAD, and which touch an unmerged file.
d19fbc3c 1174
61d72564 1175You may also use gitlink:git-mergetool[1], which lets you merge the
c64415e2
BF
1176unmerged files using external tools such as emacs or kdiff3.
1177
ef561ac7 1178Each time you resolve the conflicts in a file and update the index:
d19fbc3c
BF
1179
1180-------------------------------------------------
1181$ git add file.txt
d19fbc3c
BF
1182-------------------------------------------------
1183
ef561ac7
BF
1184the different stages of that file will be "collapsed", after which
1185git-diff will (by default) no longer show diffs for that file.
d19fbc3c
BF
1186
1187[[undoing-a-merge]]
ae25c67a 1188Undoing a merge
d19fbc3c
BF
1189---------------
1190
1191If you get stuck and decide to just give up and throw the whole mess
1192away, you can always return to the pre-merge state with
1193
1194-------------------------------------------------
1195$ git reset --hard HEAD
1196-------------------------------------------------
1197
1198Or, if you've already commited the merge that you want to throw away,
1199
1200-------------------------------------------------
1c73bb0e 1201$ git reset --hard ORIG_HEAD
d19fbc3c
BF
1202-------------------------------------------------
1203
1204However, this last command can be dangerous in some cases--never
1205throw away a commit you have already committed if that commit may
1206itself have been merged into another branch, as doing so may confuse
1207further merges.
1208
e34caace 1209[[fast-forwards]]
d19fbc3c
BF
1210Fast-forward merges
1211-------------------
1212
1213There is one special case not mentioned above, which is treated
1214differently. Normally, a merge results in a merge commit, with two
1215parents, one pointing at each of the two lines of development that
1216were merged.
1217
59723040
BF
1218However, if the current branch is a descendant of the other--so every
1219commit present in the one is already contained in the other--then git
1220just performs a "fast forward"; the head of the current branch is moved
1221forward to point at the head of the merged-in branch, without any new
1222commits being created.
d19fbc3c 1223
e34caace 1224[[fixing-mistakes]]
b684f830
BF
1225Fixing mistakes
1226---------------
1227
1228If you've messed up the working tree, but haven't yet committed your
1229mistake, you can return the entire working tree to the last committed
1230state with
1231
1232-------------------------------------------------
1233$ git reset --hard HEAD
1234-------------------------------------------------
1235
1236If you make a commit that you later wish you hadn't, there are two
1237fundamentally different ways to fix the problem:
1238
1239 1. You can create a new commit that undoes whatever was done
1240 by the previous commit. This is the correct thing if your
1241 mistake has already been made public.
1242
1243 2. You can go back and modify the old commit. You should
1244 never do this if you have already made the history public;
1245 git does not normally expect the "history" of a project to
1246 change, and cannot correctly perform repeated merges from
1247 a branch that has had its history changed.
1248
e34caace 1249[[reverting-a-commit]]
b684f830
BF
1250Fixing a mistake with a new commit
1251~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1252
1253Creating a new commit that reverts an earlier change is very easy;
1254just pass the gitlink:git-revert[1] command a reference to the bad
1255commit; for example, to revert the most recent commit:
1256
1257-------------------------------------------------
1258$ git revert HEAD
1259-------------------------------------------------
1260
1261This will create a new commit which undoes the change in HEAD. You
1262will be given a chance to edit the commit message for the new commit.
1263
1264You can also revert an earlier change, for example, the next-to-last:
1265
1266-------------------------------------------------
1267$ git revert HEAD^
1268-------------------------------------------------
1269
1270In this case git will attempt to undo the old change while leaving
1271intact any changes made since then. If more recent changes overlap
1272with the changes to be reverted, then you will be asked to fix
1273conflicts manually, just as in the case of <<resolving-a-merge,
1274resolving a merge>>.
1275
365aa199 1276[[fixing-a-mistake-by-editing-history]]
b684f830
BF
1277Fixing a mistake by editing history
1278~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1279
1280If the problematic commit is the most recent commit, and you have not
1281yet made that commit public, then you may just
1282<<undoing-a-merge,destroy it using git-reset>>.
1283
1284Alternatively, you
1285can edit the working directory and update the index to fix your
1286mistake, just as if you were going to <<how-to-make-a-commit,create a
1287new commit>>, then run
1288
1289-------------------------------------------------
1290$ git commit --amend
1291-------------------------------------------------
1292
1293which will replace the old commit by a new commit incorporating your
1294changes, giving you a chance to edit the old commit message first.
1295
1296Again, you should never do this to a commit that may already have
1297been merged into another branch; use gitlink:git-revert[1] instead in
1298that case.
1299
1300It is also possible to edit commits further back in the history, but
1301this is an advanced topic to be left for
1302<<cleaning-up-history,another chapter>>.
1303
e34caace 1304[[checkout-of-path]]
b684f830
BF
1305Checking out an old version of a file
1306~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1307
1308In the process of undoing a previous bad change, you may find it
1309useful to check out an older version of a particular file using
1310gitlink:git-checkout[1]. We've used git checkout before to switch
1311branches, but it has quite different behavior if it is given a path
1312name: the command
1313
1314-------------------------------------------------
1315$ git checkout HEAD^ path/to/file
1316-------------------------------------------------
1317
1318replaces path/to/file by the contents it had in the commit HEAD^, and
1319also updates the index to match. It does not change branches.
1320
1321If you just want to look at an old version of the file, without
1322modifying the working directory, you can do that with
1323gitlink:git-show[1]:
1324
1325-------------------------------------------------
ed4eb0d8 1326$ git show HEAD^:path/to/file
b684f830
BF
1327-------------------------------------------------
1328
1329which will display the given version of the file.
1330
e34caace 1331[[ensuring-good-performance]]
d19fbc3c
BF
1332Ensuring good performance
1333-------------------------
1334
1335On large repositories, git depends on compression to keep the history
1336information from taking up to much space on disk or in memory.
1337
1338This compression is not performed automatically. Therefore you
17217090 1339should occasionally run gitlink:git-gc[1]:
d19fbc3c
BF
1340
1341-------------------------------------------------
1342$ git gc
1343-------------------------------------------------
1344
17217090
BF
1345to recompress the archive. This can be very time-consuming, so
1346you may prefer to run git-gc when you are not doing other work.
d19fbc3c 1347
e34caace
BF
1348
1349[[ensuring-reliability]]
11e016a3
BF
1350Ensuring reliability
1351--------------------
1352
e34caace 1353[[checking-for-corruption]]
11e016a3
BF
1354Checking the repository for corruption
1355~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1356
1191ee18
BF
1357The gitlink:git-fsck[1] command runs a number of self-consistency checks
1358on the repository, and reports on any problems. This may take some
21dcb3b7
BF
1359time. The most common warning by far is about "dangling" objects:
1360
1361-------------------------------------------------
04e50e94 1362$ git fsck
21dcb3b7
BF
1363dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1364dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1365dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1366dangling blob 218761f9d90712d37a9c5e36f406f92202db07eb
1367dangling commit bf093535a34a4d35731aa2bd90fe6b176302f14f
1368dangling commit 8e4bec7f2ddaa268bef999853c25755452100f8e
1369dangling tree d50bb86186bf27b681d25af89d3b5b68382e4085
1370dangling tree b24c2473f1fd3d91352a624795be026d64c8841f
1371...
1372-------------------------------------------------
1373
59723040
BF
1374Dangling objects are not a problem. At worst they may take up a little
1375extra disk space. They can sometimes provide a last-resort method of
1376recovery lost work--see <<dangling-objects>> for details. However, if
1377you want, you may remove them with gitlink:git-prune[1] or the --prune
1191ee18 1378option to gitlink:git-gc[1]:
21dcb3b7
BF
1379
1380-------------------------------------------------
1381$ git gc --prune
1382-------------------------------------------------
1383
1191ee18
BF
1384This may be time-consuming. Unlike most other git operations (including
1385git-gc when run without any options), it is not safe to prune while
1386other git operations are in progress in the same repository.
21dcb3b7 1387
e34caace 1388[[recovering-lost-changes]]
11e016a3
BF
1389Recovering lost changes
1390~~~~~~~~~~~~~~~~~~~~~~~
1391
e34caace 1392[[reflogs]]
559e4d7a
BF
1393Reflogs
1394^^^^^^^
1395
1396Say you modify a branch with gitlink:git-reset[1] --hard, and then
1397realize that the branch was the only reference you had to that point in
1398history.
1399
1400Fortunately, git also keeps a log, called a "reflog", of all the
1401previous values of each branch. So in this case you can still find the
1402old history using, for example,
1403
1404-------------------------------------------------
1405$ git log master@{1}
1406-------------------------------------------------
1407
1408This lists the commits reachable from the previous version of the head.
1409This syntax can be used to with any git command that accepts a commit,
1410not just with git log. Some other examples:
1411
1412-------------------------------------------------
1413$ git show master@{2} # See where the branch pointed 2,
1414$ git show master@{3} # 3, ... changes ago.
1415$ gitk master@{yesterday} # See where it pointed yesterday,
1416$ gitk master@{"1 week ago"} # ... or last week
953f3d6f
BF
1417$ git log --walk-reflogs master # show reflog entries for master
1418-------------------------------------------------
1419
1420A separate reflog is kept for the HEAD, so
1421
1422-------------------------------------------------
1423$ git show HEAD@{"1 week ago"}
559e4d7a
BF
1424-------------------------------------------------
1425
953f3d6f
BF
1426will show what HEAD pointed to one week ago, not what the current branch
1427pointed to one week ago. This allows you to see the history of what
1428you've checked out.
1429
559e4d7a 1430The reflogs are kept by default for 30 days, after which they may be
036be17e 1431pruned. See gitlink:git-reflog[1] and gitlink:git-gc[1] to learn
559e4d7a
BF
1432how to control this pruning, and see the "SPECIFYING REVISIONS"
1433section of gitlink:git-rev-parse[1] for details.
1434
1435Note that the reflog history is very different from normal git history.
1436While normal history is shared by every repository that works on the
1437same project, the reflog history is not shared: it tells you only about
1438how the branches in your local repository have changed over time.
1439
59723040 1440[[dangling-object-recovery]]
559e4d7a
BF
1441Examining dangling objects
1442^^^^^^^^^^^^^^^^^^^^^^^^^^
1443
59723040
BF
1444In some situations the reflog may not be able to save you. For example,
1445suppose you delete a branch, then realize you need the history it
1446contained. The reflog is also deleted; however, if you have not yet
1447pruned the repository, then you may still be able to find the lost
1448commits in the dangling objects that git-fsck reports. See
1449<<dangling-objects>> for the details.
559e4d7a
BF
1450
1451-------------------------------------------------
1452$ git fsck
1453dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1454dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1455dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1456...
