user-manual: fix inconsistent example
[git/git.git] / Documentation / user-manual.txt
CommitLineData
d19fbc3c
BF
1Git User's Manual
2_________________
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
ef89f701 7Chapter 1 gives a brief overview of git commands, without any
b181d57f 8explanation; you may prefer to skip to chapter 2 on a first reading.
ef89f701
BF
9
10Chapters 2 and 3 explain how to fetch and study a project using
d5cd5de4
BF
11git--the tools you'd need to build and test a particular version of a
12software project, to search for regressions, and so on.
6bd9b682 13
ef89f701 14Chapter 4 explains how to do development with git, and chapter 5 how
d5cd5de4 15to share that development with others.
6bd9b682
BF
16
17Further chapters cover more specialized topics.
18
d19fbc3c
BF
19Comprehensive reference documentation is available through the man
20pages. For a command such as "git clone", just use
21
22------------------------------------------------
23$ man git-clone
24------------------------------------------------
25
ef89f701
BF
26Git Quick Start
27===============
28
29This is a quick summary of the major commands; the following chapters
30will explain how these work in more detail.
31
32Creating a new repository
33-------------------------
34
35From a tarball:
36
37-----------------------------------------------
38$ tar xzf project.tar.gz
39$ cd project
40$ git init
41Initialized empty Git repository in .git/
42$ git add .
43$ git commit
44-----------------------------------------------
45
46From a remote repository:
47
48-----------------------------------------------
49$ git clone git://example.com/pub/project.git
50$ cd project
51-----------------------------------------------
52
53Managing branches
54-----------------
55
56-----------------------------------------------
57$ git branch # list all branches in this repo
58$ git checkout test # switch working directory to branch "test"
59$ git branch new # create branch "new" starting at current HEAD
60$ git branch -d new # delete branch "new"
61-----------------------------------------------
62
63Instead of basing new branch on current HEAD (the default), use:
64
65-----------------------------------------------
66$ git branch new test # branch named "test"
67$ git branch new v2.6.15 # tag named v2.6.15
68$ git branch new HEAD^ # commit before the most recent
69$ git branch new HEAD^^ # commit before that
70$ git branch new test~10 # ten commits before tip of branch "test"
71-----------------------------------------------
72
73Create and switch to a new branch at the same time:
74
75-----------------------------------------------
76$ git checkout -b new v2.6.15
77-----------------------------------------------
78
79Update and examine branches from the repository you cloned from:
80
81-----------------------------------------------
82$ git fetch # update
83$ git branch -r # list
84 origin/master
85 origin/next
86 ...
87$ git branch checkout -b masterwork origin/master
88-----------------------------------------------
89
90Fetch a branch from a different repository, and give it a new
91name in your repository:
92
93-----------------------------------------------
94$ git fetch git://example.com/project.git theirbranch:mybranch
95$ git fetch git://example.com/project.git v2.6.15:mybranch
96-----------------------------------------------
97
98Keep a list of repositories you work with regularly:
99
100-----------------------------------------------
101$ git remote add example git://example.com/project.git
b181d57f 102$ git remote # list remote repositories
ef89f701
BF
103example
104origin
b181d57f 105$ git remote show example # get details
ef89f701
BF
106* remote example
107 URL: git://example.com/project.git
108 Tracked remote branches
109 master next ...
b181d57f
BF
110$ git fetch example # update branches from example
111$ git branch -r # list all remote branches
ef89f701
BF
112-----------------------------------------------
113
114
115Exploring history
116-----------------
117
118-----------------------------------------------
119$ gitk # visualize and browse history
120$ git log # list all commits
121$ git log src/ # ...modifying src/
122$ git log v2.6.15..v2.6.16 # ...in v2.6.16, not in v2.6.15
123$ git log master..test # ...in branch test, not in branch master
124$ git log test..master # ...in branch master, but not in test
125$ git log test...master # ...in one branch, not in both
126$ git log -S'foo()' # ...where difference contain "foo()"
127$ git log --since="2 weeks ago"
128$ git log -p # show patches as well
129$ git show # most recent commit
130$ git diff v2.6.15..v2.6.16 # diff between two tagged versions
131$ git diff v2.6.15..HEAD # diff with current head
132$ git grep "foo()" # search working directory for "foo()"
133$ git grep v2.6.15 "foo()" # search old tree for "foo()"
134$ git show v2.6.15:a.txt # look at old version of a.txt
135-----------------------------------------------
136
b181d57f 137Search for regressions:
ef89f701
BF
138
139-----------------------------------------------
140$ git bisect start
141$ git bisect bad # current version is bad
142$ git bisect good v2.6.13-rc2 # last known good revision
143Bisecting: 675 revisions left to test after this
144 # test here, then:
145$ git bisect good # if this revision is good, or
146$ git bisect bad # if this revision is bad.
147 # repeat until done.
148-----------------------------------------------
149
150Making changes
151--------------
152
153Make sure git knows who to blame:
154
155------------------------------------------------
156$ cat >~/.gitconfig <<\EOF
157[user]
158name = Your Name Comes Here
159email = you@yourdomain.example.com
160EOF
161------------------------------------------------
162
163Select file contents to include in the next commit, then make the
164commit:
165
166-----------------------------------------------
167$ git add a.txt # updated file
168$ git add b.txt # new file
169$ git rm c.txt # old file
170$ git commit
171-----------------------------------------------
172
173Or, prepare and create the commit in one step:
174
175-----------------------------------------------
b181d57f 176$ git commit d.txt # use latest content only of d.txt
ef89f701
BF
177$ git commit -a # use latest content of all tracked files
178-----------------------------------------------
179
180Merging
181-------
182
183-----------------------------------------------
184$ git merge test # merge branch "test" into the current branch
185$ git pull git://example.com/project.git master
186 # fetch and merge in remote branch
187$ git pull . test # equivalent to git merge test
188-----------------------------------------------
189
e4add70c
BF
190Sharing your changes
191--------------------
ef89f701
BF
192
193Importing or exporting patches:
194
195-----------------------------------------------
196$ git format-patch origin..HEAD # format a patch for each commit
197 # in HEAD but not in origin
198$ git-am mbox # import patches from the mailbox "mbox"
199-----------------------------------------------
200
ef89f701
BF
201Fetch a branch in a different git repository, then merge into the
202current branch:
203
204-----------------------------------------------
205$ git pull git://example.com/project.git theirbranch
206-----------------------------------------------
207
208Store the fetched branch into a local branch before merging into the
209current branch:
210
211-----------------------------------------------
212$ git pull git://example.com/project.git theirbranch:mybranch
213-----------------------------------------------
214
e4add70c
BF
215After creating commits on a local branch, update the remote
216branch with your commits:
217
218-----------------------------------------------
219$ git push ssh://example.com/project.git mybranch:theirbranch
220-----------------------------------------------
221
222When remote and local branch are both named "test":
223
224-----------------------------------------------
225$ git push ssh://example.com/project.git test
226-----------------------------------------------
227
228Shortcut version for a frequently used remote repository:
229
230-----------------------------------------------
231$ git remote add example ssh://example.com/project.git
232$ git push example test
233-----------------------------------------------
234
b181d57f
BF
235Repository maintenance
236----------------------
237
238Check for corruption:
239
240-----------------------------------------------
04e50e94 241$ git fsck
b181d57f
BF
242-----------------------------------------------
243
244Recompress, remove unused cruft:
245
246-----------------------------------------------
247$ git gc
248-----------------------------------------------
249
d19fbc3c
BF
250Repositories and Branches
251=========================
252
253How to get a git repository
254---------------------------
255
256It will be useful to have a git repository to experiment with as you
257read this manual.
258
259The best way to get one is by using the gitlink:git-clone[1] command
260to download a copy of an existing repository for a project that you
261are interested in. If you don't already have a project in mind, here
262are some interesting examples:
263
264------------------------------------------------
265 # git itself (approx. 10MB download):
266$ git clone git://git.kernel.org/pub/scm/git/git.git
267 # the linux kernel (approx. 150MB download):
268$ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
269------------------------------------------------
270
271The initial clone may be time-consuming for a large project, but you
272will only need to clone once.
273
274The clone command creates a new directory named after the project
275("git" or "linux-2.6" in the examples above). After you cd into this
276directory, you will see that it contains a copy of the project files,
277together with a special top-level directory named ".git", which
278contains all the information about the history of the project.
279
d5cd5de4 280In most of the following, examples will be taken from one of the two
d19fbc3c
BF
281repositories above.
282
283How to check out a different version of a project
284-------------------------------------------------
285
286Git is best thought of as a tool for storing the history of a
287collection of files. It stores the history as a compressed
288collection of interrelated snapshots (versions) of the project's
289contents.
290
291A single git repository may contain multiple branches. Each branch
292is a bookmark referencing a particular point in the project history.
293The gitlink:git-branch[1] command shows you the list of branches:
294
295------------------------------------------------
296$ git branch
297* master
298------------------------------------------------
299
300A freshly cloned repository contains a single branch, named "master",
301and the working directory contains the version of the project
302referred to by the master branch.
303
304Most projects also use tags. Tags, like branches, are references
305into the project's history, and can be listed using the
306gitlink:git-tag[1] command:
307
308------------------------------------------------
309$ git tag -l
310v2.6.11
311v2.6.11-tree
312v2.6.12
313v2.6.12-rc2
314v2.6.12-rc3
315v2.6.12-rc4
316v2.6.12-rc5
317v2.6.12-rc6
318v2.6.13
319...
320------------------------------------------------
321
fe4b3e59
BF
322Tags are expected to always point at the same version of a project,
323while branches are expected to advance as development progresses.
324
d19fbc3c
BF
325Create a new branch pointing to one of these versions and check it
326out using gitlink:git-checkout[1]:
327
328------------------------------------------------
329$ git checkout -b new v2.6.13
330------------------------------------------------
331
332The working directory then reflects the contents that the project had
333when it was tagged v2.6.13, and gitlink:git-branch[1] shows two
334branches, with an asterisk marking the currently checked-out branch:
335
336------------------------------------------------
337$ git branch
338 master
339* new
340------------------------------------------------
341
342If you decide that you'd rather see version 2.6.17, you can modify
343the current branch to point at v2.6.17 instead, with
344
345------------------------------------------------
346$ git reset --hard v2.6.17
347------------------------------------------------
348
349Note that if the current branch was your only reference to a
350particular point in history, then resetting that branch may leave you
351with no way to find the history it used to point to; so use this
352command carefully.
353
354Understanding History: Commits
355------------------------------
356
357Every change in the history of a project is represented by a commit.
358The gitlink:git-show[1] command shows the most recent commit on the
359current branch:
360
361------------------------------------------------
362$ git show
363commit 2b5f6dcce5bf94b9b119e9ed8d537098ec61c3d2
364Author: Jamal Hadi Salim <hadi@cyberus.ca>
365Date: Sat Dec 2 22:22:25 2006 -0800
366
367 [XFRM]: Fix aevent structuring to be more complete.
368
369 aevents can not uniquely identify an SA. We break the ABI with this
370 patch, but consensus is that since it is not yet utilized by any
371 (known) application then it is fine (better do it now than later).
372
373 Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
374 Signed-off-by: David S. Miller <davem@davemloft.net>
375
376diff --git a/Documentation/networking/xfrm_sync.txt b/Documentation/networking/xfrm_sync.txt
377index 8be626f..d7aac9d 100644
378--- a/Documentation/networking/xfrm_sync.txt
379+++ b/Documentation/networking/xfrm_sync.txt
380@@ -47,10 +47,13 @@ aevent_id structure looks like:
381
382 struct xfrm_aevent_id {
383 struct xfrm_usersa_id sa_id;
384+ xfrm_address_t saddr;
385 __u32 flags;
386+ __u32 reqid;
387 };
388...
389------------------------------------------------
390
391As you can see, a commit shows who made the latest change, what they
392did, and why.
393
35121930
BF
394Every commit has a 40-hexdigit id, sometimes called the "object name" or the
395"SHA1 id", shown on the first line of the "git show" output. You can usually
396refer to a commit by a shorter name, such as a tag or a branch name, but this
397longer name can also be useful. Most importantly, it is a globally unique
398name for this commit: so if you tell somebody else the object name (for
399example in email), then you are guaranteed that name will refer to the same
400commit in their repository that it does in yours (assuming their repository
401has that commit at all). Since the object name is computed as a hash over the
402contents of the commit, you are guaranteed that the commit can never change
403without its name also changing.
404
405In fact, in <<git-internals>> we shall see that everything stored in git
406history, including file data and directory contents, is stored in an object
407with a name that is a hash of its contents.
d19fbc3c
BF
408
409Understanding history: commits, parents, and reachability
410~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
411
412Every commit (except the very first commit in a project) also has a
413parent commit which shows what happened before this commit.
414Following the chain of parents will eventually take you back to the
415beginning of the project.
416
417However, the commits do not form a simple list; git allows lines of
418development to diverge and then reconverge, and the point where two
419lines of development reconverge is called a "merge". The commit
420representing a merge can therefore have more than one parent, with
421each parent representing the most recent commit on one of the lines
422of development leading to that point.
423
424The best way to see how this works is using the gitlink:gitk[1]
425command; running gitk now on a git repository and looking for merge
426commits will help understand how the git organizes history.
427
428In the following, we say that commit X is "reachable" from commit Y
429if commit X is an ancestor of commit Y. Equivalently, you could say
430that Y is a descendent of X, or that there is a chain of parents
431leading from commit Y to commit X.
