Add a birdview-on-the-source-code section to the user manual
[git/git.git] / Documentation / user-manual.txt
CommitLineData
71f4b183
CW
1Git User's Manual (for version 1.5.1 or newer)
2______________________________________________
d19fbc3c
BF
3
4This manual is designed to be readable by someone with basic unix
79c96c57 5command-line skills, but no previous knowledge of git.
d19fbc3c 6
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
e34caace 26[[git-quick-start]]
ef89f701
BF
27Git Quick Start
28===============
29
30This is a quick summary of the major commands; the following chapters
31will explain how these work in more detail.
32
e34caace 33[[quick-creating-a-new-repository]]
ef89f701
BF
34Creating a new repository
35-------------------------
36
37From a tarball:
38
39-----------------------------------------------
40$ tar xzf project.tar.gz
41$ cd project
42$ git init
43Initialized empty Git repository in .git/
44$ git add .
45$ git commit
46-----------------------------------------------
47
48From a remote repository:
49
50-----------------------------------------------
51$ git clone git://example.com/pub/project.git
52$ cd project
53-----------------------------------------------
54
e34caace 55[[managing-branches]]
ef89f701
BF
56Managing branches
57-----------------
58
59-----------------------------------------------
c64415e2 60$ git branch # list all local branches in this repo
ef89f701
BF
61$ git checkout test # switch working directory to branch "test"
62$ git branch new # create branch "new" starting at current HEAD
63$ git branch -d new # delete branch "new"
64-----------------------------------------------
65
66Instead of basing new branch on current HEAD (the default), use:
67
68-----------------------------------------------
69$ git branch new test # branch named "test"
70$ git branch new v2.6.15 # tag named v2.6.15
71$ git branch new HEAD^ # commit before the most recent
72$ git branch new HEAD^^ # commit before that
73$ git branch new test~10 # ten commits before tip of branch "test"
74-----------------------------------------------
75
76Create and switch to a new branch at the same time:
77
78-----------------------------------------------
79$ git checkout -b new v2.6.15
80-----------------------------------------------
81
82Update and examine branches from the repository you cloned from:
83
84-----------------------------------------------
85$ git fetch # update
86$ git branch -r # list
87 origin/master
88 origin/next
89 ...
04483524 90$ git checkout -b masterwork origin/master
ef89f701
BF
91-----------------------------------------------
92
93Fetch a branch from a different repository, and give it a new
94name in your repository:
95
96-----------------------------------------------
97$ git fetch git://example.com/project.git theirbranch:mybranch
98$ git fetch git://example.com/project.git v2.6.15:mybranch
99-----------------------------------------------
100
101Keep a list of repositories you work with regularly:
102
103-----------------------------------------------
104$ git remote add example git://example.com/project.git
b181d57f 105$ git remote # list remote repositories
ef89f701
BF
106example
107origin
b181d57f 108$ git remote show example # get details
ef89f701
BF
109* remote example
110 URL: git://example.com/project.git
111 Tracked remote branches
112 master next ...
b181d57f
BF
113$ git fetch example # update branches from example
114$ git branch -r # list all remote branches
ef89f701
BF
115-----------------------------------------------
116
117
e34caace 118[[exploring-history]]
ef89f701
BF
119Exploring history
120-----------------
121
122-----------------------------------------------
123$ gitk # visualize and browse history
124$ git log # list all commits
125$ git log src/ # ...modifying src/
126$ git log v2.6.15..v2.6.16 # ...in v2.6.16, not in v2.6.15
127$ git log master..test # ...in branch test, not in branch master
128$ git log test..master # ...in branch master, but not in test
129$ git log test...master # ...in one branch, not in both
130$ git log -S'foo()' # ...where difference contain "foo()"
131$ git log --since="2 weeks ago"
132$ git log -p # show patches as well
133$ git show # most recent commit
134$ git diff v2.6.15..v2.6.16 # diff between two tagged versions
135$ git diff v2.6.15..HEAD # diff with current head
136$ git grep "foo()" # search working directory for "foo()"
137$ git grep v2.6.15 "foo()" # search old tree for "foo()"
138$ git show v2.6.15:a.txt # look at old version of a.txt
139-----------------------------------------------
140
b181d57f 141Search for regressions:
ef89f701
BF
142
143-----------------------------------------------
144$ git bisect start
145$ git bisect bad # current version is bad
146$ git bisect good v2.6.13-rc2 # last known good revision
147Bisecting: 675 revisions left to test after this
148 # test here, then:
149$ git bisect good # if this revision is good, or
150$ git bisect bad # if this revision is bad.
151 # repeat until done.