1457-------------------------------------------------
1458
aacd404e 1459You can examine
559e4d7a
BF
1460one of those dangling commits with, for example,
1461
1462------------------------------------------------
1463$ gitk 7281251ddd --not --all
1464------------------------------------------------
1465
1466which does what it sounds like: it says that you want to see the commit
1467history that is described by the dangling commit(s), but not the
1468history that is described by all your existing branches and tags. Thus
1469you get exactly the history reachable from that commit that is lost.
1470(And notice that it might not be just one commit: we only report the
1471"tip of the line" as being dangling, but there might be a whole deep
79c96c57 1472and complex commit history that was dropped.)
559e4d7a
BF
1473
1474If you decide you want the history back, you can always create a new
1475reference pointing to it, for example, a new branch:
1476
1477------------------------------------------------
1478$ git branch recovered-branch 7281251ddd
1479------------------------------------------------
1480
59723040
BF
1481Other types of dangling objects (blobs and trees) are also possible, and
1482dangling objects can arise in other situations.
1483
11e016a3 1484
e34caace 1485[[sharing-development]]
d19fbc3c 1486Sharing development with others
b684f830 1487===============================
d19fbc3c
BF
1488
1489[[getting-updates-with-git-pull]]
1490Getting updates with git pull
b684f830 1491-----------------------------
d19fbc3c
BF
1492
1493After you clone a repository and make a few changes of your own, you
1494may wish to check the original repository for updates and merge them
1495into your own work.
1496
1497We have already seen <<Updating-a-repository-with-git-fetch,how to
1498keep remote tracking branches up to date>> with gitlink:git-fetch[1],
1499and how to merge two branches. So you can merge in changes from the
1500original repository's master branch with:
1501
1502-------------------------------------------------
1503$ git fetch
1504$ git merge origin/master
1505-------------------------------------------------
1506
1507However, the gitlink:git-pull[1] command provides a way to do this in
1508one step:
1509
1510-------------------------------------------------
1511$ git pull origin master
1512-------------------------------------------------
1513
1514In fact, "origin" is normally the default repository to pull from,
1515and the default branch is normally the HEAD of the remote repository,
1516so often you can accomplish the above with just
1517
1518-------------------------------------------------
1519$ git pull
1520-------------------------------------------------
1521
c64415e2
BF
1522See the descriptions of the branch.<name>.remote and branch.<name>.merge
1523options in gitlink:git-config[1] to learn how to control these defaults
1524depending on the current branch. Also note that the --track option to
1525gitlink:git-branch[1] and gitlink:git-checkout[1] can be used to
1526automatically set the default remote branch to pull from at the time
1527that a branch is created:
1528
1529-------------------------------------------------
1530$ git checkout --track -b origin/maint maint
1531-------------------------------------------------
d19fbc3c
BF
1532
1533In addition to saving you keystrokes, "git pull" also helps you by
1534producing a default commit message documenting the branch and
1535repository that you pulled from.
1536
1537(But note that no such commit will be created in the case of a
1538<<fast-forwards,fast forward>>; instead, your branch will just be
79c96c57 1539updated to point to the latest commit from the upstream branch.)
d19fbc3c 1540
1191ee18
BF
1541The git-pull command can also be given "." as the "remote" repository,
1542in which case it just merges in a branch from the current repository; so
4c63ff45
BF
1543the commands
1544
1545-------------------------------------------------
1546$ git pull . branch
1547$ git merge branch
1548-------------------------------------------------
1549
1550are roughly equivalent. The former is actually very commonly used.
1551
e34caace 1552[[submitting-patches]]
d19fbc3c 1553Submitting patches to a project
b684f830 1554-------------------------------
d19fbc3c
BF
1555
1556If you just have a few changes, the simplest way to submit them may
1557just be to send them as patches in email:
1558
036be17e 1559First, use gitlink:git-format-patch[1]; for example:
d19fbc3c
BF
1560
1561-------------------------------------------------
eb6ae7f4 1562$ git format-patch origin
d19fbc3c
BF
1563-------------------------------------------------
1564
1565will produce a numbered series of files in the current directory, one
1566for each patch in the current branch but not in origin/HEAD.
1567
1568You can then import these into your mail client and send them by
1569hand. However, if you have a lot to send at once, you may prefer to
1570use the gitlink:git-send-email[1] script to automate the process.
1571Consult the mailing list for your project first to determine how they
1572prefer such patches be handled.
1573
e34caace 1574[[importing-patches]]
d19fbc3c 1575Importing patches to a project
b684f830 1576------------------------------
d19fbc3c
BF
1577
1578Git also provides a tool called gitlink:git-am[1] (am stands for
1579"apply mailbox"), for importing such an emailed series of patches.
1580Just save all of the patch-containing messages, in order, into a
1581single mailbox file, say "patches.mbox", then run
1582
1583-------------------------------------------------
eb6ae7f4 1584$ git am -3 patches.mbox
d19fbc3c
BF
1585-------------------------------------------------
1586
1587Git will apply each patch in order; if any conflicts are found, it
1588will stop, and you can fix the conflicts as described in
01997b4a
BF
1589"<<resolving-a-merge,Resolving a merge>>". (The "-3" option tells
1590git to perform a merge; if you would prefer it just to abort and
1591leave your tree and index untouched, you may omit that option.)
1592
1593Once the index is updated with the results of the conflict
1594resolution, instead of creating a new commit, just run
d19fbc3c
BF
1595
1596-------------------------------------------------
1597$ git am --resolved
1598-------------------------------------------------
1599
1600and git will create the commit for you and continue applying the
1601remaining patches from the mailbox.
1602
1603The final result will be a series of commits, one for each patch in
1604the original mailbox, with authorship and commit log message each
1605taken from the message containing each patch.
1606
1607[[setting-up-a-public-repository]]
1608Setting up a public repository
b684f830 1609------------------------------
d19fbc3c
BF
1610
1611Another way to submit changes to a project is to simply tell the
1612maintainer of that project to pull from your repository, exactly as
1613you did in the section "<<getting-updates-with-git-pull, Getting
1614updates with git pull>>".
1615
1616If you and maintainer both have accounts on the same machine, then
1617then you can just pull changes from each other's repositories
79c96c57
MC
1618directly; note that all of the commands (gitlink:git-clone[1],
1619git-fetch[1], git-pull[1], etc.) that accept a URL as an argument
21f13ee2 1620will also accept a local directory name; so, for example, you can
d19fbc3c
BF
1621use
1622
1623-------------------------------------------------
1624$ git clone /path/to/repository
1625$ git pull /path/to/other/repository
1626-------------------------------------------------
1627
1628If this sort of setup is inconvenient or impossible, another (more
1629common) option is to set up a public repository on a public server.
1630This also allows you to cleanly separate private work in progress
1631from publicly visible work.
1632
1633You will continue to do your day-to-day work in your personal
1634repository, but periodically "push" changes from your personal
1635repository into your public repository, allowing other developers to
1636pull from that repository. So the flow of changes, in a situation
1637where there is one other developer with a public repository, looks
1638like this:
1639
1640 you push
1641 your personal repo ------------------> your public repo
1642 ^ |
1643 | |
1644 | you pull | they pull
1645 | |
1646 | |
1647 | they push V
1648 their public repo <------------------- their repo
1649
1650Now, assume your personal repository is in the directory ~/proj. We
1651first create a new clone of the repository:
1652
1653-------------------------------------------------
52c80037 1654$ git clone --bare ~/proj proj.git
d19fbc3c
BF
1655-------------------------------------------------
1656
52c80037
BF
1657The resulting directory proj.git contains a "bare" git repository--it is
1658just the contents of the ".git" directory, without a checked-out copy of
1659a working directory.
d19fbc3c 1660
c64415e2 1661Next, copy proj.git to the server where you plan to host the
d19fbc3c
BF
1662public repository. You can use scp, rsync, or whatever is most
1663convenient.
1664
1665If somebody else maintains the public server, they may already have
1666set up a git service for you, and you may skip to the section
1667"<<pushing-changes-to-a-public-repository,Pushing changes to a public
1668repository>>", below.
1669
1670Otherwise, the following sections explain how to export your newly
1671created public repository:
1672
1673[[exporting-via-http]]
1674Exporting a git repository via http
b684f830 1675-----------------------------------
d19fbc3c
BF
1676
1677The git protocol gives better performance and reliability, but on a
1678host with a web server set up, http exports may be simpler to set up.
1679
1680All you need to do is place the newly created bare git repository in
1681a directory that is exported by the web server, and make some
1682adjustments to give web clients some extra information they need:
1683
1684-------------------------------------------------
1685$ mv proj.git /home/you/public_html/proj.git
1686$ cd proj.git
c64415e2 1687$ git --bare update-server-info
d19fbc3c
BF
1688$ chmod a+x hooks/post-update
1689-------------------------------------------------
1690
1691(For an explanation of the last two lines, see
1692gitlink:git-update-server-info[1], and the documentation
1693link:hooks.txt[Hooks used by git].)
1694
1695Advertise the url of proj.git. Anybody else should then be able to
1696clone or pull from that url, for example with a commandline like:
1697
1698-------------------------------------------------
1699$ git clone http://yourserver.com/~you/proj.git
1700-------------------------------------------------
1701
1702(See also
1703link:howto/setup-git-server-over-http.txt[setup-git-server-over-http]
1704for a slightly more sophisticated setup using WebDAV which also
1705allows pushing over http.)
1706
1707[[exporting-via-git]]
1708Exporting a git repository via the git protocol
b684f830 1709-----------------------------------------------
d19fbc3c
BF
1710
1711This is the preferred method.
1712
1713For now, we refer you to the gitlink:git-daemon[1] man page for
1714instructions. (See especially the examples section.)
1715
1716[[pushing-changes-to-a-public-repository]]
1717Pushing changes to a public repository
b684f830 1718--------------------------------------
d19fbc3c
BF
1719
1720Note that the two techniques outline above (exporting via
1721<<exporting-via-http,http>> or <<exporting-via-git,git>>) allow other
1722maintainers to fetch your latest changes, but they do not allow write
1723access, which you will need to update the public repository with the
1724latest changes created in your private repository.
1725
1726The simplest way to do this is using gitlink:git-push[1] and ssh; to
1727update the remote branch named "master" with the latest state of your
1728branch named "master", run
1729
1730-------------------------------------------------
1731$ git push ssh://yourserver.com/~you/proj.git master:master
1732-------------------------------------------------
1733
1734or just
1735
1736-------------------------------------------------
1737$ git push ssh://yourserver.com/~you/proj.git master
1738-------------------------------------------------
1739
1740As with git-fetch, git-push will complain if this does not result in
1741a <<fast-forwards,fast forward>>. Normally this is a sign of
1742something wrong. However, if you are sure you know what you're
1743doing, you may force git-push to perform the update anyway by
1744proceeding the branch name by a plus sign:
1745
1746-------------------------------------------------
1747$ git push ssh://yourserver.com/~you/proj.git +master
1748-------------------------------------------------
1749
1750As with git-fetch, you may also set up configuration options to
1751save typing; so, for example, after
1752
1753-------------------------------------------------
c64415e2 1754$ cat >>.git/config <<EOF
d19fbc3c
BF
1755[remote "public-repo"]
1756 url = ssh://yourserver.com/~you/proj.git
1757EOF
1758-------------------------------------------------
1759
1760you should be able to perform the above push with just
1761
1762-------------------------------------------------
1763$ git push public-repo master
1764-------------------------------------------------
1765
1766See the explanations of the remote.<name>.url, branch.<name>.remote,
9d13bda3 1767and remote.<name>.push options in gitlink:git-config[1] for
d19fbc3c
BF
1768details.