432
3dff5379
PR
433Understanding history: History diagrams
434~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c
BF
435
436We will sometimes represent git history using diagrams like the one
437below. Commits are shown as "o", and the links between them with
438lines drawn with - / and \. Time goes left to right:
439
440 o--o--o <-- Branch A
441 /
442 o--o--o <-- master
443 \
444 o--o--o <-- Branch B
445
446If we need to talk about a particular commit, the character "o" may
447be replaced with another letter or number.
448
449Understanding history: What is a branch?
450~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
451
452Though we've been using the word "branch" to mean a kind of reference
453to a particular commit, the word branch is also commonly used to
454refer to the line of commits leading up to that point. In the
455example above, git may think of the branch named "A" as just a
456pointer to one particular commit, but we may refer informally to the
457line of three commits leading up to that point as all being part of
458"branch A".
459
460If we need to make it clear that we're just talking about the most
461recent commit on the branch, we may refer to that commit as the
462"head" of the branch.
463
464Manipulating branches
465---------------------
466
467Creating, deleting, and modifying branches is quick and easy; here's
468a summary of the commands:
469
470git branch::
471 list all branches
472git branch <branch>::
473 create a new branch named <branch>, referencing the same
474 point in history as the current branch
475git branch <branch> <start-point>::
476 create a new branch named <branch>, referencing
477 <start-point>, which may be specified any way you like,
478 including using a branch name or a tag name
479git branch -d <branch>::
480 delete the branch <branch>; if the branch you are deleting
481 points to a commit which is not reachable from this branch,
482 this command will fail with a warning.
483git branch -D <branch>::
484 even if the branch points to a commit not reachable
485 from the current branch, you may know that that commit
486 is still reachable from some other branch or tag. In that
487 case it is safe to use this command to force git to delete
488 the branch.
489git checkout <branch>::
490 make the current branch <branch>, updating the working
491 directory to reflect the version referenced by <branch>
492git checkout -b <new> <start-point>::
493 create a new branch <new> referencing <start-point>, and
494 check it out.
495
496It is also useful to know that the special symbol "HEAD" can always
497be used to refer to the current branch.
498
499Examining branches from a remote repository
500-------------------------------------------
501
502The "master" branch that was created at the time you cloned is a copy
503of the HEAD in the repository that you cloned from. That repository
504may also have had other branches, though, and your local repository
505keeps branches which track each of those remote branches, which you
506can view using the "-r" option to gitlink:git-branch[1]:
507
508------------------------------------------------
509$ git branch -r
510 origin/HEAD
511 origin/html
512 origin/maint
513 origin/man
514 origin/master
515 origin/next
516 origin/pu
517 origin/todo
518------------------------------------------------
519
520You cannot check out these remote-tracking branches, but you can
521examine them on a branch of your own, just as you would a tag:
522
523------------------------------------------------
524$ git checkout -b my-todo-copy origin/todo
525------------------------------------------------
526
527Note that the name "origin" is just the name that git uses by default
528to refer to the repository that you cloned from.
529
530[[how-git-stores-references]]
f60b9642
BF
531Naming branches, tags, and other references
532-------------------------------------------
d19fbc3c
BF
533
534Branches, remote-tracking branches, and tags are all references to
f60b9642
BF
535commits. All references are named with a slash-separated path name
536starting with "refs"; the names we've been using so far are actually
537shorthand:
d19fbc3c 538
f60b9642
BF
539 - The branch "test" is short for "refs/heads/test".
540 - The tag "v2.6.18" is short for "refs/tags/v2.6.18".
541 - "origin/master" is short for "refs/remotes/origin/master".
d19fbc3c 542
f60b9642
BF
543The full name is occasionally useful if, for example, there ever
544exists a tag and a branch with the same name.
d19fbc3c 545
f60b9642
BF
546As another useful shortcut, if the repository "origin" posesses only
547a single branch, you can refer to that branch as just "origin".
d19fbc3c 548
f60b9642
BF
549More generally, if you have defined a remote repository named
550"example", you can refer to the branch in that repository as
551"example". And for a repository with multiple branches, this will
552refer to the branch designated as the "HEAD" branch.
d19fbc3c
BF
553
554For the complete list of paths which git checks for references, and
f60b9642
BF
555the order it uses to decide which to choose when there are multiple
556references with the same shorthand name, see the "SPECIFYING
557REVISIONS" section of gitlink:git-rev-parse[1].
d19fbc3c
BF
558
559[[Updating-a-repository-with-git-fetch]]
560Updating a repository with git fetch
561------------------------------------
562
563Eventually the developer cloned from will do additional work in her
564repository, creating new commits and advancing the branches to point
565at the new commits.
566
567The command "git fetch", with no arguments, will update all of the
568remote-tracking branches to the latest version found in her
569repository. It will not touch any of your own branches--not even the
570"master" branch that was created for you on clone.
571
d5cd5de4
BF
572Fetching branches from other repositories
573-----------------------------------------
574
575You can also track branches from repositories other than the one you
576cloned from, using gitlink:git-remote[1]:
577
578-------------------------------------------------
579$ git remote add linux-nfs git://linux-nfs.org/pub/nfs-2.6.git
580$ git fetch
581* refs/remotes/linux-nfs/master: storing branch 'master' ...
582 commit: bf81b46
583-------------------------------------------------
584
585New remote-tracking branches will be stored under the shorthand name
586that you gave "git remote add", in this case linux-nfs:
587
588-------------------------------------------------
589$ git branch -r
590linux-nfs/master
591origin/master
592-------------------------------------------------
593
594If you run "git fetch <remote>" later, the tracking branches for the
595named <remote> will be updated.
596
597If you examine the file .git/config, you will see that git has added
598a new stanza:
599
600-------------------------------------------------
601$ cat .git/config
602...
603[remote "linux-nfs"]
923642fe
BF
604 url = git://linux-nfs.org/pub/nfs-2.6.git
605 fetch = +refs/heads/*:refs/remotes/linux-nfs/*
d5cd5de4
BF
606...
607-------------------------------------------------
608
fc90c536
BF
609This is what causes git to track the remote's branches; you may modify
610or delete these configuration options by editing .git/config with a
611text editor. (See the "CONFIGURATION FILE" section of
612gitlink:git-config[1] for details.)
d5cd5de4 613
d19fbc3c
BF
614Exploring git history
615=====================
616
617Git is best thought of as a tool for storing the history of a
618collection of files. It does this by storing compressed snapshots of
619the contents of a file heirarchy, together with "commits" which show
620the relationships between these snapshots.
621
622Git provides extremely flexible and fast tools for exploring the
623history of a project.
624
aacd404e 625We start with one specialized tool that is useful for finding the
d19fbc3c
BF
626commit that introduced a bug into a project.
627
628How to use bisect to find a regression
629--------------------------------------
630
631Suppose version 2.6.18 of your project worked, but the version at
632"master" crashes. Sometimes the best way to find the cause of such a
633regression is to perform a brute-force search through the project's
634history to find the particular commit that caused the problem. The
635gitlink:git-bisect[1] command can help you do this:
636
637-------------------------------------------------
638$ git bisect start
639$ git bisect good v2.6.18
640$ git bisect bad master
641Bisecting: 3537 revisions left to test after this
642[65934a9a028b88e83e2b0f8b36618fe503349f8e] BLOCK: Make USB storage depend on SCSI rather than selecting it [try #6]
643-------------------------------------------------
644
645If you run "git branch" at this point, you'll see that git has
646temporarily moved you to a new branch named "bisect". This branch
647points to a commit (with commit id 65934...) that is reachable from
648v2.6.19 but not from v2.6.18. Compile and test it, and see whether
649it crashes. Assume it does crash. Then:
650
651-------------------------------------------------
652$ git bisect bad
653Bisecting: 1769 revisions left to test after this
654[7eff82c8b1511017ae605f0c99ac275a7e21b867] i2c-core: Drop useless bitmaskings
655-------------------------------------------------
656
657checks out an older version. Continue like this, telling git at each
658stage whether the version it gives you is good or bad, and notice
659that the number of revisions left to test is cut approximately in
660half each time.
661
662After about 13 tests (in this case), it will output the commit id of
663the guilty commit. You can then examine the commit with
664gitlink:git-show[1], find out who wrote it, and mail them your bug
665report with the commit id. Finally, run
666
667-------------------------------------------------
668$ git bisect reset
669-------------------------------------------------
670
671to return you to the branch you were on before and delete the
672temporary "bisect" branch.
673
674Note that the version which git-bisect checks out for you at each
675point is just a suggestion, and you're free to try a different
676version if you think it would be a good idea. For example,
677occasionally you may land on a commit that broke something unrelated;
678run
679
680-------------------------------------------------
681$ git bisect-visualize
682-------------------------------------------------
683
684which will run gitk and label the commit it chose with a marker that
685says "bisect". Chose a safe-looking commit nearby, note its commit
686id, and check it out with:
687
688-------------------------------------------------
689$ git reset --hard fb47ddb2db...
690-------------------------------------------------
691
692then test, run "bisect good" or "bisect bad" as appropriate, and
693continue.
694
695Naming commits
696--------------
697
698We have seen several ways of naming commits already:
699
d55ae921 700 - 40-hexdigit object name
d19fbc3c
BF
701 - branch name: refers to the commit at the head of the given
702 branch
703 - tag name: refers to the commit pointed to by the given tag
704 (we've seen branches and tags are special cases of
705 <<how-git-stores-references,references>>).
706 - HEAD: refers to the head of the current branch
707
eb6ae7f4 708There are many more; see the "SPECIFYING REVISIONS" section of the
aec053bb 709gitlink:git-rev-parse[1] man page for the complete list of ways to
d19fbc3c
BF
710name revisions. Some examples:
711
712-------------------------------------------------
d55ae921 713$ git show fb47ddb2 # the first few characters of the object name
d19fbc3c
BF
714 # are usually enough to specify it uniquely
715$ git show HEAD^ # the parent of the HEAD commit
716$ git show HEAD^^ # the grandparent
717$ git show HEAD~4 # the great-great-grandparent
718-------------------------------------------------
719
720Recall that merge commits may have more than one parent; by default,
721^ and ~ follow the first parent listed in the commit, but you can
722also choose:
723
724-------------------------------------------------
725$ git show HEAD^1 # show the first parent of HEAD
726$ git show HEAD^2 # show the second parent of HEAD
727-------------------------------------------------
728
729In addition to HEAD, there are several other special names for
730commits:
731
732Merges (to be discussed later), as well as operations such as
733git-reset, which change the currently checked-out commit, generally
734set ORIG_HEAD to the value HEAD had before the current operation.
735
736The git-fetch operation always stores the head of the last fetched
737branch in FETCH_HEAD. For example, if you run git fetch without
738specifying a local branch as the target of the operation
739
740-------------------------------------------------
741$ git fetch git://example.com/proj.git theirbranch
742-------------------------------------------------
743
744the fetched commits will still be available from FETCH_HEAD.
745
746When we discuss merges we'll also see the special name MERGE_HEAD,
747which refers to the other branch that we're merging in to the current
748branch.
749
aec053bb 750The gitlink:git-rev-parse[1] command is a low-level command that is
d55ae921
BF
751occasionally useful for translating some name for a commit to the object
752name for that commit:
aec053bb
BF
753
754-------------------------------------------------
755$ git rev-parse origin
756e05db0fd4f31dde7005f075a84f96b360d05984b
757-------------------------------------------------
758
d19fbc3c
BF
759Creating tags
760-------------
761
762We can also create a tag to refer to a particular commit; after
763running
764
765-------------------------------------------------
766$ git-tag stable-1 1b2e1d63ff
767-------------------------------------------------
768
769You can use stable-1 to refer to the commit 1b2e1d63ff.
770
771This creates a "lightweight" tag. If the tag is a tag you wish to
772share with others, and possibly sign cryptographically, then you
773should create a tag object instead; see the gitlink:git-tag[1] man
774page for details.
775
776Browsing revisions
777------------------
778
779The gitlink:git-log[1] command can show lists of commits. On its
780own, it shows all commits reachable from the parent commit; but you
781can also make more specific requests:
782
783-------------------------------------------------
784$ git log v2.5.. # commits since (not reachable from) v2.5
785$ git log test..master # commits reachable from master but not test
786$ git log master..test # ...reachable from test but not master
787$ git log master...test # ...reachable from either test or master,
788 # but not both
789$ git log --since="2 weeks ago" # commits from the last 2 weeks
790$ git log Makefile # commits which modify Makefile
791$ git log fs/ # ... which modify any file under fs/
792$ git log -S'foo()' # commits which add or remove any file data
793 # matching the string 'foo()'
794-------------------------------------------------
795
796And of course you can combine all of these; the following finds
797commits since v2.5 which touch the Makefile or any file under fs:
798
799-------------------------------------------------
800$ git log v2.5.. Makefile fs/
801-------------------------------------------------
802
803You can also ask git log to show patches:
804
805-------------------------------------------------
806$ git log -p
807-------------------------------------------------
808
809See the "--pretty" option in the gitlink:git-log[1] man page for more
810display options.
811
812Note that git log starts with the most recent commit and works
813backwards through the parents; however, since git history can contain
3dff5379 814multiple independent lines of development, the particular order that
d19fbc3c
BF
815commits are listed in may be somewhat arbitrary.