1769
e34caace 1770[[setting-up-a-shared-repository]]
d19fbc3c 1771Setting up a shared repository
b684f830 1772------------------------------
d19fbc3c
BF
1773
1774Another way to collaborate is by using a model similar to that
1775commonly used in CVS, where several developers with special rights
1776all push to and pull from a single shared repository. See
1777link:cvs-migration.txt[git for CVS users] for instructions on how to
1778set this up.
1779
e34caace 1780[[setting-up-gitweb]]
b684f830
BF
1781Allow web browsing of a repository
1782----------------------------------
d19fbc3c 1783
a8cd1402
BF
1784The gitweb cgi script provides users an easy way to browse your
1785project's files and history without having to install git; see the file
04483524 1786gitweb/INSTALL in the git source tree for instructions on setting it up.
d19fbc3c 1787
e34caace 1788[[sharing-development-examples]]
b684f830
BF
1789Examples
1790--------
d19fbc3c 1791
b684f830 1792TODO: topic branches, typical roles as in everyday.txt, ?
d19fbc3c 1793
d19fbc3c 1794
d19fbc3c 1795[[cleaning-up-history]]
4c63ff45
BF
1796Rewriting history and maintaining patch series
1797==============================================
1798
1799Normally commits are only added to a project, never taken away or
1800replaced. Git is designed with this assumption, and violating it will
1801cause git's merge machinery (for example) to do the wrong thing.
1802
1803However, there is a situation in which it can be useful to violate this
1804assumption.
1805
e34caace 1806[[patch-series]]
4c63ff45
BF
1807Creating the perfect patch series
1808---------------------------------
1809
1810Suppose you are a contributor to a large project, and you want to add a
1811complicated feature, and to present it to the other developers in a way
1812that makes it easy for them to read your changes, verify that they are
1813correct, and understand why you made each change.
1814
b181d57f 1815If you present all of your changes as a single patch (or commit), they
79c96c57 1816may find that it is too much to digest all at once.
4c63ff45
BF
1817
1818If you present them with the entire history of your work, complete with
1819mistakes, corrections, and dead ends, they may be overwhelmed.
1820
1821So the ideal is usually to produce a series of patches such that:
1822
1823 1. Each patch can be applied in order.
1824
1825 2. Each patch includes a single logical change, together with a
1826 message explaining the change.
1827
1828 3. No patch introduces a regression: after applying any initial
1829 part of the series, the resulting project still compiles and
1830 works, and has no bugs that it didn't have before.
1831
1832 4. The complete series produces the same end result as your own
1833 (probably much messier!) development process did.
1834
b181d57f
BF
1835We will introduce some tools that can help you do this, explain how to
1836use them, and then explain some of the problems that can arise because
1837you are rewriting history.
4c63ff45 1838
e34caace 1839[[using-git-rebase]]
4c63ff45
BF
1840Keeping a patch series up to date using git-rebase
1841--------------------------------------------------
1842
79c96c57
MC
1843Suppose that you create a branch "mywork" on a remote-tracking branch
1844"origin", and create some commits on top of it:
4c63ff45
BF
1845
1846-------------------------------------------------
1847$ git checkout -b mywork origin
1848$ vi file.txt
1849$ git commit
1850$ vi otherfile.txt
1851$ git commit
1852...
1853-------------------------------------------------
1854
1855You have performed no merges into mywork, so it is just a simple linear
1856sequence of patches on top of "origin":
1857
1dc71a91 1858................................................
4c63ff45
BF
1859 o--o--o <-- origin
1860 \
1861 o--o--o <-- mywork
1dc71a91 1862................................................
4c63ff45
BF
1863
1864Some more interesting work has been done in the upstream project, and
1865"origin" has advanced:
1866
1dc71a91 1867................................................
4c63ff45
BF
1868 o--o--O--o--o--o <-- origin
1869 \
1870 a--b--c <-- mywork
1dc71a91 1871................................................
4c63ff45
BF
1872
1873At this point, you could use "pull" to merge your changes back in;
1874the result would create a new merge commit, like this:
1875
1dc71a91 1876................................................
4c63ff45
BF
1877 o--o--O--o--o--o <-- origin
1878 \ \
1879 a--b--c--m <-- mywork
1dc71a91 1880................................................
4c63ff45
BF
1881
1882However, if you prefer to keep the history in mywork a simple series of
1883commits without any merges, you may instead choose to use
1884gitlink:git-rebase[1]:
1885
1886-------------------------------------------------
1887$ git checkout mywork
1888$ git rebase origin
1889-------------------------------------------------
1890
b181d57f
BF
1891This will remove each of your commits from mywork, temporarily saving
1892them as patches (in a directory named ".dotest"), update mywork to
1893point at the latest version of origin, then apply each of the saved
1894patches to the new mywork. The result will look like:
4c63ff45
BF
1895
1896
1dc71a91 1897................................................
4c63ff45
BF
1898 o--o--O--o--o--o <-- origin
1899 \
1900 a'--b'--c' <-- mywork
1dc71a91 1901................................................
4c63ff45 1902
b181d57f
BF
1903In the process, it may discover conflicts. In that case it will stop
1904and allow you to fix the conflicts; after fixing conflicts, use "git
1905add" to update the index with those contents, and then, instead of
1906running git-commit, just run
4c63ff45
BF
1907
1908-------------------------------------------------
1909$ git rebase --continue
1910-------------------------------------------------
1911
1912and git will continue applying the rest of the patches.
1913
1914At any point you may use the --abort option to abort this process and
1915return mywork to the state it had before you started the rebase:
1916
1917-------------------------------------------------
1918$ git rebase --abort
1919-------------------------------------------------
1920
e34caace 1921[[modifying-one-commit]]
365aa199
BF
1922Modifying a single commit
1923-------------------------
1924
1925We saw in <<fixing-a-mistake-by-editing-history>> that you can replace the
1926most recent commit using
1927
1928-------------------------------------------------
1929$ git commit --amend
1930-------------------------------------------------
1931
1932which will replace the old commit by a new commit incorporating your
1933changes, giving you a chance to edit the old commit message first.
1934
1935You can also use a combination of this and gitlink:git-rebase[1] to edit
1936commits further back in your history. First, tag the problematic commit with
1937
1938-------------------------------------------------
1939$ git tag bad mywork~5
1940-------------------------------------------------
1941
1942(Either gitk or git-log may be useful for finding the commit.)
1943
25d9f3fa
BF
1944Then check out that commit, edit it, and rebase the rest of the series
1945on top of it (note that we could check out the commit on a temporary
1946branch, but instead we're using a <<detached-head,detached head>>):
365aa199
BF
1947
1948-------------------------------------------------
25d9f3fa 1949$ git checkout bad
365aa199
BF
1950$ # make changes here and update the index
1951$ git commit --amend
25d9f3fa 1952$ git rebase --onto HEAD bad mywork
365aa199
BF
1953-------------------------------------------------
1954
25d9f3fa
BF
1955When you're done, you'll be left with mywork checked out, with the top
1956patches on mywork reapplied on top of your modified commit. You can
365aa199
BF
1957then clean up with
1958
1959-------------------------------------------------
365aa199
BF
1960$ git tag -d bad
1961-------------------------------------------------
1962
1963Note that the immutable nature of git history means that you haven't really
1964"modified" existing commits; instead, you have replaced the old commits with
1965new commits having new object names.
1966
e34caace 1967[[reordering-patch-series]]
4c63ff45
BF
1968Reordering or selecting from a patch series
1969-------------------------------------------
1970
b181d57f
BF
1971Given one existing commit, the gitlink:git-cherry-pick[1] command
1972allows you to apply the change introduced by that commit and create a
1973new commit that records it. So, for example, if "mywork" points to a
1974series of patches on top of "origin", you might do something like:
1975
1976-------------------------------------------------
1977$ git checkout -b mywork-new origin
1978$ gitk origin..mywork &
1979-------------------------------------------------
1980
1981And browse through the list of patches in the mywork branch using gitk,
1982applying them (possibly in a different order) to mywork-new using
1983cherry-pick, and possibly modifying them as you go using commit
1984--amend.
1985
1986Another technique is to use git-format-patch to create a series of
1987patches, then reset the state to before the patches:
4c63ff45 1988
b181d57f
BF
1989-------------------------------------------------
1990$ git format-patch origin
1991$ git reset --hard origin
1992-------------------------------------------------
4c63ff45 1993
b181d57f
BF
1994Then modify, reorder, or eliminate patches as preferred before applying
1995them again with gitlink:git-am[1].
4c63ff45 1996
e34caace 1997[[patch-series-tools]]
4c63ff45
BF
1998Other tools
1999-----------
2000
b181d57f 2001There are numerous other tools, such as stgit, which exist for the
79c96c57 2002purpose of maintaining a patch series. These are outside of the scope of
b181d57f 2003this manual.
4c63ff45 2004
e34caace 2005[[problems-with-rewriting-history]]
4c63ff45
BF
2006Problems with rewriting history
2007-------------------------------
2008
b181d57f
BF
2009The primary problem with rewriting the history of a branch has to do
2010with merging. Suppose somebody fetches your branch and merges it into
2011their branch, with a result something like this:
2012
1dc71a91 2013................................................
b181d57f
BF
2014 o--o--O--o--o--o <-- origin
2015 \ \
2016 t--t--t--m <-- their branch:
1dc71a91 2017................................................
b181d57f
BF
2018
2019Then suppose you modify the last three commits:
2020
1dc71a91 2021................................................
b181d57f
BF
2022 o--o--o <-- new head of origin
2023 /
2024 o--o--O--o--o--o <-- old head of origin
1dc71a91 2025................................................
b181d57f
BF
2026
2027If we examined all this history together in one repository, it will
2028look like:
2029
1dc71a91 2030................................................
b181d57f
BF
2031 o--o--o <-- new head of origin
2032 /
2033 o--o--O--o--o--o <-- old head of origin
2034 \ \
2035 t--t--t--m <-- their branch:
1dc71a91 2036................................................
b181d57f
BF
2037
2038Git has no way of knowing that the new head is an updated version of
2039the old head; it treats this situation exactly the same as it would if
2040two developers had independently done the work on the old and new heads
2041in parallel. At this point, if someone attempts to merge the new head
2042in to their branch, git will attempt to merge together the two (old and
2043new) lines of development, instead of trying to replace the old by the
2044new. The results are likely to be unexpected.
2045
2046You may still choose to publish branches whose history is rewritten,
2047and it may be useful for others to be able to fetch those branches in
2048order to examine or test them, but they should not attempt to pull such
2049branches into their own work.
2050
2051For true distributed development that supports proper merging,
2052published branches should never be rewritten.
2053
e34caace 2054[[advanced-branch-management]]
b181d57f
BF
2055Advanced branch management
2056==========================
4c63ff45 2057
e34caace 2058[[fetching-individual-branches]]
b181d57f
BF
2059Fetching individual branches
2060----------------------------
2061
2062Instead of using gitlink:git-remote[1], you can also choose just
2063to update one branch at a time, and to store it locally under an
2064arbitrary name:
2065
2066-------------------------------------------------
2067$ git fetch origin todo:my-todo-work
2068-------------------------------------------------
2069
2070The first argument, "origin", just tells git to fetch from the
2071repository you originally cloned from. The second argument tells git
2072to fetch the branch named "todo" from the remote repository, and to
2073store it locally under the name refs/heads/my-todo-work.
2074
2075You can also fetch branches from other repositories; so
2076
2077-------------------------------------------------
2078$ git fetch git://example.com/proj.git master:example-master
2079-------------------------------------------------
2080
2081will create a new branch named "example-master" and store in it the
2082branch named "master" from the repository at the given URL. If you
2083already have a branch named example-master, it will attempt to
59723040
BF
2084<<fast-forwards,fast-forward>> to the commit given by example.com's
2085master branch. In more detail:
b181d57f 2086
59723040
BF
2087[[fetch-fast-forwards]]
2088git fetch and fast-forwards
2089---------------------------
b181d57f
BF
2090
2091In the previous example, when updating an existing branch, "git
2092fetch" checks to make sure that the most recent commit on the remote
2093branch is a descendant of the most recent commit on your copy of the
2094branch before updating your copy of the branch to point at the new
59723040 2095commit. Git calls this process a <<fast-forwards,fast forward>>.
b181d57f
BF
2096
2097A fast forward looks something like this:
2098
1dc71a91 2099................................................
b181d57f
BF
2100 o--o--o--o <-- old head of the branch
2101 \
2102 o--o--o <-- new head of the branch
1dc71a91 2103................................................
b181d57f
BF
2104
2105
2106In some cases it is possible that the new head will *not* actually be
2107a descendant of the old head. For example, the developer may have
2108realized she made a serious mistake, and decided to backtrack,
2109resulting in a situation like:
2110
1dc71a91 2111................................................
b181d57f
BF
2112 o--o--o--o--a--b <-- old head of the branch
2113 \
2114 o--o--o <-- new head of the branch
1dc71a91 2115................................................
b181d57f
BF
2116
2117In this case, "git fetch" will fail, and print out a warning.
2118
2119In that case, you can still force git to update to the new head, as
2120described in the following section. However, note that in the
2121situation above this may mean losing the commits labeled "a" and "b",
2122unless you've already created a reference of your own pointing to
2123them.
2124
e34caace 2125[[forcing-fetch]]
b181d57f
BF
2126Forcing git fetch to do non-fast-forward updates
2127------------------------------------------------
2128
2129If git fetch fails because the new head of a branch is not a
2130descendant of the old head, you may force the update with:
2131
2132-------------------------------------------------
2133$ git fetch git://example.com/proj.git +master:refs/remotes/example/master
2134-------------------------------------------------
2135
c64415e2
BF
2136Note the addition of the "+" sign. Alternatively, you can use the "-f"
2137flag to force updates of all the fetched branches, as in:
2138
2139-------------------------------------------------
2140$ git fetch -f origin
2141-------------------------------------------------
2142
2143Be aware that commits that the old version of example/master pointed at
2144may be lost, as we saw in the previous section.
b181d57f 2145
e34caace 2146[[remote-branch-configuration]]
b181d57f
BF
2147Configuring remote branches
2148---------------------------
2149
2150We saw above that "origin" is just a shortcut to refer to the
79c96c57 2151repository that you originally cloned from. This information is
b181d57f 2152stored in git configuration variables, which you can see using
9d13bda3 2153gitlink:git-config[1]:
b181d57f
BF
2154
2155-------------------------------------------------
9d13bda3 2156$ git config -l
b181d57f
BF
2157core.repositoryformatversion=0
2158core.filemode=true
2159core.logallrefupdates=true
2160remote.origin.url=git://git.kernel.org/pub/scm/git/git.git
2161remote.origin.fetch=+refs/heads/*:refs/remotes/origin/*
2162branch.master.remote=origin
2163branch.master.merge=refs/heads/master
2164-------------------------------------------------
2165
2166If there are other repositories that you also use frequently, you can
2167create similar configuration options to save typing; for example,
2168after
2169
2170-------------------------------------------------
9d13bda3 2171$ git config remote.example.url git://example.com/proj.git
b181d57f
BF
2172-------------------------------------------------
2173
2174then the following two commands will do the same thing:
2175
2176-------------------------------------------------
2177$ git fetch git://example.com/proj.git master:refs/remotes/example/master
2178$ git fetch example master:refs/remotes/example/master
2179-------------------------------------------------
2180
2181Even better, if you add one more option:
2182
2183-------------------------------------------------
9d13bda3 2184$ git config remote.example.fetch master:refs/remotes/example/master
b181d57f
BF
2185-------------------------------------------------
2186
2187then the following commands will all do the same thing:
2188
2189-------------------------------------------------
52c80037
BF
2190$ git fetch git://example.com/proj.git master:refs/remotes/example/master
2191$ git fetch example master:refs/remotes/example/master
b181d57f
BF
2192$ git fetch example
2193-------------------------------------------------
2194
2195You can also add a "+" to force the update each time:
2196
2197-------------------------------------------------
9d13bda3 2198$ git config remote.example.fetch +master:ref/remotes/example/master
b181d57f
BF
2199-------------------------------------------------
2200
2201Don't do this unless you're sure you won't mind "git fetch" possibly
2202throwing away commits on mybranch.
2203
2204Also note that all of the above configuration can be performed by
2205directly editing the file .git/config instead of using
9d13bda3 2206gitlink:git-config[1].
b181d57f 2207
9d13bda3 2208See gitlink:git-config[1] for more details on the configuration
b181d57f 2209options mentioned above.
d19fbc3c 2210
d19fbc3c 2211
35121930 2212[[git-internals]]
d19fbc3c
BF
2213Git internals
2214=============
2215
a536b08b
BF
2216Git depends on two fundamental abstractions: the "object database", and
2217the "current directory cache" aka "index".
b181d57f 2218
e34caace 2219[[the-object-database]]
b181d57f
BF
2220The Object Database
2221-------------------
2222
2223The object database is literally just a content-addressable collection
2224of objects. All objects are named by their content, which is
2225approximated by the SHA1 hash of the object itself. Objects may refer
2226to other objects (by referencing their SHA1 hash), and so you can
2227build up a hierarchy of objects.
2228
c64415e2 2229All objects have a statically determined "type" which is
b181d57f
BF
2230determined at object creation time, and which identifies the format of
2231the object (i.e. how it is used, and how it can refer to other
2232objects). There are currently four different object types: "blob",
a536b08b 2233"tree", "commit", and "tag".
b181d57f 2234
a536b08b
BF
2235A <<def_blob_object,"blob" object>> cannot refer to any other object,
2236and is, as the name implies, a pure storage object containing some
2237user data. It is used to actually store the file data, i.e. a blob
2238object is associated with some particular version of some file.
b181d57f 2239
a536b08b
BF
2240A <<def_tree_object,"tree" object>> is an object that ties one or more
2241"blob" objects into a directory structure. In addition, a tree object
2242can refer to other tree objects, thus creating a directory hierarchy.
b181d57f 2243
a536b08b
BF
2244A <<def_commit_object,"commit" object>> ties such directory hierarchies
2245together into a <<def_DAG,directed acyclic graph>> of revisions - each
2246"commit" is associated with exactly one tree (the directory hierarchy at
2247the time of the commit). In addition, a "commit" refers to one or more
2248"parent" commit objects that describe the history of how we arrived at
2249that directory hierarchy.
b181d57f
BF
2250
2251As a special case, a commit object with no parents is called the "root"
c64415e2 2252commit, and is the point of an initial project commit. Each project
b181d57f
BF
2253must have at least one root, and while you can tie several different
2254root objects together into one project by creating a commit object which
2255has two or more separate roots as its ultimate parents, that's probably
2256just going to confuse people. So aim for the notion of "one root object
2257per project", even if git itself does not enforce that.
2258
a536b08b
BF
2259A <<def_tag_object,"tag" object>> symbolically identifies and can be
2260used to sign other objects. It contains the identifier and type of
2261another object, a symbolic name (of course!) and, optionally, a
2262signature.
b181d57f
BF
2263
2264Regardless of object type, all objects share the following
2265characteristics: they are all deflated with zlib, and have a header
2266that not only specifies their type, but also provides size information
2267about the data in the object. It's worth noting that the SHA1 hash
2268that is used to name the object is the hash of the original data
2269plus this header, so `sha1sum` 'file' does not match the object name
2270for 'file'.
2271(Historical note: in the dawn of the age of git the hash
2272was the sha1 of the 'compressed' object.)
2273
2274As a result, the general consistency of an object can always be tested
2275independently of the contents or the type of the object: all objects can
2276be validated by verifying that (a) their hashes match the content of the
2277file and (b) the object successfully inflates to a stream of bytes that
2278forms a sequence of <ascii type without space> + <space> + <ascii decimal
2279size> + <byte\0> + <binary object data>.
2280
2281The structured objects can further have their structure and
2282connectivity to other objects verified. This is generally done with
04e50e94 2283the `git-fsck` program, which generates a full dependency graph
b181d57f
BF
2284of all objects, and verifies their internal consistency (in addition
2285to just verifying their superficial consistency through the hash).
2286
2287The object types in some more detail:
2288
e34caace 2289[[blob-object]]
b181d57f
BF
2290Blob Object
2291-----------
2292
2293A "blob" object is nothing but a binary blob of data, and doesn't
2294refer to anything else. There is no signature or any other
2295verification of the data, so while the object is consistent (it 'is'
2296indexed by its sha1 hash, so the data itself is certainly correct), it
2297has absolutely no other attributes. No name associations, no
2298permissions. It is purely a blob of data (i.e. normally "file
2299contents").
2300
2301In particular, since the blob is entirely defined by its data, if two
2302files in a directory tree (or in multiple different versions of the
2303repository) have the same contents, they will share the same blob
2304object. The object is totally independent of its location in the
2305directory tree, and renaming a file does not change the object that
2306file is associated with in any way.
2307
2308A blob is typically created when gitlink:git-update-index[1]
2309is run, and its data can be accessed by gitlink:git-cat-file[1].