816
817Generating diffs
818----------------
819
820You can generate diffs between any two versions using
821gitlink:git-diff[1]:
822
823-------------------------------------------------
824$ git diff master..test
825-------------------------------------------------
826
827Sometimes what you want instead is a set of patches:
828
829-------------------------------------------------
830$ git format-patch master..test
831-------------------------------------------------
832
833will generate a file with a patch for each commit reachable from test
834but not from master. Note that if master also has commits which are
835not reachable from test, then the combined result of these patches
836will not be the same as the diff produced by the git-diff example.
837
838Viewing old file versions
839-------------------------
840
841You can always view an old version of a file by just checking out the
842correct revision first. But sometimes it is more convenient to be
843able to view an old version of a single file without checking
844anything out; this command does that:
845
846-------------------------------------------------
847$ git show v2.5:fs/locks.c
848-------------------------------------------------
849
850Before the colon may be anything that names a commit, and after it
851may be any path to a file tracked by git.
852
aec053bb
BF
853Examples
854--------
855
856Check whether two branches point at the same history
2f99710c 857~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
aec053bb
BF
858
859Suppose you want to check whether two branches point at the same point
860in history.
861
862-------------------------------------------------
863$ git diff origin..master
864-------------------------------------------------
865
69f7ad73
BF
866will tell you whether the contents of the project are the same at the
867two branches; in theory, however, it's possible that the same project
868contents could have been arrived at by two different historical
d55ae921 869routes. You could compare the object names:
aec053bb
BF
870
871-------------------------------------------------
872$ git rev-list origin
873e05db0fd4f31dde7005f075a84f96b360d05984b
874$ git rev-list master
875e05db0fd4f31dde7005f075a84f96b360d05984b
876-------------------------------------------------
877
69f7ad73
BF
878Or you could recall that the ... operator selects all commits
879contained reachable from either one reference or the other but not
880both: so
aec053bb
BF
881
882-------------------------------------------------
883$ git log origin...master
884-------------------------------------------------
885
886will return no commits when the two branches are equal.
887
b181d57f
BF
888Find first tagged version including a given fix
889~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
aec053bb 890
69f7ad73
BF
891Suppose you know that the commit e05db0fd fixed a certain problem.
892You'd like to find the earliest tagged release that contains that
893fix.
894
895Of course, there may be more than one answer--if the history branched
896after commit e05db0fd, then there could be multiple "earliest" tagged
897releases.
898
899You could just visually inspect the commits since e05db0fd:
900
901-------------------------------------------------
902$ gitk e05db0fd..
903-------------------------------------------------
904
b181d57f
BF
905Or you can use gitlink:git-name-rev[1], which will give the commit a
906name based on any tag it finds pointing to one of the commit's
907descendants:
908
909-------------------------------------------------
910$ git name-rev e05db0fd
911e05db0fd tags/v1.5.0-rc1^0~23
912-------------------------------------------------
913
914The gitlink:git-describe[1] command does the opposite, naming the
915revision using a tag on which the given commit is based:
916
917-------------------------------------------------
918$ git describe e05db0fd
919v1.5.0-rc0-ge05db0f
920-------------------------------------------------
921
922but that may sometimes help you guess which tags might come after the
923given commit.
924
925If you just want to verify whether a given tagged version contains a
926given commit, you could use gitlink:git-merge-base[1]:
927
928-------------------------------------------------
929$ git merge-base e05db0fd v1.5.0-rc1
930e05db0fd4f31dde7005f075a84f96b360d05984b
931-------------------------------------------------
932
933The merge-base command finds a common ancestor of the given commits,
934and always returns one or the other in the case where one is a
935descendant of the other; so the above output shows that e05db0fd
936actually is an ancestor of v1.5.0-rc1.
937
938Alternatively, note that
939
940-------------------------------------------------
4a7979ca 941$ git log v1.5.0-rc1..e05db0fd
b181d57f
BF
942-------------------------------------------------
943
4a7979ca 944will produce empty output if and only if v1.5.0-rc1 includes e05db0fd,
b181d57f 945because it outputs only commits that are not reachable from v1.5.0-rc1.
aec053bb 946
4a7979ca
BF
947As yet another alternative, the gitlink:git-show-branch[1] command lists
948the commits reachable from its arguments with a display on the left-hand
949side that indicates which arguments that commit is reachable from. So,
950you can run something like
951
952-------------------------------------------------
953$ git show-branch e05db0fd v1.5.0-rc0 v1.5.0-rc1 v1.5.0-rc2
954! [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
955available
956 ! [v1.5.0-rc0] GIT v1.5.0 preview
957 ! [v1.5.0-rc1] GIT v1.5.0-rc1
958 ! [v1.5.0-rc2] GIT v1.5.0-rc2
959...
960-------------------------------------------------
961
962then search for a line that looks like
963
964-------------------------------------------------
965+ ++ [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
966available
967-------------------------------------------------
968
969Which shows that e05db0fd is reachable from itself, from v1.5.0-rc1, and
970from v1.5.0-rc2, but not from v1.5.0-rc0.
971
972
d19fbc3c
BF
973Developing with git
974===================
975
976Telling git your name
977---------------------
978
979Before creating any commits, you should introduce yourself to git. The
980easiest way to do so is:
981
982------------------------------------------------
983$ cat >~/.gitconfig <<\EOF
984[user]
985 name = Your Name Comes Here
986 email = you@yourdomain.example.com
987EOF
988------------------------------------------------
989
fc90c536
BF
990(See the "CONFIGURATION FILE" section of gitlink:git-config[1] for
991details on the configuration file.)
992
d19fbc3c
BF
993
994Creating a new repository
995-------------------------
996
997Creating a new repository from scratch is very easy:
998
999-------------------------------------------------
1000$ mkdir project
1001$ cd project
f1d2b477 1002$ git init
d19fbc3c
BF
1003-------------------------------------------------
1004
1005If you have some initial content (say, a tarball):
1006
1007-------------------------------------------------
1008$ tar -xzvf project.tar.gz
1009$ cd project
f1d2b477 1010$ git init
d19fbc3c
BF
1011$ git add . # include everything below ./ in the first commit:
1012$ git commit
1013-------------------------------------------------
1014
1015[[how-to-make-a-commit]]
1016how to make a commit
1017--------------------
1018
1019Creating a new commit takes three steps:
1020
1021 1. Making some changes to the working directory using your
1022 favorite editor.
1023 2. Telling git about your changes.
1024 3. Creating the commit using the content you told git about
1025 in step 2.
1026
1027In practice, you can interleave and repeat steps 1 and 2 as many
1028times as you want: in order to keep track of what you want committed
1029at step 3, git maintains a snapshot of the tree's contents in a
1030special staging area called "the index."
1031
01997b4a
BF
1032At the beginning, the content of the index will be identical to
1033that of the HEAD. The command "git diff --cached", which shows
1034the difference between the HEAD and the index, should therefore
1035produce no output at that point.
eb6ae7f4 1036
d19fbc3c
BF
1037Modifying the index is easy:
1038
1039To update the index with the new contents of a modified file, use
1040
1041-------------------------------------------------
1042$ git add path/to/file
1043-------------------------------------------------
1044
1045To add the contents of a new file to the index, use
1046
1047-------------------------------------------------
1048$ git add path/to/file
1049-------------------------------------------------
1050
eb6ae7f4 1051To remove a file from the index and from the working tree,
d19fbc3c
BF
1052
1053-------------------------------------------------
1054$ git rm path/to/file
1055-------------------------------------------------
1056
1057After each step you can verify that
1058
1059-------------------------------------------------
1060$ git diff --cached
1061-------------------------------------------------
1062
1063always shows the difference between the HEAD and the index file--this
1064is what you'd commit if you created the commit now--and that
1065
1066-------------------------------------------------
1067$ git diff
1068-------------------------------------------------
1069
1070shows the difference between the working tree and the index file.
1071
1072Note that "git add" always adds just the current contents of a file
1073to the index; further changes to the same file will be ignored unless
1074you run git-add on the file again.
1075
1076When you're ready, just run
1077
1078-------------------------------------------------
1079$ git commit
1080-------------------------------------------------
1081
1082and git will prompt you for a commit message and then create the new
3dff5379 1083commit. Check to make sure it looks like what you expected with
d19fbc3c
BF
1084
1085-------------------------------------------------
1086$ git show
1087-------------------------------------------------
1088
1089As a special shortcut,
1090
1091-------------------------------------------------
1092$ git commit -a
1093-------------------------------------------------
1094
1095will update the index with any files that you've modified or removed
1096and create a commit, all in one step.
1097
1098A number of commands are useful for keeping track of what you're
1099about to commit:
1100
1101-------------------------------------------------
1102$ git diff --cached # difference between HEAD and the index; what
1103 # would be commited if you ran "commit" now.
1104$ git diff # difference between the index file and your
1105 # working directory; changes that would not
1106 # be included if you ran "commit" now.
1107$ git status # a brief per-file summary of the above.
1108-------------------------------------------------
1109
1110creating good commit messages
1111-----------------------------
1112
1113Though not required, it's a good idea to begin the commit message
1114with a single short (less than 50 character) line summarizing the
1115change, followed by a blank line and then a more thorough
1116description. Tools that turn commits into email, for example, use
1117the first line on the Subject line and the rest of the commit in the
1118body.
1119
1120how to merge
1121------------
1122
1123You can rejoin two diverging branches of development using
1124gitlink:git-merge[1]:
1125
1126-------------------------------------------------
1127$ git merge branchname
1128-------------------------------------------------
1129
1130merges the development in the branch "branchname" into the current
1131branch. If there are conflicts--for example, if the same file is
1132modified in two different ways in the remote branch and the local
1133branch--then you are warned; the output may look something like this:
1134
1135-------------------------------------------------
1136$ git pull . next
1137Trying really trivial in-index merge...
1138fatal: Merge requires file-level merging
1139Nope.
1140Merging HEAD with 77976da35a11db4580b80ae27e8d65caf5208086
1141Merging:
114215e2162 world
114377976da goodbye
1144found 1 common ancestor(s):
1145d122ed4 initial
1146Auto-merging file.txt
1147CONFLICT (content): Merge conflict in file.txt
1148Automatic merge failed; fix conflicts and then commit the result.
1149-------------------------------------------------
1150
1151Conflict markers are left in the problematic files, and after
1152you resolve the conflicts manually, you can update the index
1153with the contents and run git commit, as you normally would when
1154creating a new file.
1155
1156If you examine the resulting commit using gitk, you will see that it
1157has two parents, one pointing to the top of the current branch, and
1158one to the top of the other branch.
1159
1160In more detail:
1161
1162[[resolving-a-merge]]
1163Resolving a merge
1164-----------------
1165
1166When a merge isn't resolved automatically, git leaves the index and
1167the working tree in a special state that gives you all the
1168information you need to help resolve the merge.
1169
1170Files with conflicts are marked specially in the index, so until you
ef561ac7
BF
1171resolve the problem and update the index, gitlink:git-commit[1] will
1172fail:
d19fbc3c
BF
1173
1174-------------------------------------------------
1175$ git commit
1176file.txt: needs merge
1177-------------------------------------------------
1178
ef561ac7
BF
1179Also, gitlink:git-status[1] will list those files as "unmerged", and the
1180files with conflicts will have conflict markers added, like this:
1181
1182-------------------------------------------------
1183<<<<<<< HEAD:file.txt
1184Hello world
1185=======
1186Goodbye
1187>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1188-------------------------------------------------
1189
1190All you need to do is edit the files to resolve the conflicts, and then
1191
1192-------------------------------------------------
1193$ git add file.txt
1194$ git commit
1195-------------------------------------------------
1196
1197Note that the commit message will already be filled in for you with
1198some information about the merge. Normally you can just use this
1199default message unchanged, but you may add additional commentary of
1200your own if desired.
1201
1202The above is all you need to know to resolve a simple merge. But git
1203also provides more information to help resolve conflicts:
1204
1205Getting conflict-resolution help during a merge
1206~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c
BF
1207
1208All of the changes that git was able to merge automatically are
1209already added to the index file, so gitlink:git-diff[1] shows only
ef561ac7 1210the conflicts. It uses an unusual syntax:
d19fbc3c
BF
1211
1212-------------------------------------------------
1213$ git diff
1214diff --cc file.txt
1215index 802992c,2b60207..0000000
1216--- a/file.txt
1217+++ b/file.txt
1218@@@ -1,1 -1,1 +1,5 @@@
1219++<<<<<<< HEAD:file.txt
1220 +Hello world
1221++=======
1222+ Goodbye
1223++>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1224-------------------------------------------------
1225
1226Recall that the commit which will be commited after we resolve this
1227conflict will have two parents instead of the usual one: one parent
1228will be HEAD, the tip of the current branch; the other will be the
1229tip of the other branch, which is stored temporarily in MERGE_HEAD.
1230
ef561ac7
BF
1231During the merge, the index holds three versions of each file. Each of
1232these three "file stages" represents a different version of the file:
1233
1234-------------------------------------------------
1235$ git show :1:file.txt # the file in a common ancestor of both branches
1236$ git show :2:file.txt # the version from HEAD, but including any
1237 # nonconflicting changes from MERGE_HEAD
1238$ git show :3:file.txt # the version from MERGE_HEAD, but including any
1239 # nonconflicting changes from HEAD.