2310
e34caace 2311[[tree-object]]
b181d57f
BF
2312Tree Object
2313-----------
2314
2315The next hierarchical object type is the "tree" object. A tree object
2316is a list of mode/name/blob data, sorted by name. Alternatively, the
2317mode data may specify a directory mode, in which case instead of
2318naming a blob, that name is associated with another TREE object.
2319
2320Like the "blob" object, a tree object is uniquely determined by the
2321set contents, and so two separate but identical trees will always
2322share the exact same object. This is true at all levels, i.e. it's
2323true for a "leaf" tree (which does not refer to any other trees, only
2324blobs) as well as for a whole subdirectory.
2325
2326For that reason a "tree" object is just a pure data abstraction: it
2327has no history, no signatures, no verification of validity, except
2328that since the contents are again protected by the hash itself, we can
2329trust that the tree is immutable and its contents never change.
2330
2331So you can trust the contents of a tree to be valid, the same way you
2332can trust the contents of a blob, but you don't know where those
2333contents 'came' from.
2334
2335Side note on trees: since a "tree" object is a sorted list of
2336"filename+content", you can create a diff between two trees without
2337actually having to unpack two trees. Just ignore all common parts,
2338and your diff will look right. In other words, you can effectively
2339(and efficiently) tell the difference between any two random trees by
2340O(n) where "n" is the size of the difference, rather than the size of
2341the tree.
2342
2343Side note 2 on trees: since the name of a "blob" depends entirely and
2344exclusively on its contents (i.e. there are no names or permissions
2345involved), you can see trivial renames or permission changes by
2346noticing that the blob stayed the same. However, renames with data
2347changes need a smarter "diff" implementation.
2348
2349A tree is created with gitlink:git-write-tree[1] and
2350its data can be accessed by gitlink:git-ls-tree[1].
2351Two trees can be compared with gitlink:git-diff-tree[1].
2352
e34caace 2353[[commit-object]]
b181d57f
BF
2354Commit Object
2355-------------
2356
2357The "commit" object is an object that introduces the notion of
2358history into the picture. In contrast to the other objects, it
2359doesn't just describe the physical state of a tree, it describes how
2360we got there, and why.
2361
2362A "commit" is defined by the tree-object that it results in, the
2363parent commits (zero, one or more) that led up to that point, and a
2364comment on what happened. Again, a commit is not trusted per se:
2365the contents are well-defined and "safe" due to the cryptographically
2366strong signatures at all levels, but there is no reason to believe
2367that the tree is "good" or that the merge information makes sense.
2368The parents do not have to actually have any relationship with the
2369result, for example.
2370
c64415e2 2371Note on commits: unlike some SCM's, commits do not contain
b181d57f
BF
2372rename information or file mode change information. All of that is
2373implicit in the trees involved (the result tree, and the result trees
2374of the parents), and describing that makes no sense in this idiotic
2375file manager.
2376
2377A commit is created with gitlink:git-commit-tree[1] and
2378its data can be accessed by gitlink:git-cat-file[1].
2379
e34caace 2380[[trust]]
b181d57f
BF
2381Trust
2382-----
2383
2384An aside on the notion of "trust". Trust is really outside the scope
2385of "git", but it's worth noting a few things. First off, since
2386everything is hashed with SHA1, you 'can' trust that an object is
2387intact and has not been messed with by external sources. So the name
2388of an object uniquely identifies a known state - just not a state that
2389you may want to trust.
2390
2391Furthermore, since the SHA1 signature of a commit refers to the
2392SHA1 signatures of the tree it is associated with and the signatures
2393of the parent, a single named commit specifies uniquely a whole set
2394of history, with full contents. You can't later fake any step of the
2395way once you have the name of a commit.
2396
2397So to introduce some real trust in the system, the only thing you need
2398to do is to digitally sign just 'one' special note, which includes the
2399name of a top-level commit. Your digital signature shows others
2400that you trust that commit, and the immutability of the history of
2401commits tells others that they can trust the whole history.
2402
2403In other words, you can easily validate a whole archive by just
2404sending out a single email that tells the people the name (SHA1 hash)
2405of the top commit, and digitally sign that email using something
2406like GPG/PGP.
2407
2408To assist in this, git also provides the tag object...
2409
e34caace 2410[[tag-object]]
b181d57f
BF
2411Tag Object
2412----------
2413
2414Git provides the "tag" object to simplify creating, managing and
2415exchanging symbolic and signed tokens. The "tag" object at its
2416simplest simply symbolically identifies another object by containing
2417the sha1, type and symbolic name.
2418
2419However it can optionally contain additional signature information
2420(which git doesn't care about as long as there's less than 8k of
2421it). This can then be verified externally to git.
2422
2423Note that despite the tag features, "git" itself only handles content
2424integrity; the trust framework (and signature provision and
2425verification) has to come from outside.
2426
2427A tag is created with gitlink:git-mktag[1],
2428its data can be accessed by gitlink:git-cat-file[1],
2429and the signature can be verified by
2430gitlink:git-verify-tag[1].
2431
2432
e34caace 2433[[the-index]]
b181d57f
BF
2434The "index" aka "Current Directory Cache"
2435-----------------------------------------
2436
2437The index is a simple binary file, which contains an efficient
c64415e2 2438representation of the contents of a virtual directory. It
b181d57f
BF
2439does so by a simple array that associates a set of names, dates,
2440permissions and content (aka "blob") objects together. The cache is
2441always kept ordered by name, and names are unique (with a few very
2442specific rules) at any point in time, but the cache has no long-term
2443meaning, and can be partially updated at any time.
2444
2445In particular, the index certainly does not need to be consistent with
2446the current directory contents (in fact, most operations will depend on
2447different ways to make the index 'not' be consistent with the directory
2448hierarchy), but it has three very important attributes:
2449
2450'(a) it can re-generate the full state it caches (not just the
2451directory structure: it contains pointers to the "blob" objects so
2452that it can regenerate the data too)'
2453
2454As a special case, there is a clear and unambiguous one-way mapping
2455from a current directory cache to a "tree object", which can be
2456efficiently created from just the current directory cache without
2457actually looking at any other data. So a directory cache at any one
2458time uniquely specifies one and only one "tree" object (but has
2459additional data to make it easy to match up that tree object with what
2460has happened in the directory)
2461
2462'(b) it has efficient methods for finding inconsistencies between that
2463cached state ("tree object waiting to be instantiated") and the
2464current state.'
2465
2466'(c) it can additionally efficiently represent information about merge
2467conflicts between different tree objects, allowing each pathname to be
2468associated with sufficient information about the trees involved that
2469you can create a three-way merge between them.'
2470
79c96c57 2471Those are the ONLY three things that the directory cache does. It's a
b181d57f
BF
2472cache, and the normal operation is to re-generate it completely from a
2473known tree object, or update/compare it with a live tree that is being
2474developed. If you blow the directory cache away entirely, you generally
2475haven't lost any information as long as you have the name of the tree
2476that it described.
2477
2478At the same time, the index is at the same time also the
2479staging area for creating new trees, and creating a new tree always
2480involves a controlled modification of the index file. In particular,
2481the index file can have the representation of an intermediate tree that
2482has not yet been instantiated. So the index can be thought of as a
2483write-back cache, which can contain dirty information that has not yet
2484been written back to the backing store.
2485
2486
2487
e34caace 2488[[the-workflow]]
b181d57f
BF
2489The Workflow
2490------------
2491
2492Generally, all "git" operations work on the index file. Some operations
2493work *purely* on the index file (showing the current state of the
2494index), but most operations move data to and from the index file. Either
2495from the database or from the working directory. Thus there are four
2496main combinations:
2497
e34caace 2498[[working-directory-to-index]]
b181d57f
BF
2499working directory -> index
2500~~~~~~~~~~~~~~~~~~~~~~~~~~
2501
2502You update the index with information from the working directory with
2503the gitlink:git-update-index[1] command. You
2504generally update the index information by just specifying the filename
2505you want to update, like so:
2506
2507-------------------------------------------------
2508$ git-update-index filename
2509-------------------------------------------------
2510
2511but to avoid common mistakes with filename globbing etc, the command
2512will not normally add totally new entries or remove old entries,
2513i.e. it will normally just update existing cache entries.
2514
2515To tell git that yes, you really do realize that certain files no
2516longer exist, or that new files should be added, you
2517should use the `--remove` and `--add` flags respectively.
2518
2519NOTE! A `--remove` flag does 'not' mean that subsequent filenames will
2520necessarily be removed: if the files still exist in your directory
2521structure, the index will be updated with their new status, not
2522removed. The only thing `--remove` means is that update-cache will be
2523considering a removed file to be a valid thing, and if the file really
2524does not exist any more, it will update the index accordingly.
2525
2526As a special case, you can also do `git-update-index --refresh`, which
2527will refresh the "stat" information of each index to match the current
2528stat information. It will 'not' update the object status itself, and
2529it will only update the fields that are used to quickly test whether
2530an object still matches its old backing store object.
2531
e34caace 2532[[index-to-object-database]]
b181d57f
BF
2533index -> object database
2534~~~~~~~~~~~~~~~~~~~~~~~~
2535
2536You write your current index file to a "tree" object with the program
2537
2538-------------------------------------------------
2539$ git-write-tree
2540-------------------------------------------------
2541
2542that doesn't come with any options - it will just write out the
2543current index into the set of tree objects that describe that state,
2544and it will return the name of the resulting top-level tree. You can
2545use that tree to re-generate the index at any time by going in the
2546other direction:
2547
e34caace 2548[[object-database-to-index]]
b181d57f
BF
2549object database -> index
2550~~~~~~~~~~~~~~~~~~~~~~~~
2551
2552You read a "tree" file from the object database, and use that to
2553populate (and overwrite - don't do this if your index contains any
2554unsaved state that you might want to restore later!) your current
2555index. Normal operation is just
2556
2557-------------------------------------------------
2558$ git-read-tree <sha1 of tree>
2559-------------------------------------------------
2560
2561and your index file will now be equivalent to the tree that you saved
2562earlier. However, that is only your 'index' file: your working
2563directory contents have not been modified.
2564
e34caace 2565[[index-to-working-directory]]
b181d57f
BF
2566index -> working directory
2567~~~~~~~~~~~~~~~~~~~~~~~~~~
2568
2569You update your working directory from the index by "checking out"
2570files. This is not a very common operation, since normally you'd just
2571keep your files updated, and rather than write to your working
2572directory, you'd tell the index files about the changes in your
2573working directory (i.e. `git-update-index`).
2574
2575However, if you decide to jump to a new version, or check out somebody
2576else's version, or just restore a previous tree, you'd populate your
2577index file with read-tree, and then you need to check out the result
2578with
2579
2580-------------------------------------------------
2581$ git-checkout-index filename
2582-------------------------------------------------
2583
2584or, if you want to check out all of the index, use `-a`.
2585
2586NOTE! git-checkout-index normally refuses to overwrite old files, so
2587if you have an old version of the tree already checked out, you will
2588need to use the "-f" flag ('before' the "-a" flag or the filename) to
2589'force' the checkout.