1240-------------------------------------------------
1241
1242Since the stage 2 and stage 3 versions have already been updated with
1243nonconflicting changes, the only remaining differences between them are
1244the important ones; thus gitlink:git-diff[1] can use the information in
1245the index to show only those conflicts.
1246
1247The diff above shows the differences between the working-tree version of
1248file.txt and the stage 2 and stage 3 versions. So instead of preceding
1249each line by a single "+" or "-", it now uses two columns: the first
1250column is used for differences between the first parent and the working
1251directory copy, and the second for differences between the second parent
1252and the working directory copy. (See the "COMBINED DIFF FORMAT" section
1253of gitlink:git-diff-files[1] for a details of the format.)
1254
1255After resolving the conflict in the obvious way (but before updating the
1256index), the diff will look like:
d19fbc3c
BF
1257
1258-------------------------------------------------
1259$ git diff
1260diff --cc file.txt
1261index 802992c,2b60207..0000000
1262--- a/file.txt
1263+++ b/file.txt
1264@@@ -1,1 -1,1 +1,1 @@@
1265- Hello world
1266 -Goodbye
1267++Goodbye world
1268-------------------------------------------------
1269
1270This shows that our resolved version deleted "Hello world" from the
1271first parent, deleted "Goodbye" from the second parent, and added
1272"Goodbye world", which was previously absent from both.
1273
ef561ac7
BF
1274Some special diff options allow diffing the working directory against
1275any of these stages:
1276
1277-------------------------------------------------
1278$ git diff -1 file.txt # diff against stage 1
1279$ git diff --base file.txt # same as the above
1280$ git diff -2 file.txt # diff against stage 2
1281$ git diff --ours file.txt # same as the above
1282$ git diff -3 file.txt # diff against stage 3
1283$ git diff --theirs file.txt # same as the above.
1284-------------------------------------------------
1285
1286The gitlink:git-log[1] and gitk[1] commands also provide special help
1287for merges:
d19fbc3c
BF
1288
1289-------------------------------------------------
1290$ git log --merge
ef561ac7 1291$ gitk --merge
d19fbc3c
BF
1292-------------------------------------------------
1293
ef561ac7
BF
1294These will display all commits which exist only on HEAD or on
1295MERGE_HEAD, and which touch an unmerged file.
d19fbc3c 1296
ef561ac7 1297Each time you resolve the conflicts in a file and update the index:
d19fbc3c
BF
1298
1299-------------------------------------------------
1300$ git add file.txt
d19fbc3c
BF
1301-------------------------------------------------
1302
ef561ac7
BF
1303the different stages of that file will be "collapsed", after which
1304git-diff will (by default) no longer show diffs for that file.
d19fbc3c
BF
1305
1306[[undoing-a-merge]]
1307undoing a merge
1308---------------
1309
1310If you get stuck and decide to just give up and throw the whole mess
1311away, you can always return to the pre-merge state with
1312
1313-------------------------------------------------
1314$ git reset --hard HEAD
1315-------------------------------------------------
1316
1317Or, if you've already commited the merge that you want to throw away,
1318
1319-------------------------------------------------
1c73bb0e 1320$ git reset --hard ORIG_HEAD
d19fbc3c
BF
1321-------------------------------------------------
1322
1323However, this last command can be dangerous in some cases--never
1324throw away a commit you have already committed if that commit may
1325itself have been merged into another branch, as doing so may confuse
1326further merges.
1327
1328Fast-forward merges
1329-------------------
1330
1331There is one special case not mentioned above, which is treated
1332differently. Normally, a merge results in a merge commit, with two
1333parents, one pointing at each of the two lines of development that
1334were merged.
1335
1336However, if one of the two lines of development is completely
1337contained within the other--so every commit present in the one is
1338already contained in the other--then git just performs a
1339<<fast-forwards,fast forward>>; the head of the current branch is
1340moved forward to point at the head of the merged-in branch, without
1341any new commits being created.
1342
b684f830
BF
1343Fixing mistakes
1344---------------
1345
1346If you've messed up the working tree, but haven't yet committed your
1347mistake, you can return the entire working tree to the last committed
1348state with
1349
1350-------------------------------------------------
1351$ git reset --hard HEAD
1352-------------------------------------------------
1353
1354If you make a commit that you later wish you hadn't, there are two
1355fundamentally different ways to fix the problem:
1356
1357 1. You can create a new commit that undoes whatever was done
1358 by the previous commit. This is the correct thing if your
1359 mistake has already been made public.
1360
1361 2. You can go back and modify the old commit. You should
1362 never do this if you have already made the history public;
1363 git does not normally expect the "history" of a project to
1364 change, and cannot correctly perform repeated merges from
1365 a branch that has had its history changed.
1366
1367Fixing a mistake with a new commit
1368~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1369
1370Creating a new commit that reverts an earlier change is very easy;
1371just pass the gitlink:git-revert[1] command a reference to the bad
1372commit; for example, to revert the most recent commit:
1373
1374-------------------------------------------------
1375$ git revert HEAD
1376-------------------------------------------------
1377
1378This will create a new commit which undoes the change in HEAD. You
1379will be given a chance to edit the commit message for the new commit.
1380
1381You can also revert an earlier change, for example, the next-to-last:
1382
1383-------------------------------------------------
1384$ git revert HEAD^
1385-------------------------------------------------
1386
1387In this case git will attempt to undo the old change while leaving
1388intact any changes made since then. If more recent changes overlap
1389with the changes to be reverted, then you will be asked to fix
1390conflicts manually, just as in the case of <<resolving-a-merge,
1391resolving a merge>>.
1392
365aa199 1393[[fixing-a-mistake-by-editing-history]]
b684f830
BF
1394Fixing a mistake by editing history
1395~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1396
1397If the problematic commit is the most recent commit, and you have not
1398yet made that commit public, then you may just
1399<<undoing-a-merge,destroy it using git-reset>>.
1400
1401Alternatively, you
1402can edit the working directory and update the index to fix your
1403mistake, just as if you were going to <<how-to-make-a-commit,create a
1404new commit>>, then run
1405
1406-------------------------------------------------
1407$ git commit --amend
1408-------------------------------------------------
1409
1410which will replace the old commit by a new commit incorporating your
1411changes, giving you a chance to edit the old commit message first.
1412
1413Again, you should never do this to a commit that may already have
1414been merged into another branch; use gitlink:git-revert[1] instead in
1415that case.
1416
1417It is also possible to edit commits further back in the history, but
1418this is an advanced topic to be left for
1419<<cleaning-up-history,another chapter>>.
1420
1421Checking out an old version of a file
1422~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1423
1424In the process of undoing a previous bad change, you may find it
1425useful to check out an older version of a particular file using
1426gitlink:git-checkout[1]. We've used git checkout before to switch
1427branches, but it has quite different behavior if it is given a path
1428name: the command
1429
1430-------------------------------------------------
1431$ git checkout HEAD^ path/to/file
1432-------------------------------------------------
1433
1434replaces path/to/file by the contents it had in the commit HEAD^, and
1435also updates the index to match. It does not change branches.
1436
1437If you just want to look at an old version of the file, without
1438modifying the working directory, you can do that with
1439gitlink:git-show[1]:
1440
1441-------------------------------------------------
1442$ git show HEAD^ path/to/file
1443-------------------------------------------------
1444
1445which will display the given version of the file.
1446
d19fbc3c
BF
1447Ensuring good performance
1448-------------------------
1449
1450On large repositories, git depends on compression to keep the history
1451information from taking up to much space on disk or in memory.
1452
1453This compression is not performed automatically. Therefore you
17217090 1454should occasionally run gitlink:git-gc[1]:
d19fbc3c
BF
1455
1456-------------------------------------------------
1457$ git gc
1458-------------------------------------------------
1459
17217090
BF
1460to recompress the archive. This can be very time-consuming, so
1461you may prefer to run git-gc when you are not doing other work.
d19fbc3c 1462
11e016a3
BF
1463Ensuring reliability
1464--------------------
1465
1466Checking the repository for corruption
1467~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1468
1191ee18
BF
1469The gitlink:git-fsck[1] command runs a number of self-consistency checks
1470on the repository, and reports on any problems. This may take some
21dcb3b7
BF
1471time. The most common warning by far is about "dangling" objects:
1472
1473-------------------------------------------------
04e50e94 1474$ git fsck
21dcb3b7
BF
1475dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1476dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1477dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1478dangling blob 218761f9d90712d37a9c5e36f406f92202db07eb
1479dangling commit bf093535a34a4d35731aa2bd90fe6b176302f14f
1480dangling commit 8e4bec7f2ddaa268bef999853c25755452100f8e
1481dangling tree d50bb86186bf27b681d25af89d3b5b68382e4085
1482dangling tree b24c2473f1fd3d91352a624795be026d64c8841f
1483...
1484-------------------------------------------------
1485
1191ee18
BF
1486Dangling objects are objects that are harmless, but also unnecessary;
1487you can remove them at any time with gitlink:git-prune[1] or the --prune
1488option to gitlink:git-gc[1]:
21dcb3b7
BF
1489
1490-------------------------------------------------
1491$ git gc --prune
1492-------------------------------------------------
1493
1191ee18
BF
1494This may be time-consuming. Unlike most other git operations (including
1495git-gc when run without any options), it is not safe to prune while
1496other git operations are in progress in the same repository.
21dcb3b7 1497
988b27d3 1498For more about dangling objects, see <<dangling-objects>>.
21dcb3b7 1499
11e016a3
BF
1500
1501Recovering lost changes
1502~~~~~~~~~~~~~~~~~~~~~~~
1503
559e4d7a
BF
1504Reflogs
1505^^^^^^^
1506
1507Say you modify a branch with gitlink:git-reset[1] --hard, and then
1508realize that the branch was the only reference you had to that point in
1509history.
1510
1511Fortunately, git also keeps a log, called a "reflog", of all the
1512previous values of each branch. So in this case you can still find the
1513old history using, for example,
1514
1515-------------------------------------------------
1516$ git log master@{1}
1517-------------------------------------------------
1518
1519This lists the commits reachable from the previous version of the head.
1520This syntax can be used to with any git command that accepts a commit,
1521not just with git log. Some other examples:
1522
1523-------------------------------------------------
1524$ git show master@{2} # See where the branch pointed 2,
1525$ git show master@{3} # 3, ... changes ago.
1526$ gitk master@{yesterday} # See where it pointed yesterday,
1527$ gitk master@{"1 week ago"} # ... or last week
1528-------------------------------------------------
1529
1530The reflogs are kept by default for 30 days, after which they may be
036be17e 1531pruned. See gitlink:git-reflog[1] and gitlink:git-gc[1] to learn
559e4d7a
BF
1532how to control this pruning, and see the "SPECIFYING REVISIONS"
1533section of gitlink:git-rev-parse[1] for details.
1534
1535Note that the reflog history is very different from normal git history.
1536While normal history is shared by every repository that works on the
1537same project, the reflog history is not shared: it tells you only about
1538how the branches in your local repository have changed over time.
1539
1540Examining dangling objects
1541^^^^^^^^^^^^^^^^^^^^^^^^^^
1542
1543In some situations the reflog may not be able to save you. For
1544example, suppose you delete a branch, then realize you need the history
79c96c57 1545it contained. The reflog is also deleted; however, if you have not
559e4d7a
BF
1546yet pruned the repository, then you may still be able to find
1547the lost commits; run git-fsck and watch for output that mentions
1548"dangling commits":
1549
1550-------------------------------------------------
1551$ git fsck
1552dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1553dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1554dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1555...
1556-------------------------------------------------
1557
aacd404e 1558You can examine
559e4d7a
BF
1559one of those dangling commits with, for example,
1560
1561------------------------------------------------
1562$ gitk 7281251ddd --not --all
1563------------------------------------------------
1564
1565which does what it sounds like: it says that you want to see the commit
1566history that is described by the dangling commit(s), but not the
1567history that is described by all your existing branches and tags. Thus
1568you get exactly the history reachable from that commit that is lost.
1569(And notice that it might not be just one commit: we only report the
1570"tip of the line" as being dangling, but there might be a whole deep
79c96c57 1571and complex commit history that was dropped.)
559e4d7a
BF
1572
1573If you decide you want the history back, you can always create a new
1574reference pointing to it, for example, a new branch:
1575
1576------------------------------------------------
1577$ git branch recovered-branch 7281251ddd
1578------------------------------------------------
1579
11e016a3 1580
d19fbc3c 1581Sharing development with others
b684f830 1582===============================
d19fbc3c
BF
1583
1584[[getting-updates-with-git-pull]]
1585Getting updates with git pull
b684f830 1586-----------------------------
d19fbc3c
BF
1587
1588After you clone a repository and make a few changes of your own, you
1589may wish to check the original repository for updates and merge them
1590into your own work.
1591
1592We have already seen <<Updating-a-repository-with-git-fetch,how to
1593keep remote tracking branches up to date>> with gitlink:git-fetch[1],
1594and how to merge two branches. So you can merge in changes from the
1595original repository's master branch with:
1596
1597-------------------------------------------------
1598$ git fetch
1599$ git merge origin/master
1600-------------------------------------------------
1601
1602However, the gitlink:git-pull[1] command provides a way to do this in
1603one step:
1604
1605-------------------------------------------------
1606$ git pull origin master
1607-------------------------------------------------
1608
1609In fact, "origin" is normally the default repository to pull from,
1610and the default branch is normally the HEAD of the remote repository,
1611so often you can accomplish the above with just
1612
1613-------------------------------------------------
1614$ git pull
1615-------------------------------------------------
1616
1617See the descriptions of the branch.<name>.remote and
9d13bda3 1618branch.<name>.merge options in gitlink:git-config[1] to learn
d19fbc3c
BF
1619how to control these defaults depending on the current branch.
1620
1621In addition to saving you keystrokes, "git pull" also helps you by
1622producing a default commit message documenting the branch and
1623repository that you pulled from.
1624
1625(But note that no such commit will be created in the case of a
1626<<fast-forwards,fast forward>>; instead, your branch will just be
79c96c57 1627updated to point to the latest commit from the upstream branch.)
d19fbc3c 1628
1191ee18
BF
1629The git-pull command can also be given "." as the "remote" repository,
1630in which case it just merges in a branch from the current repository; so
4c63ff45
BF
1631the commands
1632
1633-------------------------------------------------
1634$ git pull . branch
1635$ git merge branch
1636-------------------------------------------------
1637
1638are roughly equivalent. The former is actually very commonly used.
1639
d19fbc3c 1640Submitting patches to a project
b684f830 1641-------------------------------
d19fbc3c
BF
1642
1643If you just have a few changes, the simplest way to submit them may
1644just be to send them as patches in email:
1645
036be17e 1646First, use gitlink:git-format-patch[1]; for example:
d19fbc3c
BF
1647
1648-------------------------------------------------
eb6ae7f4 1649$ git format-patch origin
d19fbc3c
BF
1650-------------------------------------------------
1651
1652will produce a numbered series of files in the current directory, one
1653for each patch in the current branch but not in origin/HEAD.
1654
1655You can then import these into your mail client and send them by
1656hand. However, if you have a lot to send at once, you may prefer to
1657use the gitlink:git-send-email[1] script to automate the process.
1658Consult the mailing list for your project first to determine how they
1659prefer such patches be handled.
1660
1661Importing patches to a project
b684f830 1662------------------------------
d19fbc3c
BF
1663
1664Git also provides a tool called gitlink:git-am[1] (am stands for
1665"apply mailbox"), for importing such an emailed series of patches.
1666Just save all of the patch-containing messages, in order, into a
1667single mailbox file, say "patches.mbox", then run
1668
1669-------------------------------------------------
eb6ae7f4 1670$ git am -3 patches.mbox
d19fbc3c
BF
1671-------------------------------------------------
1672
1673Git will apply each patch in order; if any conflicts are found, it
1674will stop, and you can fix the conflicts as described in
01997b4a
BF
1675"<<resolving-a-merge,Resolving a merge>>". (The "-3" option tells
1676git to perform a merge; if you would prefer it just to abort and
1677leave your tree and index untouched, you may omit that option.)
1678
1679Once the index is updated with the results of the conflict
1680resolution, instead of creating a new commit, just run
d19fbc3c
BF
1681
1682-------------------------------------------------
1683$ git am --resolved
1684-------------------------------------------------
1685
1686and git will create the commit for you and continue applying the
1687remaining patches from the mailbox.
1688
1689The final result will be a series of commits, one for each patch in
1690the original mailbox, with authorship and commit log message each
1691taken from the message containing each patch.
1692
1693[[setting-up-a-public-repository]]
1694Setting up a public repository
b684f830 1695------------------------------
d19fbc3c
BF
1696
1697Another way to submit changes to a project is to simply tell the
1698maintainer of that project to pull from your repository, exactly as
1699you did in the section "<<getting-updates-with-git-pull, Getting
1700updates with git pull>>".
1701
1702If you and maintainer both have accounts on the same machine, then
1703then you can just pull changes from each other's repositories
79c96c57
MC
1704directly; note that all of the commands (gitlink:git-clone[1],
1705git-fetch[1], git-pull[1], etc.) that accept a URL as an argument
d19fbc3c
BF
1706will also accept a local file patch; so, for example, you can
1707use
1708
1709-------------------------------------------------
1710$ git clone /path/to/repository
1711$ git pull /path/to/other/repository
1712-------------------------------------------------
1713
1714If this sort of setup is inconvenient or impossible, another (more
1715common) option is to set up a public repository on a public server.
1716This also allows you to cleanly separate private work in progress
1717from publicly visible work.
1718
1719You will continue to do your day-to-day work in your personal
1720repository, but periodically "push" changes from your personal
1721repository into your public repository, allowing other developers to
1722pull from that repository. So the flow of changes, in a situation
1723where there is one other developer with a public repository, looks
1724like this:
1725
1726 you push
1727 your personal repo ------------------> your public repo
1728 ^ |
1729 | |
1730 | you pull | they pull
1731 | |
1732 | |
1733 | they push V
1734 their public repo <------------------- their repo
1735
1736Now, assume your personal repository is in the directory ~/proj. We
1737first create a new clone of the repository:
1738
1739-------------------------------------------------
1740$ git clone --bare proj-clone.git
1741-------------------------------------------------
1742
1743The resulting directory proj-clone.git will contains a "bare" git
1744repository--it is just the contents of the ".git" directory, without
1745a checked-out copy of a working directory.
1746
1747Next, copy proj-clone.git to the server where you plan to host the
1748public repository. You can use scp, rsync, or whatever is most
1749convenient.
1750
1751If somebody else maintains the public server, they may already have
1752set up a git service for you, and you may skip to the section
1753"<<pushing-changes-to-a-public-repository,Pushing changes to a public
1754repository>>", below.
1755
1756Otherwise, the following sections explain how to export your newly
1757created public repository:
1758
1759[[exporting-via-http]]
1760Exporting a git repository via http
b684f830 1761-----------------------------------
d19fbc3c
BF
1762
1763The git protocol gives better performance and reliability, but on a
1764host with a web server set up, http exports may be simpler to set up.
1765
1766All you need to do is place the newly created bare git repository in
1767a directory that is exported by the web server, and make some
1768adjustments to give web clients some extra information they need:
1769
1770-------------------------------------------------
1771$ mv proj.git /home/you/public_html/proj.git
1772$ cd proj.git
1773$ git update-server-info
1774$ chmod a+x hooks/post-update
1775-------------------------------------------------
1776
1777(For an explanation of the last two lines, see
1778gitlink:git-update-server-info[1], and the documentation
1779link:hooks.txt[Hooks used by git].)
1780
1781Advertise the url of proj.git. Anybody else should then be able to
1782clone or pull from that url, for example with a commandline like:
1783
1784-------------------------------------------------
1785$ git clone http://yourserver.com/~you/proj.git
1786-------------------------------------------------
1787
1788(See also
1789link:howto/setup-git-server-over-http.txt[setup-git-server-over-http]
1790for a slightly more sophisticated setup using WebDAV which also
1791allows pushing over http.)
1792
1793[[exporting-via-git]]
1794Exporting a git repository via the git protocol
b684f830 1795-----------------------------------------------
d19fbc3c
BF
1796
1797This is the preferred method.
1798
1799For now, we refer you to the gitlink:git-daemon[1] man page for
1800instructions. (See especially the examples section.)
1801
1802[[pushing-changes-to-a-public-repository]]
1803Pushing changes to a public repository
b684f830 1804--------------------------------------
d19fbc3c
BF
1805
1806Note that the two techniques outline above (exporting via
1807<<exporting-via-http,http>> or <<exporting-via-git,git>>) allow other
1808maintainers to fetch your latest changes, but they do not allow write
1809access, which you will need to update the public repository with the
1810latest changes created in your private repository.
1811
1812The simplest way to do this is using gitlink:git-push[1] and ssh; to
1813update the remote branch named "master" with the latest state of your
1814branch named "master", run
1815
1816-------------------------------------------------
1817$ git push ssh://yourserver.com/~you/proj.git master:master
1818-------------------------------------------------
1819
1820or just
1821
1822-------------------------------------------------
1823$ git push ssh://yourserver.com/~you/proj.git master
1824-------------------------------------------------
1825
1826As with git-fetch, git-push will complain if this does not result in
1827a <<fast-forwards,fast forward>>. Normally this is a sign of
1828something wrong. However, if you are sure you know what you're
1829doing, you may force git-push to perform the update anyway by
1830proceeding the branch name by a plus sign:
1831
1832-------------------------------------------------
1833$ git push ssh://yourserver.com/~you/proj.git +master
1834-------------------------------------------------
1835
1836As with git-fetch, you may also set up configuration options to
1837save typing; so, for example, after
1838
1839-------------------------------------------------
1840$ cat >.git/config <<EOF
1841[remote "public-repo"]
1842 url = ssh://yourserver.com/~you/proj.git
1843EOF
1844-------------------------------------------------
1845
1846you should be able to perform the above push with just
1847
1848-------------------------------------------------
1849$ git push public-repo master
1850-------------------------------------------------
1851
1852See the explanations of the remote.<name>.url, branch.<name>.remote,
9d13bda3 1853and remote.<name>.push options in gitlink:git-config[1] for
d19fbc3c
BF
1854details.
1855
1856Setting up a shared repository
b684f830 1857------------------------------
d19fbc3c
BF
1858
1859Another way to collaborate is by using a model similar to that
1860commonly used in CVS, where several developers with special rights
1861all push to and pull from a single shared repository. See
1862link:cvs-migration.txt[git for CVS users] for instructions on how to
1863set this up.
1864
b684f830
BF
1865Allow web browsing of a repository
1866----------------------------------
d19fbc3c 1867
a8cd1402
BF
1868The gitweb cgi script provides users an easy way to browse your
1869project's files and history without having to install git; see the file
1870gitweb/README in the git source tree for instructions on setting it up.
d19fbc3c 1871
b684f830
BF
1872Examples
1873--------
d19fbc3c 1874
b684f830 1875TODO: topic branches, typical roles as in everyday.txt, ?
d19fbc3c 1876
d19fbc3c 1877
d19fbc3c 1878[[cleaning-up-history]]
4c63ff45
BF
1879Rewriting history and maintaining patch series
1880==============================================
1881
1882Normally commits are only added to a project, never taken away or
1883replaced. Git is designed with this assumption, and violating it will
1884cause git's merge machinery (for example) to do the wrong thing.
1885
1886However, there is a situation in which it can be useful to violate this
1887assumption.
1888
1889Creating the perfect patch series
1890---------------------------------
1891
1892Suppose you are a contributor to a large project, and you want to add a
1893complicated feature, and to present it to the other developers in a way
1894that makes it easy for them to read your changes, verify that they are
1895correct, and understand why you made each change.
1896
b181d57f 1897If you present all of your changes as a single patch (or commit), they
79c96c57 1898may find that it is too much to digest all at once.
4c63ff45
BF
1899
1900If you present them with the entire history of your work, complete with
1901mistakes, corrections, and dead ends, they may be overwhelmed.
1902
1903So the ideal is usually to produce a series of patches such that:
1904
1905 1. Each patch can be applied in order.
1906
1907 2. Each patch includes a single logical change, together with a
1908 message explaining the change.
1909
1910 3. No patch introduces a regression: after applying any initial
1911 part of the series, the resulting project still compiles and
1912 works, and has no bugs that it didn't have before.
1913
1914 4. The complete series produces the same end result as your own
1915 (probably much messier!) development process did.
1916
b181d57f
BF
1917We will introduce some tools that can help you do this, explain how to
1918use them, and then explain some of the problems that can arise because
1919you are rewriting history.
4c63ff45
BF
1920
1921Keeping a patch series up to date using git-rebase
1922--------------------------------------------------
1923
79c96c57
MC
1924Suppose that you create a branch "mywork" on a remote-tracking branch
1925"origin", and create some commits on top of it:
4c63ff45
BF
1926
1927-------------------------------------------------
1928$ git checkout -b mywork origin
1929$ vi file.txt
1930$ git commit
1931$ vi otherfile.txt
1932$ git commit
1933...
1934-------------------------------------------------
1935
1936You have performed no merges into mywork, so it is just a simple linear
1937sequence of patches on top of "origin":
1938
1939
1940 o--o--o <-- origin
1941 \
1942 o--o--o <-- mywork
1943
1944Some more interesting work has been done in the upstream project, and
1945"origin" has advanced:
1946
1947 o--o--O--o--o--o <-- origin
1948 \
1949 a--b--c <-- mywork
1950
1951At this point, you could use "pull" to merge your changes back in;
1952the result would create a new merge commit, like this:
1953
1954
1955 o--o--O--o--o--o <-- origin
1956 \ \
1957 a--b--c--m <-- mywork
1958
1959However, if you prefer to keep the history in mywork a simple series of
1960commits without any merges, you may instead choose to use
1961gitlink:git-rebase[1]:
1962
1963-------------------------------------------------
1964$ git checkout mywork
1965$ git rebase origin
1966-------------------------------------------------
1967
b181d57f
BF
1968This will remove each of your commits from mywork, temporarily saving
1969them as patches (in a directory named ".dotest"), update mywork to
1970point at the latest version of origin, then apply each of the saved
1971patches to the new mywork. The result will look like:
4c63ff45
BF
1972
1973
1974 o--o--O--o--o--o <-- origin
1975 \
1976 a'--b'--c' <-- mywork
1977
b181d57f
BF
1978In the process, it may discover conflicts. In that case it will stop
1979and allow you to fix the conflicts; after fixing conflicts, use "git
1980add" to update the index with those contents, and then, instead of
1981running git-commit, just run
4c63ff45
BF
1982
1983-------------------------------------------------
1984$ git rebase --continue
1985-------------------------------------------------
1986
1987and git will continue applying the rest of the patches.