2590
2591
2592Finally, there are a few odds and ends which are not purely moving
2593from one representation to the other:
2594
e34caace 2595[[tying-it-all-together]]
b181d57f
BF
2596Tying it all together
2597~~~~~~~~~~~~~~~~~~~~~
2598
2599To commit a tree you have instantiated with "git-write-tree", you'd
2600create a "commit" object that refers to that tree and the history
2601behind it - most notably the "parent" commits that preceded it in
2602history.
2603
2604Normally a "commit" has one parent: the previous state of the tree
2605before a certain change was made. However, sometimes it can have two
2606or more parent commits, in which case we call it a "merge", due to the
2607fact that such a commit brings together ("merges") two or more
2608previous states represented by other commits.
2609
2610In other words, while a "tree" represents a particular directory state
2611of a working directory, a "commit" represents that state in "time",
2612and explains how we got there.
2613
2614You create a commit object by giving it the tree that describes the
2615state at the time of the commit, and a list of parents:
2616
2617-------------------------------------------------
2618$ git-commit-tree <tree> -p <parent> [-p <parent2> ..]
2619-------------------------------------------------
2620
2621and then giving the reason for the commit on stdin (either through
2622redirection from a pipe or file, or by just typing it at the tty).
2623
2624git-commit-tree will return the name of the object that represents
2625that commit, and you should save it away for later use. Normally,
2626you'd commit a new `HEAD` state, and while git doesn't care where you
2627save the note about that state, in practice we tend to just write the
2628result to the file pointed at by `.git/HEAD`, so that we can always see
2629what the last committed state was.
2630
2631Here is an ASCII art by Jon Loeliger that illustrates how
2632various pieces fit together.
2633
2634------------
2635
2636 commit-tree
2637 commit obj
2638 +----+
2639 | |
2640 | |
2641 V V
2642 +-----------+
2643 | Object DB |
2644 | Backing |
2645 | Store |
2646 +-----------+
2647 ^
2648 write-tree | |
2649 tree obj | |
2650 | | read-tree
2651 | | tree obj
2652 V
2653 +-----------+
2654 | Index |
2655 | "cache" |
2656 +-----------+
2657 update-index ^
2658 blob obj | |
2659 | |
2660 checkout-index -u | | checkout-index
2661 stat | | blob obj
2662 V
2663 +-----------+
2664 | Working |
2665 | Directory |
2666 +-----------+
2667
2668------------
2669
2670
e34caace 2671[[examining-the-data]]
b181d57f
BF
2672Examining the data
2673------------------
2674
2675You can examine the data represented in the object database and the
2676index with various helper tools. For every object, you can use
2677gitlink:git-cat-file[1] to examine details about the
2678object:
2679
2680-------------------------------------------------
2681$ git-cat-file -t <objectname>
2682-------------------------------------------------
2683
2684shows the type of the object, and once you have the type (which is
2685usually implicit in where you find the object), you can use
2686
2687-------------------------------------------------
2688$ git-cat-file blob|tree|commit|tag <objectname>
2689-------------------------------------------------
2690
2691to show its contents. NOTE! Trees have binary content, and as a result
2692there is a special helper for showing that content, called
2693`git-ls-tree`, which turns the binary content into a more easily
2694readable form.
2695
2696It's especially instructive to look at "commit" objects, since those
2697tend to be small and fairly self-explanatory. In particular, if you
2698follow the convention of having the top commit name in `.git/HEAD`,
2699you can do
2700
2701-------------------------------------------------
2702$ git-cat-file commit HEAD
2703-------------------------------------------------
2704
2705to see what the top commit was.
2706
e34caace 2707[[merging-multiple-trees]]
b181d57f 2708Merging multiple trees
d19fbc3c
BF
2709----------------------
2710
b181d57f
BF
2711Git helps you do a three-way merge, which you can expand to n-way by
2712repeating the merge procedure arbitrary times until you finally
2713"commit" the state. The normal situation is that you'd only do one
2714three-way merge (two parents), and commit it, but if you like to, you
2715can do multiple parents in one go.
2716
2717To do a three-way merge, you need the two sets of "commit" objects
2718that you want to merge, use those to find the closest common parent (a
2719third "commit" object), and then use those commit objects to find the
2720state of the directory ("tree" object) at these points.
2721
2722To get the "base" for the merge, you first look up the common parent
2723of two commits with
2724
2725-------------------------------------------------
2726$ git-merge-base <commit1> <commit2>
2727-------------------------------------------------
2728
2729which will return you the commit they are both based on. You should
2730now look up the "tree" objects of those commits, which you can easily
2731do with (for example)
2732
2733-------------------------------------------------
2734$ git-cat-file commit <commitname> | head -1
2735-------------------------------------------------
2736
2737since the tree object information is always the first line in a commit
2738object.
2739
1191ee18 2740Once you know the three trees you are going to merge (the one "original"
c64415e2 2741tree, aka the common tree, and the two "result" trees, aka the branches
1191ee18
BF
2742you want to merge), you do a "merge" read into the index. This will
2743complain if it has to throw away your old index contents, so you should
b181d57f 2744make sure that you've committed those - in fact you would normally
1191ee18
BF
2745always do a merge against your last commit (which should thus match what
2746you have in your current index anyway).
b181d57f
BF
2747
2748To do the merge, do
2749
2750-------------------------------------------------
2751$ git-read-tree -m -u <origtree> <yourtree> <targettree>
2752-------------------------------------------------
2753
2754which will do all trivial merge operations for you directly in the
2755index file, and you can just write the result out with
2756`git-write-tree`.
2757
2758
e34caace 2759[[merging-multiple-trees-2]]
b181d57f
BF
2760Merging multiple trees, continued
2761---------------------------------
2762
2763Sadly, many merges aren't trivial. If there are files that have
2764been added.moved or removed, or if both branches have modified the
2765same file, you will be left with an index tree that contains "merge
2766entries" in it. Such an index tree can 'NOT' be written out to a tree
2767object, and you will have to resolve any such merge clashes using
2768other tools before you can write out the result.
2769
2770You can examine such index state with `git-ls-files --unmerged`
2771command. An example:
2772
2773------------------------------------------------
2774$ git-read-tree -m $orig HEAD $target
2775$ git-ls-files --unmerged
2776100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1 hello.c
2777100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2 hello.c
2778100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello.c
2779------------------------------------------------
2780
2781Each line of the `git-ls-files --unmerged` output begins with
2782the blob mode bits, blob SHA1, 'stage number', and the
2783filename. The 'stage number' is git's way to say which tree it
2784came from: stage 1 corresponds to `$orig` tree, stage 2 `HEAD`
2785tree, and stage3 `$target` tree.
2786
2787Earlier we said that trivial merges are done inside
2788`git-read-tree -m`. For example, if the file did not change
2789from `$orig` to `HEAD` nor `$target`, or if the file changed
2790from `$orig` to `HEAD` and `$orig` to `$target` the same way,
2791obviously the final outcome is what is in `HEAD`. What the
2792above example shows is that file `hello.c` was changed from
2793`$orig` to `HEAD` and `$orig` to `$target` in a different way.
2794You could resolve this by running your favorite 3-way merge
c64415e2
BF
2795program, e.g. `diff3`, `merge`, or git's own merge-file, on
2796the blob objects from these three stages yourself, like this:
b181d57f
BF
2797
2798------------------------------------------------
2799$ git-cat-file blob 263414f... >hello.c~1
2800$ git-cat-file blob 06fa6a2... >hello.c~2
2801$ git-cat-file blob cc44c73... >hello.c~3
c64415e2 2802$ git merge-file hello.c~2 hello.c~1 hello.c~3
b181d57f
BF
2803------------------------------------------------
2804
2805This would leave the merge result in `hello.c~2` file, along
2806with conflict markers if there are conflicts. After verifying
2807the merge result makes sense, you can tell git what the final
2808merge result for this file is by:
2809
2810-------------------------------------------------
2811$ mv -f hello.c~2 hello.c
2812$ git-update-index hello.c
2813-------------------------------------------------
2814
2815When a path is in unmerged state, running `git-update-index` for
2816that path tells git to mark the path resolved.
2817
2818The above is the description of a git merge at the lowest level,
2819to help you understand what conceptually happens under the hood.
2820In practice, nobody, not even git itself, uses three `git-cat-file`
2821for this. There is `git-merge-index` program that extracts the
2822stages to temporary files and calls a "merge" script on it:
2823
2824-------------------------------------------------
2825$ git-merge-index git-merge-one-file hello.c
2826-------------------------------------------------
2827
207dfa07 2828and that is what higher level `git merge -s resolve` is implemented with.
b181d57f 2829
e34caace 2830[[pack-files]]
b181d57f
BF
2831How git stores objects efficiently: pack files
2832----------------------------------------------
2833
2834We've seen how git stores each object in a file named after the
2835object's SHA1 hash.
2836
2837Unfortunately this system becomes inefficient once a project has a
2838lot of objects. Try this on an old project:
2839
2840------------------------------------------------
2841$ git count-objects
28426930 objects, 47620 kilobytes
2843------------------------------------------------
2844
2845The first number is the number of objects which are kept in
2846individual files. The second is the amount of space taken up by
2847those "loose" objects.
2848
2849You can save space and make git faster by moving these loose objects in
2850to a "pack file", which stores a group of objects in an efficient
2851compressed format; the details of how pack files are formatted can be
2852found in link:technical/pack-format.txt[technical/pack-format.txt].
2853
2854To put the loose objects into a pack, just run git repack:
2855
2856------------------------------------------------
2857$ git repack
2858Generating pack...
2859Done counting 6020 objects.
2860Deltifying 6020 objects.
2861 100% (6020/6020) done
2862Writing 6020 objects.
2863 100% (6020/6020) done
2864Total 6020, written 6020 (delta 4070), reused 0 (delta 0)
2865Pack pack-3e54ad29d5b2e05838c75df582c65257b8d08e1c created.
2866------------------------------------------------
2867
2868You can then run
2869
2870------------------------------------------------
2871$ git prune
2872------------------------------------------------
2873
2874to remove any of the "loose" objects that are now contained in the
2875pack. This will also remove any unreferenced objects (which may be
2876created when, for example, you use "git reset" to remove a commit).
2877You can verify that the loose objects are gone by looking at the
2878.git/objects directory or by running
2879
2880------------------------------------------------
2881$ git count-objects
28820 objects, 0 kilobytes
2883------------------------------------------------
2884
2885Although the object files are gone, any commands that refer to those
2886objects will work exactly as they did before.
2887
2888The gitlink:git-gc[1] command performs packing, pruning, and more for
2889you, so is normally the only high-level command you need.
d19fbc3c 2890
59723040 2891[[dangling-objects]]
21dcb3b7 2892Dangling objects
61b41790 2893----------------
21dcb3b7 2894
04e50e94 2895The gitlink:git-fsck[1] command will sometimes complain about dangling
21dcb3b7
BF
2896objects. They are not a problem.
2897
1191ee18
BF
2898The most common cause of dangling objects is that you've rebased a
2899branch, or you have pulled from somebody else who rebased a branch--see
2900<<cleaning-up-history>>. In that case, the old head of the original
59723040
BF
2901branch still exists, as does everything it pointed to. The branch
2902pointer itself just doesn't, since you replaced it with another one.
1191ee18 2903
59723040 2904There are also other situations that cause dangling objects. For
1191ee18
BF
2905example, a "dangling blob" may arise because you did a "git add" of a
2906file, but then, before you actually committed it and made it part of the
2907bigger picture, you changed something else in that file and committed
2908that *updated* thing - the old state that you added originally ends up
2909not being pointed to by any commit or tree, so it's now a dangling blob
2910object.
2911
2912Similarly, when the "recursive" merge strategy runs, and finds that
2913there are criss-cross merges and thus more than one merge base (which is
2914fairly unusual, but it does happen), it will generate one temporary
2915midway tree (or possibly even more, if you had lots of criss-crossing
2916merges and more than two merge bases) as a temporary internal merge
2917base, and again, those are real objects, but the end result will not end
2918up pointing to them, so they end up "dangling" in your repository.
2919
2920Generally, dangling objects aren't anything to worry about. They can
2921even be very useful: if you screw something up, the dangling objects can
2922be how you recover your old tree (say, you did a rebase, and realized
2923that you really didn't want to - you can look at what dangling objects
2924you have, and decide to reset your head to some old dangling state).
21dcb3b7 2925
59723040 2926For commits, you can just use:
21dcb3b7
BF
2927
2928------------------------------------------------
2929$ gitk <dangling-commit-sha-goes-here> --not --all
2930------------------------------------------------
2931
59723040
BF
2932This asks for all the history reachable from the given commit but not
2933from any branch, tag, or other reference. If you decide it's something
2934you want, you can always create a new reference to it, e.g.,
2935
2936------------------------------------------------
2937$ git branch recovered-branch <dangling-commit-sha-goes-here>
2938------------------------------------------------
2939
2940For blobs and trees, you can't do the same, but you can still examine
2941them. You can just do
21dcb3b7
BF
2942
2943------------------------------------------------
2944$ git show <dangling-blob/tree-sha-goes-here>
2945------------------------------------------------
2946
1191ee18
BF
2947to show what the contents of the blob were (or, for a tree, basically
2948what the "ls" for that directory was), and that may give you some idea
2949of what the operation was that left that dangling object.
21dcb3b7 2950
1191ee18
BF
2951Usually, dangling blobs and trees aren't very interesting. They're
2952almost always the result of either being a half-way mergebase (the blob
2953will often even have the conflict markers from a merge in it, if you
2954have had conflicting merges that you fixed up by hand), or simply
2955because you interrupted a "git fetch" with ^C or something like that,
2956leaving _some_ of the new objects in the object database, but just
2957dangling and useless.
21dcb3b7
BF
2958
2959Anyway, once you are sure that you're not interested in any dangling
2960state, you can just prune all unreachable objects:
2961
2962------------------------------------------------
2963$ git prune
2964------------------------------------------------
2965
1191ee18
BF
2966and they'll be gone. But you should only run "git prune" on a quiescent
2967repository - it's kind of like doing a filesystem fsck recovery: you
2968don't want to do that while the filesystem is mounted.
21dcb3b7 2969
04e50e94
BF
2970(The same is true of "git-fsck" itself, btw - but since
2971git-fsck never actually *changes* the repository, it just reports
2972on what it found, git-fsck itself is never "dangerous" to run.
21dcb3b7
BF
2973Running it while somebody is actually changing the repository can cause
2974confusing and scary messages, but it won't actually do anything bad. In
2975contrast, running "git prune" while somebody is actively changing the
2976repository is a *BAD* idea).
2977
126640af 2978[[birdview-on-the-source-code]]
a5fc33b4
BF
2979A birds-eye view of Git's source code
2980-------------------------------------
126640af 2981
a5fc33b4
BF
2982It is not always easy for new developers to find their way through Git's
2983source code. This section gives you a little guidance to show where to
2984start.
126640af 2985
a5fc33b4 2986A good place to start is with the contents of the initial commit, with:
126640af
JS
2987
2988----------------------------------------------------
a5fc33b4 2989$ git checkout e83c5163
126640af
JS
2990----------------------------------------------------
2991
a5fc33b4
BF
2992The initial revision lays the foundation for almost everything git has
2993today, but is small enough to read in one sitting.
126640af 2994
a5fc33b4
BF
2995Note that terminology has changed since that revision. For example, the
2996README in that revision uses the word "changeset" to describe what we
2997now call a <<def_commit_object,commit>>.
126640af 2998
a5fc33b4 2999Also, we do not call it "cache" any more, but "index", however, the
126640af
JS
3000file is still called `cache.h`. Remark: Not much reason to change it now,
3001especially since there is no good single name for it anyway, because it is
3002basically _the_ header file which is included by _all_ of Git's C sources.
3003
a5fc33b4
BF
3004If you grasp the ideas in that initial commit, you should check out a
3005more recent version and skim `cache.h`, `object.h` and `commit.h`.
126640af
JS
3006
3007In the early days, Git (in the tradition of UNIX) was a bunch of programs
3008which were extremely simple, and which you used in scripts, piping the
3009output of one into another. This turned out to be good for initial
3010development, since it was easier to test new things. However, recently
3011many of these parts have become builtins, and some of the core has been
3012"libified", i.e. put into libgit.a for performance, portability reasons,
3013and to avoid code duplication.
3014
3015By now, you know what the index is (and find the corresponding data
3016structures in `cache.h`), and that there are just a couple of object types
3017(blobs, trees, commits and tags) which inherit their common structure from
3018`struct object`, which is their first member (and thus, you can cast e.g.
3019`(struct object *)commit` to achieve the _same_ as `&commit->object`, i.e.
3020get at the object name and flags).
3021
3022Now is a good point to take a break to let this information sink in.
3023
3024Next step: get familiar with the object naming. Read <<naming-commits>>.
3025There are quite a few ways to name an object (and not only revisions!).
3026All of these are handled in `sha1_name.c`. Just have a quick look at
3027the function `get_sha1()`. A lot of the special handling is done by
3028functions like `get_sha1_basic()` or the likes.
3029
3030This is just to get you into the groove for the most libified part of Git:
3031the revision walker.
3032
3033Basically, the initial version of `git log` was a shell script:
3034
3035----------------------------------------------------------------
3036$ git-rev-list --pretty $(git-rev-parse --default HEAD "$@") | \
3037 LESS=-S ${PAGER:-less}
3038----------------------------------------------------------------
3039
3040What does this mean?
3041
3042`git-rev-list` is the original version of the revision walker, which
3043_always_ printed a list of revisions to stdout. It is still functional,
3044and needs to, since most new Git programs start out as scripts using
3045`git-rev-list`.
3046
3047`git-rev-parse` is not as important any more; it was only used to filter out
3048options that were relevant for the different plumbing commands that were
3049called by the script.
3050
3051Most of what `git-rev-list` did is contained in `revision.c` and
3052`revision.h`. It wraps the options in a struct named `rev_info`, which
3053controls how and what revisions are walked, and more.
3054
3055The original job of `git-rev-parse` is now taken by the function
3056`setup_revisions()`, which parses the revisions and the common command line
3057options for the revision walker. This information is stored in the struct
3058`rev_info` for later consumption. You can do your own command line option
3059parsing after calling `setup_revisions()`. After that, you have to call
3060`prepare_revision_walk()` for initialization, and then you can get the
3061commits one by one with the function `get_revision()`.
3062
3063If you are interested in more details of the revision walking process,
3064just have a look at the first implementation of `cmd_log()`; call
3065`git-show v1.3.0~155^2~4` and scroll down to that function (note that you
3066no longer need to call `setup_pager()` directly).
3067
3068Nowadays, `git log` is a builtin, which means that it is _contained_ in the
3069command `git`. The source side of a builtin is
3070
3071- a function called `cmd_<bla>`, typically defined in `builtin-<bla>.c`,
3072 and declared in `builtin.h`,
3073
3074- an entry in the `commands[]` array in `git.c`, and
3075
3076- an entry in `BUILTIN_OBJECTS` in the `Makefile`.
3077
3078Sometimes, more than one builtin is contained in one source file. For
3079example, `cmd_whatchanged()` and `cmd_log()` both reside in `builtin-log.c`,
3080since they share quite a bit of code. In that case, the commands which are
3081_not_ named like the `.c` file in which they live have to be listed in
3082`BUILT_INS` in the `Makefile`.
3083
3084`git log` looks more complicated in C than it does in the original script,
3085but that allows for a much greater flexibility and performance.
3086
3087Here again it is a good point to take a pause.
3088
3089Lesson three is: study the code. Really, it is the best way to learn about
3090the organization of Git (after you know the basic concepts).
3091
3092So, think about something which you are interested in, say, "how can I
3093access a blob just knowing the object name of it?". The first step is to
3094find a Git command with which you can do it. In this example, it is either
3095`git show` or `git cat-file`.
3096
3097For the sake of clarity, let's stay with `git cat-file`, because it
3098
3099- is plumbing, and
3100
3101- was around even in the initial commit (it literally went only through
3102 some 20 revisions as `cat-file.c`, was renamed to `builtin-cat-file.c`
3103 when made a builtin, and then saw less than 10 versions).
3104
3105So, look into `builtin-cat-file.c`, search for `cmd_cat_file()` and look what
3106it does.
3107
3108------------------------------------------------------------------
3109 git_config(git_default_config);
3110 if (argc != 3)
3111 usage("git-cat-file [-t|-s|-e|-p|<type>] <sha1>");
3112 if (get_sha1(argv[2], sha1))
3113 die("Not a valid object name %s", argv[2]);
3114------------------------------------------------------------------
3115
3116Let's skip over the obvious details; the only really interesting part
3117here is the call to `get_sha1()`. It tries to interpret `argv[2]` as an
3118object name, and if it refers to an object which is present in the current
3119repository, it writes the resulting SHA-1 into the variable `sha1`.
3120
3121Two things are interesting here:
3122
3123- `get_sha1()` returns 0 on _success_. This might surprise some new
3124 Git hackers, but there is a long tradition in UNIX to return different
3125 negative numbers in case of different errors -- and 0 on success.
3126
3127- the variable `sha1` in the function signature of `get_sha1()` is `unsigned
a5fc33b4 3128 char \*`, but is actually expected to be a pointer to `unsigned
126640af 3129 char[20]`. This variable will contain the 160-bit SHA-1 of the given
a5fc33b4 3130 commit. Note that whenever a SHA-1 is passed as `unsigned char \*`, it
126640af 3131 is the binary representation, as opposed to the ASCII representation in
a5fc33b4 3132 hex characters, which is passed as `char *`.