1988
1989At any point you may use the --abort option to abort this process and
1990return mywork to the state it had before you started the rebase:
1991
1992-------------------------------------------------
1993$ git rebase --abort
1994-------------------------------------------------
1995
365aa199
BF
1996Modifying a single commit
1997-------------------------
1998
1999We saw in <<fixing-a-mistake-by-editing-history>> that you can replace the
2000most recent commit using
2001
2002-------------------------------------------------
2003$ git commit --amend
2004-------------------------------------------------
2005
2006which will replace the old commit by a new commit incorporating your
2007changes, giving you a chance to edit the old commit message first.
2008
2009You can also use a combination of this and gitlink:git-rebase[1] to edit
2010commits further back in your history. First, tag the problematic commit with
2011
2012-------------------------------------------------
2013$ git tag bad mywork~5
2014-------------------------------------------------
2015
2016(Either gitk or git-log may be useful for finding the commit.)
2017
2018Then check out a new branch at that commit, edit it, and rebase the rest of
2019the series on top of it:
2020
2021-------------------------------------------------
2022$ git checkout -b TMP bad
2023$ # make changes here and update the index
2024$ git commit --amend
2025$ git rebase --onto TMP bad mywork
2026-------------------------------------------------
2027
2028When you're done, you'll be left with mywork checked out, with the top patches
2029on mywork reapplied on top of the modified commit you created in TMP. You can
2030then clean up with
2031
2032-------------------------------------------------
2033$ git branch -d TMP
2034$ git tag -d bad
2035-------------------------------------------------
2036
2037Note that the immutable nature of git history means that you haven't really
2038"modified" existing commits; instead, you have replaced the old commits with
2039new commits having new object names.
2040
4c63ff45
BF
2041Reordering or selecting from a patch series
2042-------------------------------------------
2043
b181d57f
BF
2044Given one existing commit, the gitlink:git-cherry-pick[1] command
2045allows you to apply the change introduced by that commit and create a
2046new commit that records it. So, for example, if "mywork" points to a
2047series of patches on top of "origin", you might do something like:
2048
2049-------------------------------------------------
2050$ git checkout -b mywork-new origin
2051$ gitk origin..mywork &
2052-------------------------------------------------
2053
2054And browse through the list of patches in the mywork branch using gitk,
2055applying them (possibly in a different order) to mywork-new using
2056cherry-pick, and possibly modifying them as you go using commit
2057--amend.
2058
2059Another technique is to use git-format-patch to create a series of
2060patches, then reset the state to before the patches:
4c63ff45 2061
b181d57f
BF
2062-------------------------------------------------
2063$ git format-patch origin
2064$ git reset --hard origin
2065-------------------------------------------------
4c63ff45 2066
b181d57f
BF
2067Then modify, reorder, or eliminate patches as preferred before applying
2068them again with gitlink:git-am[1].
4c63ff45
BF
2069
2070Other tools
2071-----------
2072
b181d57f 2073There are numerous other tools, such as stgit, which exist for the
79c96c57 2074purpose of maintaining a patch series. These are outside of the scope of
b181d57f 2075this manual.
4c63ff45
BF
2076
2077Problems with rewriting history
2078-------------------------------
2079
b181d57f
BF
2080The primary problem with rewriting the history of a branch has to do
2081with merging. Suppose somebody fetches your branch and merges it into
2082their branch, with a result something like this:
2083
2084 o--o--O--o--o--o <-- origin
2085 \ \
2086 t--t--t--m <-- their branch:
2087
2088Then suppose you modify the last three commits:
2089
2090 o--o--o <-- new head of origin
2091 /
2092 o--o--O--o--o--o <-- old head of origin
2093
2094If we examined all this history together in one repository, it will
2095look like:
2096
2097 o--o--o <-- new head of origin
2098 /
2099 o--o--O--o--o--o <-- old head of origin
2100 \ \
2101 t--t--t--m <-- their branch:
2102
2103Git has no way of knowing that the new head is an updated version of
2104the old head; it treats this situation exactly the same as it would if
2105two developers had independently done the work on the old and new heads
2106in parallel. At this point, if someone attempts to merge the new head
2107in to their branch, git will attempt to merge together the two (old and
2108new) lines of development, instead of trying to replace the old by the
2109new. The results are likely to be unexpected.
2110
2111You may still choose to publish branches whose history is rewritten,
2112and it may be useful for others to be able to fetch those branches in
2113order to examine or test them, but they should not attempt to pull such
2114branches into their own work.
2115
2116For true distributed development that supports proper merging,
2117published branches should never be rewritten.
2118
2119Advanced branch management
2120==========================
4c63ff45 2121
b181d57f
BF
2122Fetching individual branches
2123----------------------------
2124
2125Instead of using gitlink:git-remote[1], you can also choose just
2126to update one branch at a time, and to store it locally under an
2127arbitrary name:
2128
2129-------------------------------------------------
2130$ git fetch origin todo:my-todo-work
2131-------------------------------------------------
2132
2133The first argument, "origin", just tells git to fetch from the
2134repository you originally cloned from. The second argument tells git
2135to fetch the branch named "todo" from the remote repository, and to
2136store it locally under the name refs/heads/my-todo-work.
2137
2138You can also fetch branches from other repositories; so
2139
2140-------------------------------------------------
2141$ git fetch git://example.com/proj.git master:example-master
2142-------------------------------------------------
2143
2144will create a new branch named "example-master" and store in it the
2145branch named "master" from the repository at the given URL. If you
2146already have a branch named example-master, it will attempt to
2147"fast-forward" to the commit given by example.com's master branch. So
2148next we explain what a fast-forward is:
2149
2150[[fast-forwards]]
2151Understanding git history: fast-forwards
2152----------------------------------------
2153
2154In the previous example, when updating an existing branch, "git
2155fetch" checks to make sure that the most recent commit on the remote
2156branch is a descendant of the most recent commit on your copy of the
2157branch before updating your copy of the branch to point at the new
2158commit. Git calls this process a "fast forward".
2159
2160A fast forward looks something like this:
2161
2162 o--o--o--o <-- old head of the branch
2163 \
2164 o--o--o <-- new head of the branch
2165
2166
2167In some cases it is possible that the new head will *not* actually be
2168a descendant of the old head. For example, the developer may have
2169realized she made a serious mistake, and decided to backtrack,
2170resulting in a situation like:
2171
2172 o--o--o--o--a--b <-- old head of the branch
2173 \
2174 o--o--o <-- new head of the branch
2175
2176
2177
2178In this case, "git fetch" will fail, and print out a warning.
2179
2180In that case, you can still force git to update to the new head, as
2181described in the following section. However, note that in the
2182situation above this may mean losing the commits labeled "a" and "b",
2183unless you've already created a reference of your own pointing to
2184them.
2185
2186Forcing git fetch to do non-fast-forward updates
2187------------------------------------------------
2188
2189If git fetch fails because the new head of a branch is not a
2190descendant of the old head, you may force the update with:
2191
2192-------------------------------------------------
2193$ git fetch git://example.com/proj.git +master:refs/remotes/example/master
2194-------------------------------------------------
2195
79c96c57 2196Note the addition of the "+" sign. Be aware that commits that the
b181d57f
BF
2197old version of example/master pointed at may be lost, as we saw in
2198the previous section.
2199
2200Configuring remote branches
2201---------------------------
2202
2203We saw above that "origin" is just a shortcut to refer to the
79c96c57 2204repository that you originally cloned from. This information is
b181d57f 2205stored in git configuration variables, which you can see using
9d13bda3 2206gitlink:git-config[1]:
b181d57f
BF
2207
2208-------------------------------------------------
9d13bda3 2209$ git config -l
b181d57f
BF
2210core.repositoryformatversion=0
2211core.filemode=true
2212core.logallrefupdates=true
2213remote.origin.url=git://git.kernel.org/pub/scm/git/git.git
2214remote.origin.fetch=+refs/heads/*:refs/remotes/origin/*
2215branch.master.remote=origin
2216branch.master.merge=refs/heads/master
2217-------------------------------------------------
2218
2219If there are other repositories that you also use frequently, you can
2220create similar configuration options to save typing; for example,
2221after
2222
2223-------------------------------------------------
9d13bda3 2224$ git config remote.example.url git://example.com/proj.git
b181d57f
BF
2225-------------------------------------------------
2226
2227then the following two commands will do the same thing:
2228
2229-------------------------------------------------
2230$ git fetch git://example.com/proj.git master:refs/remotes/example/master
2231$ git fetch example master:refs/remotes/example/master
2232-------------------------------------------------
2233
2234Even better, if you add one more option:
2235
2236-------------------------------------------------
9d13bda3 2237$ git config remote.example.fetch master:refs/remotes/example/master
b181d57f
BF
2238-------------------------------------------------
2239
2240then the following commands will all do the same thing:
2241
2242-------------------------------------------------
2243$ git fetch git://example.com/proj.git master:ref/remotes/example/master
2244$ git fetch example master:ref/remotes/example/master
2245$ git fetch example example/master
2246$ git fetch example
2247-------------------------------------------------
2248
2249You can also add a "+" to force the update each time:
2250
2251-------------------------------------------------
9d13bda3 2252$ git config remote.example.fetch +master:ref/remotes/example/master
b181d57f
BF
2253-------------------------------------------------
2254
2255Don't do this unless you're sure you won't mind "git fetch" possibly
2256throwing away commits on mybranch.
2257
2258Also note that all of the above configuration can be performed by
2259directly editing the file .git/config instead of using
9d13bda3 2260gitlink:git-config[1].
b181d57f 2261
9d13bda3 2262See gitlink:git-config[1] for more details on the configuration
b181d57f 2263options mentioned above.
d19fbc3c 2264
d19fbc3c 2265
35121930 2266[[git-internals]]
d19fbc3c
BF
2267Git internals
2268=============
2269
b181d57f
BF
2270There are two object abstractions: the "object database", and the
2271"current directory cache" aka "index".
2272
2273The Object Database
2274-------------------
2275
2276The object database is literally just a content-addressable collection
2277of objects. All objects are named by their content, which is
2278approximated by the SHA1 hash of the object itself. Objects may refer
2279to other objects (by referencing their SHA1 hash), and so you can
2280build up a hierarchy of objects.
2281
2282All objects have a statically determined "type" aka "tag", which is
2283determined at object creation time, and which identifies the format of
2284the object (i.e. how it is used, and how it can refer to other
2285objects). There are currently four different object types: "blob",
2286"tree", "commit" and "tag".
2287
2288A "blob" object cannot refer to any other object, and is, like the type
2289implies, a pure storage object containing some user data. It is used to
2290actually store the file data, i.e. a blob object is associated with some
2291particular version of some file.
2292
2293A "tree" object is an object that ties one or more "blob" objects into a
2294directory structure. In addition, a tree object can refer to other tree
2295objects, thus creating a directory hierarchy.
2296
2297A "commit" object ties such directory hierarchies together into
2298a DAG of revisions - each "commit" is associated with exactly one tree
2299(the directory hierarchy at the time of the commit). In addition, a
2300"commit" refers to one or more "parent" commit objects that describe the
2301history of how we arrived at that directory hierarchy.
2302
2303As a special case, a commit object with no parents is called the "root"
2304object, and is the point of an initial project commit. Each project
2305must have at least one root, and while you can tie several different
2306root objects together into one project by creating a commit object which
2307has two or more separate roots as its ultimate parents, that's probably
2308just going to confuse people. So aim for the notion of "one root object
2309per project", even if git itself does not enforce that.
2310
2311A "tag" object symbolically identifies and can be used to sign other
2312objects. It contains the identifier and type of another object, a
2313symbolic name (of course!) and, optionally, a signature.
2314
2315Regardless of object type, all objects share the following
2316characteristics: they are all deflated with zlib, and have a header
2317that not only specifies their type, but also provides size information
2318about the data in the object. It's worth noting that the SHA1 hash
2319that is used to name the object is the hash of the original data
2320plus this header, so `sha1sum` 'file' does not match the object name
2321for 'file'.
2322(Historical note: in the dawn of the age of git the hash
2323was the sha1 of the 'compressed' object.)
2324
2325As a result, the general consistency of an object can always be tested
2326independently of the contents or the type of the object: all objects can
2327be validated by verifying that (a) their hashes match the content of the
2328file and (b) the object successfully inflates to a stream of bytes that
2329forms a sequence of <ascii type without space> + <space> + <ascii decimal
2330size> + <byte\0> + <binary object data>.
2331
2332The structured objects can further have their structure and
2333connectivity to other objects verified. This is generally done with
04e50e94 2334the `git-fsck` program, which generates a full dependency graph
b181d57f
BF
2335of all objects, and verifies their internal consistency (in addition
2336to just verifying their superficial consistency through the hash).
2337
2338The object types in some more detail:
2339
2340Blob Object
2341-----------
2342
2343A "blob" object is nothing but a binary blob of data, and doesn't
2344refer to anything else. There is no signature or any other
2345verification of the data, so while the object is consistent (it 'is'
2346indexed by its sha1 hash, so the data itself is certainly correct), it
2347has absolutely no other attributes. No name associations, no
2348permissions. It is purely a blob of data (i.e. normally "file
2349contents").
2350
2351In particular, since the blob is entirely defined by its data, if two
2352files in a directory tree (or in multiple different versions of the
2353repository) have the same contents, they will share the same blob
2354object. The object is totally independent of its location in the
2355directory tree, and renaming a file does not change the object that
2356file is associated with in any way.
2357
2358A blob is typically created when gitlink:git-update-index[1]
2359is run, and its data can be accessed by gitlink:git-cat-file[1].
2360
2361Tree Object
2362-----------
2363
2364The next hierarchical object type is the "tree" object. A tree object
2365is a list of mode/name/blob data, sorted by name. Alternatively, the
2366mode data may specify a directory mode, in which case instead of
2367naming a blob, that name is associated with another TREE object.
2368
2369Like the "blob" object, a tree object is uniquely determined by the
2370set contents, and so two separate but identical trees will always
2371share the exact same object. This is true at all levels, i.e. it's
2372true for a "leaf" tree (which does not refer to any other trees, only
2373blobs) as well as for a whole subdirectory.
2374
2375For that reason a "tree" object is just a pure data abstraction: it
2376has no history, no signatures, no verification of validity, except
2377that since the contents are again protected by the hash itself, we can
2378trust that the tree is immutable and its contents never change.
2379
2380So you can trust the contents of a tree to be valid, the same way you
2381can trust the contents of a blob, but you don't know where those
2382contents 'came' from.
2383
2384Side note on trees: since a "tree" object is a sorted list of
2385"filename+content", you can create a diff between two trees without
2386actually having to unpack two trees. Just ignore all common parts,
2387and your diff will look right. In other words, you can effectively
2388(and efficiently) tell the difference between any two random trees by
2389O(n) where "n" is the size of the difference, rather than the size of
2390the tree.
2391
2392Side note 2 on trees: since the name of a "blob" depends entirely and
2393exclusively on its contents (i.e. there are no names or permissions
2394involved), you can see trivial renames or permission changes by
2395noticing that the blob stayed the same. However, renames with data
2396changes need a smarter "diff" implementation.
2397
2398A tree is created with gitlink:git-write-tree[1] and
2399its data can be accessed by gitlink:git-ls-tree[1].
2400Two trees can be compared with gitlink:git-diff-tree[1].
2401
2402Commit Object
2403-------------
2404
2405The "commit" object is an object that introduces the notion of
2406history into the picture. In contrast to the other objects, it
2407doesn't just describe the physical state of a tree, it describes how
2408we got there, and why.
2409
2410A "commit" is defined by the tree-object that it results in, the
2411parent commits (zero, one or more) that led up to that point, and a
2412comment on what happened. Again, a commit is not trusted per se:
2413the contents are well-defined and "safe" due to the cryptographically
2414strong signatures at all levels, but there is no reason to believe
2415that the tree is "good" or that the merge information makes sense.
2416The parents do not have to actually have any relationship with the
2417result, for example.
2418
2419Note on commits: unlike real SCM's, commits do not contain
2420rename information or file mode change information. All of that is
2421implicit in the trees involved (the result tree, and the result trees
2422of the parents), and describing that makes no sense in this idiotic
2423file manager.
2424
2425A commit is created with gitlink:git-commit-tree[1] and
2426its data can be accessed by gitlink:git-cat-file[1].
2427
2428Trust
2429-----
2430
2431An aside on the notion of "trust". Trust is really outside the scope
2432of "git", but it's worth noting a few things. First off, since
2433everything is hashed with SHA1, you 'can' trust that an object is
2434intact and has not been messed with by external sources. So the name
2435of an object uniquely identifies a known state - just not a state that
2436you may want to trust.
2437
2438Furthermore, since the SHA1 signature of a commit refers to the
2439SHA1 signatures of the tree it is associated with and the signatures
2440of the parent, a single named commit specifies uniquely a whole set
2441of history, with full contents. You can't later fake any step of the
2442way once you have the name of a commit.
2443
2444So to introduce some real trust in the system, the only thing you need
2445to do is to digitally sign just 'one' special note, which includes the
2446name of a top-level commit. Your digital signature shows others
2447that you trust that commit, and the immutability of the history of
2448commits tells others that they can trust the whole history.
2449
2450In other words, you can easily validate a whole archive by just
2451sending out a single email that tells the people the name (SHA1 hash)
2452of the top commit, and digitally sign that email using something
2453like GPG/PGP.
2454
2455To assist in this, git also provides the tag object...
2456
2457Tag Object
2458----------
2459
2460Git provides the "tag" object to simplify creating, managing and
2461exchanging symbolic and signed tokens. The "tag" object at its
2462simplest simply symbolically identifies another object by containing
2463the sha1, type and symbolic name.
2464
2465However it can optionally contain additional signature information
2466(which git doesn't care about as long as there's less than 8k of
2467it). This can then be verified externally to git.
2468
2469Note that despite the tag features, "git" itself only handles content
2470integrity; the trust framework (and signature provision and
2471verification) has to come from outside.
2472
2473A tag is created with gitlink:git-mktag[1],
2474its data can be accessed by gitlink:git-cat-file[1],
2475and the signature can be verified by
2476gitlink:git-verify-tag[1].
2477
2478
2479The "index" aka "Current Directory Cache"
2480-----------------------------------------
2481
2482The index is a simple binary file, which contains an efficient
2483representation of a virtual directory content at some random time. It
2484does so by a simple array that associates a set of names, dates,
2485permissions and content (aka "blob") objects together. The cache is
2486always kept ordered by name, and names are unique (with a few very
2487specific rules) at any point in time, but the cache has no long-term
2488meaning, and can be partially updated at any time.
2489
2490In particular, the index certainly does not need to be consistent with
2491the current directory contents (in fact, most operations will depend on
2492different ways to make the index 'not' be consistent with the directory
2493hierarchy), but it has three very important attributes:
2494
2495'(a) it can re-generate the full state it caches (not just the
2496directory structure: it contains pointers to the "blob" objects so
2497that it can regenerate the data too)'
2498
2499As a special case, there is a clear and unambiguous one-way mapping
2500from a current directory cache to a "tree object", which can be
2501efficiently created from just the current directory cache without
2502actually looking at any other data. So a directory cache at any one
2503time uniquely specifies one and only one "tree" object (but has
2504additional data to make it easy to match up that tree object with what
2505has happened in the directory)
2506
2507'(b) it has efficient methods for finding inconsistencies between that
2508cached state ("tree object waiting to be instantiated") and the
2509current state.'
2510
2511'(c) it can additionally efficiently represent information about merge
2512conflicts between different tree objects, allowing each pathname to be
2513associated with sufficient information about the trees involved that
2514you can create a three-way merge between them.'
2515
79c96c57 2516Those are the ONLY three things that the directory cache does. It's a
b181d57f
BF
2517cache, and the normal operation is to re-generate it completely from a
2518known tree object, or update/compare it with a live tree that is being
2519developed. If you blow the directory cache away entirely, you generally
2520haven't lost any information as long as you have the name of the tree
2521that it described.
2522
2523At the same time, the index is at the same time also the
2524staging area for creating new trees, and creating a new tree always
2525involves a controlled modification of the index file. In particular,
2526the index file can have the representation of an intermediate tree that
2527has not yet been instantiated. So the index can be thought of as a
2528write-back cache, which can contain dirty information that has not yet
2529been written back to the backing store.
2530
2531
2532
2533The Workflow
2534------------
2535
2536Generally, all "git" operations work on the index file. Some operations
2537work *purely* on the index file (showing the current state of the
2538index), but most operations move data to and from the index file. Either
2539from the database or from the working directory. Thus there are four
2540main combinations:
2541
2542working directory -> index
2543~~~~~~~~~~~~~~~~~~~~~~~~~~
2544
2545You update the index with information from the working directory with
2546the gitlink:git-update-index[1] command. You
2547generally update the index information by just specifying the filename
2548you want to update, like so:
2549
2550-------------------------------------------------
2551$ git-update-index filename
2552-------------------------------------------------
2553
2554but to avoid common mistakes with filename globbing etc, the command
2555will not normally add totally new entries or remove old entries,
2556i.e. it will normally just update existing cache entries.
2557
2558To tell git that yes, you really do realize that certain files no
2559longer exist, or that new files should be added, you
2560should use the `--remove` and `--add` flags respectively.
2561
2562NOTE! A `--remove` flag does 'not' mean that subsequent filenames will
2563necessarily be removed: if the files still exist in your directory
2564structure, the index will be updated with their new status, not
2565removed. The only thing `--remove` means is that update-cache will be
2566considering a removed file to be a valid thing, and if the file really
2567does not exist any more, it will update the index accordingly.
2568
2569As a special case, you can also do `git-update-index --refresh`, which
2570will refresh the "stat" information of each index to match the current
2571stat information. It will 'not' update the object status itself, and
2572it will only update the fields that are used to quickly test whether
2573an object still matches its old backing store object.
2574
2575index -> object database
2576~~~~~~~~~~~~~~~~~~~~~~~~
2577
2578You write your current index file to a "tree" object with the program
2579
2580-------------------------------------------------
2581$ git-write-tree
2582-------------------------------------------------
2583
2584that doesn't come with any options - it will just write out the
2585current index into the set of tree objects that describe that state,
2586and it will return the name of the resulting top-level tree. You can
2587use that tree to re-generate the index at any time by going in the
2588other direction:
2589
2590object database -> index
2591~~~~~~~~~~~~~~~~~~~~~~~~
2592
2593You read a "tree" file from the object database, and use that to
2594populate (and overwrite - don't do this if your index contains any
2595unsaved state that you might want to restore later!) your current
2596index. Normal operation is just
2597
2598-------------------------------------------------
2599$ git-read-tree <sha1 of tree>
2600-------------------------------------------------
2601
2602and your index file will now be equivalent to the tree that you saved
2603earlier. However, that is only your 'index' file: your working
2604directory contents have not been modified.
2605
2606index -> working directory
2607~~~~~~~~~~~~~~~~~~~~~~~~~~
2608
2609You update your working directory from the index by "checking out"
2610files. This is not a very common operation, since normally you'd just
2611keep your files updated, and rather than write to your working
2612directory, you'd tell the index files about the changes in your
2613working directory (i.e. `git-update-index`).
2614
2615However, if you decide to jump to a new version, or check out somebody
2616else's version, or just restore a previous tree, you'd populate your
2617index file with read-tree, and then you need to check out the result
2618with
2619
2620-------------------------------------------------
2621$ git-checkout-index filename
2622-------------------------------------------------
2623
2624or, if you want to check out all of the index, use `-a`.
2625
2626NOTE! git-checkout-index normally refuses to overwrite old files, so
2627if you have an old version of the tree already checked out, you will
2628need to use the "-f" flag ('before' the "-a" flag or the filename) to
2629'force' the checkout.
2630
2631
2632Finally, there are a few odds and ends which are not purely moving
2633from one representation to the other:
2634
2635Tying it all together
2636~~~~~~~~~~~~~~~~~~~~~
2637
2638To commit a tree you have instantiated with "git-write-tree", you'd
2639create a "commit" object that refers to that tree and the history
2640behind it - most notably the "parent" commits that preceded it in
2641history.
2642
2643Normally a "commit" has one parent: the previous state of the tree
2644before a certain change was made. However, sometimes it can have two
2645or more parent commits, in which case we call it a "merge", due to the
2646fact that such a commit brings together ("merges") two or more
2647previous states represented by other commits.
2648
2649In other words, while a "tree" represents a particular directory state
2650of a working directory, a "commit" represents that state in "time",
2651and explains how we got there.
2652
2653You create a commit object by giving it the tree that describes the
2654state at the time of the commit, and a list of parents:
2655
2656-------------------------------------------------
2657$ git-commit-tree <tree> -p <parent> [-p <parent2> ..]
2658-------------------------------------------------
2659
2660and then giving the reason for the commit on stdin (either through
2661redirection from a pipe or file, or by just typing it at the tty).
2662
2663git-commit-tree will return the name of the object that represents
2664that commit, and you should save it away for later use. Normally,
2665you'd commit a new `HEAD` state, and while git doesn't care where you
2666save the note about that state, in practice we tend to just write the
2667result to the file pointed at by `.git/HEAD`, so that we can always see
2668what the last committed state was.
2669
2670Here is an ASCII art by Jon Loeliger that illustrates how
2671various pieces fit together.
2672
2673------------
2674
2675 commit-tree
2676 commit obj
2677 +----+
2678 | |
2679 | |
2680 V V
2681 +-----------+
2682 | Object DB |
2683 | Backing |
2684 | Store |
2685 +-----------+
2686 ^
2687 write-tree | |
2688 tree obj | |
2689 | | read-tree
2690 | | tree obj
2691 V
2692 +-----------+
2693 | Index |
2694 | "cache" |
2695 +-----------+
2696 update-index ^
2697 blob obj | |
2698 | |
2699 checkout-index -u | | checkout-index
2700 stat | | blob obj
2701 V
2702 +-----------+
2703 | Working |
2704 | Directory |
2705 +-----------+
2706
2707------------
2708
2709
2710Examining the data
2711------------------
2712
2713You can examine the data represented in the object database and the
2714index with various helper tools. For every object, you can use
2715gitlink:git-cat-file[1] to examine details about the
2716object:
2717
2718-------------------------------------------------
2719$ git-cat-file -t <objectname>
2720-------------------------------------------------
2721
2722shows the type of the object, and once you have the type (which is
2723usually implicit in where you find the object), you can use
2724
2725-------------------------------------------------
2726$ git-cat-file blob|tree|commit|tag <objectname>
2727-------------------------------------------------
2728
2729to show its contents. NOTE! Trees have binary content, and as a result
2730there is a special helper for showing that content, called
2731`git-ls-tree`, which turns the binary content into a more easily
2732readable form.