126640af
JS
3133
3134You will see both of these things throughout the code.
3135
3136Now, for the meat:
3137
3138-----------------------------------------------------------------------------
3139 case 0:
3140 buf = read_object_with_reference(sha1, argv[1], &size, NULL);
3141-----------------------------------------------------------------------------
3142
3143This is how you read a blob (actually, not only a blob, but any type of
3144object). To know how the function `read_object_with_reference()` actually
3145works, find the source code for it (something like `git grep
3146read_object_with | grep ":[a-z]"` in the git repository), and read
3147the source.
3148
3149To find out how the result can be used, just read on in `cmd_cat_file()`:
3150
3151-----------------------------------
3152 write_or_die(1, buf, size);
3153-----------------------------------
3154
3155Sometimes, you do not know where to look for a feature. In many such cases,
3156it helps to search through the output of `git log`, and then `git show` the
3157corresponding commit.
3158
3159Example: If you know that there was some test case for `git bundle`, but
3160do not remember where it was (yes, you _could_ `git grep bundle t/`, but that
3161does not illustrate the point!):
3162
3163------------------------
3164$ git log --no-merges t/
3165------------------------
3166
3167In the pager (`less`), just search for "bundle", go a few lines back,
3168and see that it is in commit 18449ab0... Now just copy this object name,
3169and paste it into the command line
3170
3171-------------------
3172$ git show 18449ab0
3173-------------------
3174
3175Voila.
3176
3177Another example: Find out what to do in order to make some script a
3178builtin:
3179
3180-------------------------------------------------
3181$ git log --no-merges --diff-filter=A builtin-*.c
3182-------------------------------------------------
3183
3184You see, Git is actually the best tool to find out about the source of Git
3185itself!
3186
e34caace 3187[[glossary]]
d19fbc3c
BF
3188include::glossary.txt[]
3189
2624d9a5
BF
3190[[git-quick-start]]
3191Appendix A: Git Quick Start
3192===========================
3193
3194This is a quick summary of the major commands; the following chapters
3195will explain how these work in more detail.
3196
3197[[quick-creating-a-new-repository]]
3198Creating a new repository
3199-------------------------
3200
3201From a tarball:
3202
3203-----------------------------------------------
3204$ tar xzf project.tar.gz
3205$ cd project
3206$ git init
3207Initialized empty Git repository in .git/
3208$ git add .
3209$ git commit
3210-----------------------------------------------
3211
3212From a remote repository:
3213
3214-----------------------------------------------
3215$ git clone git://example.com/pub/project.git
3216$ cd project
3217-----------------------------------------------
3218
3219[[managing-branches]]
3220Managing branches
3221-----------------
3222
3223-----------------------------------------------
3224$ git branch # list all local branches in this repo
3225$ git checkout test # switch working directory to branch "test"
3226$ git branch new # create branch "new" starting at current HEAD
3227$ git branch -d new # delete branch "new"
3228-----------------------------------------------
3229
3230Instead of basing new branch on current HEAD (the default), use:
3231
3232-----------------------------------------------
3233$ git branch new test # branch named "test"
3234$ git branch new v2.6.15 # tag named v2.6.15
3235$ git branch new HEAD^ # commit before the most recent
3236$ git branch new HEAD^^ # commit before that
3237$ git branch new test~10 # ten commits before tip of branch "test"
3238-----------------------------------------------
3239
3240Create and switch to a new branch at the same time:
3241
3242-----------------------------------------------
3243$ git checkout -b new v2.6.15
3244-----------------------------------------------
3245
3246Update and examine branches from the repository you cloned from:
3247
3248-----------------------------------------------
3249$ git fetch # update
3250$ git branch -r # list
3251 origin/master
3252 origin/next
3253 ...
3254$ git checkout -b masterwork origin/master
3255-----------------------------------------------
3256
3257Fetch a branch from a different repository, and give it a new
3258name in your repository:
3259
3260-----------------------------------------------
3261$ git fetch git://example.com/project.git theirbranch:mybranch
3262$ git fetch git://example.com/project.git v2.6.15:mybranch
3263-----------------------------------------------
3264
3265Keep a list of repositories you work with regularly:
3266
3267-----------------------------------------------
3268$ git remote add example git://example.com/project.git
3269$ git remote # list remote repositories
3270example
3271origin
3272$ git remote show example # get details
3273* remote example
3274 URL: git://example.com/project.git
3275 Tracked remote branches
3276 master next ...
3277$ git fetch example # update branches from example
3278$ git branch -r # list all remote branches
3279-----------------------------------------------
3280
3281
3282[[exploring-history]]
3283Exploring history
3284-----------------
3285
3286-----------------------------------------------
3287$ gitk # visualize and browse history
3288$ git log # list all commits
3289$ git log src/ # ...modifying src/
3290$ git log v2.6.15..v2.6.16 # ...in v2.6.16, not in v2.6.15
3291$ git log master..test # ...in branch test, not in branch master
3292$ git log test..master # ...in branch master, but not in test
3293$ git log test...master # ...in one branch, not in both
3294$ git log -S'foo()' # ...where difference contain "foo()"
3295$ git log --since="2 weeks ago"
3296$ git log -p # show patches as well
3297$ git show # most recent commit
3298$ git diff v2.6.15..v2.6.16 # diff between two tagged versions
3299$ git diff v2.6.15..HEAD # diff with current head
3300$ git grep "foo()" # search working directory for "foo()"
3301$ git grep v2.6.15 "foo()" # search old tree for "foo()"
3302$ git show v2.6.15:a.txt # look at old version of a.txt
3303-----------------------------------------------
3304
3305Search for regressions:
3306
3307-----------------------------------------------
3308$ git bisect start
3309$ git bisect bad # current version is bad
3310$ git bisect good v2.6.13-rc2 # last known good revision
3311Bisecting: 675 revisions left to test after this
3312 # test here, then:
3313$ git bisect good # if this revision is good, or
3314$ git bisect bad # if this revision is bad.
3315 # repeat until done.
3316-----------------------------------------------
3317
3318[[making-changes]]
3319Making changes
3320--------------
3321
3322Make sure git knows who to blame:
3323
3324------------------------------------------------
3325$ cat >>~/.gitconfig <<\EOF
3326[user]
3327 name = Your Name Comes Here
3328 email = you@yourdomain.example.com
3329EOF
3330------------------------------------------------
3331
3332Select file contents to include in the next commit, then make the
3333commit:
3334
3335-----------------------------------------------
3336$ git add a.txt # updated file
3337$ git add b.txt # new file
3338$ git rm c.txt # old file
3339$ git commit
3340-----------------------------------------------
3341
3342Or, prepare and create the commit in one step:
3343
3344-----------------------------------------------
3345$ git commit d.txt # use latest content only of d.txt
3346$ git commit -a # use latest content of all tracked files
3347-----------------------------------------------
3348
3349[[merging]]
3350Merging
3351-------
3352
3353-----------------------------------------------
3354$ git merge test # merge branch "test" into the current branch
3355$ git pull git://example.com/project.git master
3356 # fetch and merge in remote branch
3357$ git pull . test # equivalent to git merge test
3358-----------------------------------------------
3359
3360[[sharing-your-changes]]
3361Sharing your changes
3362--------------------
3363
3364Importing or exporting patches:
3365
3366-----------------------------------------------
3367$ git format-patch origin..HEAD # format a patch for each commit
3368 # in HEAD but not in origin
3369$ git am mbox # import patches from the mailbox "mbox"
3370-----------------------------------------------
3371
3372Fetch a branch in a different git repository, then merge into the
3373current branch:
3374
3375-----------------------------------------------
3376$ git pull git://example.com/project.git theirbranch
3377-----------------------------------------------
3378
3379Store the fetched branch into a local branch before merging into the
3380current branch:
3381
3382-----------------------------------------------
3383$ git pull git://example.com/project.git theirbranch:mybranch
3384-----------------------------------------------
3385
3386After creating commits on a local branch, update the remote
3387branch with your commits:
3388
3389-----------------------------------------------
3390$ git push ssh://example.com/project.git mybranch:theirbranch
3391-----------------------------------------------
3392
3393When remote and local branch are both named "test":
3394
3395-----------------------------------------------
3396$ git push ssh://example.com/project.git test
3397-----------------------------------------------
3398
3399Shortcut version for a frequently used remote repository:
3400
3401-----------------------------------------------
3402$ git remote add example ssh://example.com/project.git
3403$ git push example test
3404-----------------------------------------------
3405
3406[[repository-maintenance]]
3407Repository maintenance
3408----------------------
3409
3410Check for corruption:
3411
3412-----------------------------------------------
3413$ git fsck
3414-----------------------------------------------
3415
3416Recompress, remove unused cruft:
3417
3418-----------------------------------------------
3419$ git gc
3420-----------------------------------------------
3421
3422
e34caace 3423[[todo]]
2624d9a5
BF
3424Appendix B: Notes and todo list for this manual
3425===============================================
6bd9b682
BF
3426
3427This is a work in progress.
3428
3429The basic requirements:
2f99710c
BF
3430 - It must be readable in order, from beginning to end, by
3431 someone intelligent with a basic grasp of the unix
3432 commandline, but without any special knowledge of git. If
3433 necessary, any other prerequisites should be specifically
3434 mentioned as they arise.
3435 - Whenever possible, section headings should clearly describe
3436 the task they explain how to do, in language that requires
3437 no more knowledge than necessary: for example, "importing
3438 patches into a project" rather than "the git-am command"
6bd9b682 3439
d5cd5de4
BF
3440Think about how to create a clear chapter dependency graph that will
3441allow people to get to important topics without necessarily reading
3442everything in between.
d19fbc3c 3443
aacd404e
MC
3444Say something about .gitignore.
3445
d19fbc3c
BF
3446Scan Documentation/ for other stuff left out; in particular:
3447 howto's
d19fbc3c
BF
3448 some of technical/?
3449 hooks
0b375ab0 3450 list of commands in gitlink:git[1]
d19fbc3c
BF
3451
3452Scan email archives for other stuff left out
3453
3454Scan man pages to see if any assume more background than this manual
3455provides.
3456
2f99710c 3457Simplify beginning by suggesting disconnected head instead of
b181d57f 3458temporary branch creation?
d19fbc3c 3459
2f99710c
BF
3460Add more good examples. Entire sections of just cookbook examples
3461might be a good idea; maybe make an "advanced examples" section a
3462standard end-of-chapter section?
d19fbc3c
BF
3463
3464Include cross-references to the glossary, where appropriate.
3465
9a241220
BF
3466Document shallow clones? See draft 1.5.0 release notes for some
3467documentation.
3468
3dff5379 3469Add a section on working with other version control systems, including
9a241220
BF
3470CVS, Subversion, and just imports of series of release tarballs.
3471
a8cd1402 3472More details on gitweb?
0b375ab0
BF
3473
3474Write a chapter on using plumbing and writing scripts.