2733
2734It's especially instructive to look at "commit" objects, since those
2735tend to be small and fairly self-explanatory. In particular, if you
2736follow the convention of having the top commit name in `.git/HEAD`,
2737you can do
2738
2739-------------------------------------------------
2740$ git-cat-file commit HEAD
2741-------------------------------------------------
2742
2743to see what the top commit was.
2744
2745Merging multiple trees
d19fbc3c
BF
2746----------------------
2747
b181d57f
BF
2748Git helps you do a three-way merge, which you can expand to n-way by
2749repeating the merge procedure arbitrary times until you finally
2750"commit" the state. The normal situation is that you'd only do one
2751three-way merge (two parents), and commit it, but if you like to, you
2752can do multiple parents in one go.
2753
2754To do a three-way merge, you need the two sets of "commit" objects
2755that you want to merge, use those to find the closest common parent (a
2756third "commit" object), and then use those commit objects to find the
2757state of the directory ("tree" object) at these points.
2758
2759To get the "base" for the merge, you first look up the common parent
2760of two commits with
2761
2762-------------------------------------------------
2763$ git-merge-base <commit1> <commit2>
2764-------------------------------------------------
2765
2766which will return you the commit they are both based on. You should
2767now look up the "tree" objects of those commits, which you can easily
2768do with (for example)
2769
2770-------------------------------------------------
2771$ git-cat-file commit <commitname> | head -1
2772-------------------------------------------------
2773
2774since the tree object information is always the first line in a commit
2775object.
2776
1191ee18
BF
2777Once you know the three trees you are going to merge (the one "original"
2778tree, aka the common case, and the two "result" trees, aka the branches
2779you want to merge), you do a "merge" read into the index. This will
2780complain if it has to throw away your old index contents, so you should
b181d57f 2781make sure that you've committed those - in fact you would normally
1191ee18
BF
2782always do a merge against your last commit (which should thus match what
2783you have in your current index anyway).
b181d57f
BF
2784
2785To do the merge, do
2786
2787-------------------------------------------------
2788$ git-read-tree -m -u <origtree> <yourtree> <targettree>
2789-------------------------------------------------
2790
2791which will do all trivial merge operations for you directly in the
2792index file, and you can just write the result out with
2793`git-write-tree`.
2794
2795
2796Merging multiple trees, continued
2797---------------------------------
2798
2799Sadly, many merges aren't trivial. If there are files that have
2800been added.moved or removed, or if both branches have modified the
2801same file, you will be left with an index tree that contains "merge
2802entries" in it. Such an index tree can 'NOT' be written out to a tree
2803object, and you will have to resolve any such merge clashes using
2804other tools before you can write out the result.
2805
2806You can examine such index state with `git-ls-files --unmerged`
2807command. An example:
2808
2809------------------------------------------------
2810$ git-read-tree -m $orig HEAD $target
2811$ git-ls-files --unmerged
2812100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1 hello.c
2813100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2 hello.c
2814100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello.c
2815------------------------------------------------
2816
2817Each line of the `git-ls-files --unmerged` output begins with
2818the blob mode bits, blob SHA1, 'stage number', and the
2819filename. The 'stage number' is git's way to say which tree it
2820came from: stage 1 corresponds to `$orig` tree, stage 2 `HEAD`
2821tree, and stage3 `$target` tree.
2822
2823Earlier we said that trivial merges are done inside
2824`git-read-tree -m`. For example, if the file did not change
2825from `$orig` to `HEAD` nor `$target`, or if the file changed
2826from `$orig` to `HEAD` and `$orig` to `$target` the same way,
2827obviously the final outcome is what is in `HEAD`. What the
2828above example shows is that file `hello.c` was changed from
2829`$orig` to `HEAD` and `$orig` to `$target` in a different way.
2830You could resolve this by running your favorite 3-way merge
2831program, e.g. `diff3` or `merge`, on the blob objects from
2832these three stages yourself, like this:
2833
2834------------------------------------------------
2835$ git-cat-file blob 263414f... >hello.c~1
2836$ git-cat-file blob 06fa6a2... >hello.c~2
2837$ git-cat-file blob cc44c73... >hello.c~3
2838$ merge hello.c~2 hello.c~1 hello.c~3
2839------------------------------------------------
2840
2841This would leave the merge result in `hello.c~2` file, along
2842with conflict markers if there are conflicts. After verifying
2843the merge result makes sense, you can tell git what the final
2844merge result for this file is by:
2845
2846-------------------------------------------------
2847$ mv -f hello.c~2 hello.c
2848$ git-update-index hello.c
2849-------------------------------------------------
2850
2851When a path is in unmerged state, running `git-update-index` for
2852that path tells git to mark the path resolved.
2853
2854The above is the description of a git merge at the lowest level,
2855to help you understand what conceptually happens under the hood.
2856In practice, nobody, not even git itself, uses three `git-cat-file`
2857for this. There is `git-merge-index` program that extracts the
2858stages to temporary files and calls a "merge" script on it:
2859
2860-------------------------------------------------
2861$ git-merge-index git-merge-one-file hello.c
2862-------------------------------------------------
2863
2864and that is what higher level `git resolve` is implemented with.
2865
2866How git stores objects efficiently: pack files
2867----------------------------------------------
2868
2869We've seen how git stores each object in a file named after the
2870object's SHA1 hash.
2871
2872Unfortunately this system becomes inefficient once a project has a
2873lot of objects. Try this on an old project:
2874
2875------------------------------------------------
2876$ git count-objects
28776930 objects, 47620 kilobytes
2878------------------------------------------------
2879
2880The first number is the number of objects which are kept in
2881individual files. The second is the amount of space taken up by
2882those "loose" objects.
2883
2884You can save space and make git faster by moving these loose objects in
2885to a "pack file", which stores a group of objects in an efficient
2886compressed format; the details of how pack files are formatted can be
2887found in link:technical/pack-format.txt[technical/pack-format.txt].
2888
2889To put the loose objects into a pack, just run git repack:
2890
2891------------------------------------------------
2892$ git repack
2893Generating pack...
2894Done counting 6020 objects.
2895Deltifying 6020 objects.
2896 100% (6020/6020) done
2897Writing 6020 objects.
2898 100% (6020/6020) done
2899Total 6020, written 6020 (delta 4070), reused 0 (delta 0)
2900Pack pack-3e54ad29d5b2e05838c75df582c65257b8d08e1c created.
2901------------------------------------------------
2902
2903You can then run
2904
2905------------------------------------------------
2906$ git prune
2907------------------------------------------------
2908
2909to remove any of the "loose" objects that are now contained in the
2910pack. This will also remove any unreferenced objects (which may be
2911created when, for example, you use "git reset" to remove a commit).
2912You can verify that the loose objects are gone by looking at the
2913.git/objects directory or by running
2914
2915------------------------------------------------
2916$ git count-objects
29170 objects, 0 kilobytes
2918------------------------------------------------
2919
2920Although the object files are gone, any commands that refer to those
2921objects will work exactly as they did before.
2922
2923The gitlink:git-gc[1] command performs packing, pruning, and more for
2924you, so is normally the only high-level command you need.
d19fbc3c 2925
21dcb3b7
BF
2926[[dangling-objects]]
2927Dangling objects
61b41790 2928----------------
21dcb3b7 2929
04e50e94 2930The gitlink:git-fsck[1] command will sometimes complain about dangling
21dcb3b7
BF
2931objects. They are not a problem.
2932
1191ee18
BF
2933The most common cause of dangling objects is that you've rebased a
2934branch, or you have pulled from somebody else who rebased a branch--see
2935<<cleaning-up-history>>. In that case, the old head of the original
2936branch still exists, as does obviously everything it pointed to. The
2937branch pointer itself just doesn't, since you replaced it with another
2938one.
2939
2940There are also other situations too that cause dangling objects. For
2941example, a "dangling blob" may arise because you did a "git add" of a
2942file, but then, before you actually committed it and made it part of the
2943bigger picture, you changed something else in that file and committed
2944that *updated* thing - the old state that you added originally ends up
2945not being pointed to by any commit or tree, so it's now a dangling blob
2946object.
2947
2948Similarly, when the "recursive" merge strategy runs, and finds that
2949there are criss-cross merges and thus more than one merge base (which is
2950fairly unusual, but it does happen), it will generate one temporary
2951midway tree (or possibly even more, if you had lots of criss-crossing
2952merges and more than two merge bases) as a temporary internal merge
2953base, and again, those are real objects, but the end result will not end
2954up pointing to them, so they end up "dangling" in your repository.
2955
2956Generally, dangling objects aren't anything to worry about. They can
2957even be very useful: if you screw something up, the dangling objects can
2958be how you recover your old tree (say, you did a rebase, and realized
2959that you really didn't want to - you can look at what dangling objects
2960you have, and decide to reset your head to some old dangling state).
21dcb3b7 2961
559e4d7a
BF
2962For commits, the most useful thing to do with dangling objects tends to
2963be to do a simple
21dcb3b7
BF
2964
2965------------------------------------------------
2966$ gitk <dangling-commit-sha-goes-here> --not --all
2967------------------------------------------------
2968
1191ee18
BF
2969For blobs and trees, you can't do the same, but you can examine them.
2970You can just do
21dcb3b7
BF
2971
2972------------------------------------------------
2973$ git show <dangling-blob/tree-sha-goes-here>
2974------------------------------------------------
2975
1191ee18
BF
2976to show what the contents of the blob were (or, for a tree, basically
2977what the "ls" for that directory was), and that may give you some idea
2978of what the operation was that left that dangling object.
21dcb3b7 2979
1191ee18
BF
2980Usually, dangling blobs and trees aren't very interesting. They're
2981almost always the result of either being a half-way mergebase (the blob
2982will often even have the conflict markers from a merge in it, if you
2983have had conflicting merges that you fixed up by hand), or simply
2984because you interrupted a "git fetch" with ^C or something like that,
2985leaving _some_ of the new objects in the object database, but just
2986dangling and useless.
21dcb3b7
BF
2987
2988Anyway, once you are sure that you're not interested in any dangling
2989state, you can just prune all unreachable objects:
2990
2991------------------------------------------------
2992$ git prune
2993------------------------------------------------
2994
1191ee18
BF
2995and they'll be gone. But you should only run "git prune" on a quiescent
2996repository - it's kind of like doing a filesystem fsck recovery: you
2997don't want to do that while the filesystem is mounted.
21dcb3b7 2998
04e50e94
BF
2999(The same is true of "git-fsck" itself, btw - but since
3000git-fsck never actually *changes* the repository, it just reports
3001on what it found, git-fsck itself is never "dangerous" to run.
21dcb3b7
BF
3002Running it while somebody is actually changing the repository can cause
3003confusing and scary messages, but it won't actually do anything bad. In
3004contrast, running "git prune" while somebody is actively changing the
3005repository is a *BAD* idea).
3006
d19fbc3c
BF
3007Glossary of git terms
3008=====================
3009
3010include::glossary.txt[]
3011
6bd9b682
BF
3012Notes and todo list for this manual
3013===================================
3014
3015This is a work in progress.
3016
3017The basic requirements:
2f99710c
BF
3018 - It must be readable in order, from beginning to end, by
3019 someone intelligent with a basic grasp of the unix
3020 commandline, but without any special knowledge of git. If
3021 necessary, any other prerequisites should be specifically
3022 mentioned as they arise.
3023 - Whenever possible, section headings should clearly describe
3024 the task they explain how to do, in language that requires
3025 no more knowledge than necessary: for example, "importing
3026 patches into a project" rather than "the git-am command"
6bd9b682 3027
d5cd5de4
BF
3028Think about how to create a clear chapter dependency graph that will
3029allow people to get to important topics without necessarily reading
3030everything in between.
d19fbc3c 3031
aacd404e
MC
3032Say something about .gitignore.
3033
d19fbc3c
BF
3034Scan Documentation/ for other stuff left out; in particular:
3035 howto's
d19fbc3c
BF
3036 some of technical/?
3037 hooks
0b375ab0 3038 list of commands in gitlink:git[1]
d19fbc3c
BF
3039
3040Scan email archives for other stuff left out
3041
3042Scan man pages to see if any assume more background than this manual
3043provides.
3044
2f99710c 3045Simplify beginning by suggesting disconnected head instead of
b181d57f 3046temporary branch creation?
d19fbc3c 3047
2f99710c
BF
3048Add more good examples. Entire sections of just cookbook examples
3049might be a good idea; maybe make an "advanced examples" section a
3050standard end-of-chapter section?
d19fbc3c
BF
3051
3052Include cross-references to the glossary, where appropriate.
3053
9a241220
BF
3054Document shallow clones? See draft 1.5.0 release notes for some
3055documentation.
3056
3dff5379 3057Add a section on working with other version control systems, including
9a241220
BF
3058CVS, Subversion, and just imports of series of release tarballs.
3059
a8cd1402 3060More details on gitweb?
0b375ab0
BF
3061
3062Write a chapter on using plumbing and writing scripts.