152-----------------------------------------------
153
e34caace 154[[making-changes]]
ef89f701
BF
155Making changes
156--------------
157
158Make sure git knows who to blame:
159
160------------------------------------------------
58c19d1f 161$ cat >>~/.gitconfig <<\EOF
ef89f701 162[user]
04483524
JN
163 name = Your Name Comes Here
164 email = you@yourdomain.example.com
ef89f701
BF
165EOF
166------------------------------------------------
167
168Select file contents to include in the next commit, then make the
169commit:
170
171-----------------------------------------------
172$ git add a.txt # updated file
173$ git add b.txt # new file
174$ git rm c.txt # old file
175$ git commit
176-----------------------------------------------
177
178Or, prepare and create the commit in one step:
179
180-----------------------------------------------
b181d57f 181$ git commit d.txt # use latest content only of d.txt
ef89f701
BF
182$ git commit -a # use latest content of all tracked files
183-----------------------------------------------
184
e34caace 185[[merging]]
ef89f701
BF
186Merging
187-------
188
189-----------------------------------------------
190$ git merge test # merge branch "test" into the current branch
191$ git pull git://example.com/project.git master
192 # fetch and merge in remote branch
193$ git pull . test # equivalent to git merge test
194-----------------------------------------------
195
e34caace 196[[sharing-your-changes]]
e4add70c
BF
197Sharing your changes
198--------------------
ef89f701
BF
199
200Importing or exporting patches:
201
202-----------------------------------------------
203$ git format-patch origin..HEAD # format a patch for each commit
204 # in HEAD but not in origin
04483524 205$ git am mbox # import patches from the mailbox "mbox"
ef89f701
BF
206-----------------------------------------------
207
ef89f701
BF
208Fetch a branch in a different git repository, then merge into the
209current branch:
210
211-----------------------------------------------
212$ git pull git://example.com/project.git theirbranch
213-----------------------------------------------
214
215Store the fetched branch into a local branch before merging into the
216current branch:
217
218-----------------------------------------------
219$ git pull git://example.com/project.git theirbranch:mybranch
220-----------------------------------------------
221
e4add70c
BF
222After creating commits on a local branch, update the remote
223branch with your commits:
224
225-----------------------------------------------
226$ git push ssh://example.com/project.git mybranch:theirbranch
227-----------------------------------------------
228
229When remote and local branch are both named "test":
230
231-----------------------------------------------
232$ git push ssh://example.com/project.git test
233-----------------------------------------------
234
235Shortcut version for a frequently used remote repository:
236
237-----------------------------------------------
238$ git remote add example ssh://example.com/project.git
239$ git push example test
240-----------------------------------------------
241
e34caace 242[[repository-maintenance]]
b181d57f
BF
243Repository maintenance
244----------------------
245
246Check for corruption:
247
248-----------------------------------------------
04e50e94 249$ git fsck
b181d57f
BF
250-----------------------------------------------
251
252Recompress, remove unused cruft:
253
254-----------------------------------------------
255$ git gc
256-----------------------------------------------
257
e34caace 258[[repositories-and-branches]]
d19fbc3c
BF
259Repositories and Branches
260=========================
261
e34caace 262[[how-to-get-a-git-repository]]
d19fbc3c
BF
263How to get a git repository
264---------------------------
265
266It will be useful to have a git repository to experiment with as you
267read this manual.
268
269The best way to get one is by using the gitlink:git-clone[1] command
270to download a copy of an existing repository for a project that you
271are interested in. If you don't already have a project in mind, here
272are some interesting examples:
273
274------------------------------------------------
275 # git itself (approx. 10MB download):
276$ git clone git://git.kernel.org/pub/scm/git/git.git
277 # the linux kernel (approx. 150MB download):
278$ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
279------------------------------------------------
280
281The initial clone may be time-consuming for a large project, but you
282will only need to clone once.
283
284The clone command creates a new directory named after the project
285("git" or "linux-2.6" in the examples above). After you cd into this
286directory, you will see that it contains a copy of the project files,
287together with a special top-level directory named ".git", which
288contains all the information about the history of the project.
289
d5cd5de4 290In most of the following, examples will be taken from one of the two
d19fbc3c
BF
291repositories above.
292
e34caace 293[[how-to-check-out]]
d19fbc3c
BF
294How to check out a different version of a project
295-------------------------------------------------
296
297Git is best thought of as a tool for storing the history of a
298collection of files. It stores the history as a compressed
299collection of interrelated snapshots (versions) of the project's
300contents.
301
81b6c950
BF
302A single git repository may contain multiple branches. It keeps track
303of them by keeping a list of <<def_head,heads>> which reference the
304latest version on each branch; the gitlink:git-branch[1] command shows
305you the list of branch heads:
d19fbc3c
BF
306
307------------------------------------------------
308$ git branch
309* master
310------------------------------------------------
311
4f752407
BF
312A freshly cloned repository contains a single branch head, by default
313named "master", with the working directory initialized to the state of
314the project referred to by that branch head.
d19fbc3c 315
81b6c950
BF
316Most projects also use <<def_tag,tags>>. Tags, like heads, are
317references into the project's history, and can be listed using the
d19fbc3c
BF
318gitlink:git-tag[1] command:
319
320------------------------------------------------
321$ git tag -l
322v2.6.11
323v2.6.11-tree
324v2.6.12
325v2.6.12-rc2
326v2.6.12-rc3
327v2.6.12-rc4
328v2.6.12-rc5
329v2.6.12-rc6
330v2.6.13
331...
332------------------------------------------------
333
fe4b3e59 334Tags are expected to always point at the same version of a project,
81b6c950 335while heads are expected to advance as development progresses.
fe4b3e59 336
81b6c950 337Create a new branch head pointing to one of these versions and check it
d19fbc3c
BF
338out using gitlink:git-checkout[1]:
339
340------------------------------------------------
341$ git checkout -b new v2.6.13
342------------------------------------------------
343
344The working directory then reflects the contents that the project had
345when it was tagged v2.6.13, and gitlink:git-branch[1] shows two
346branches, with an asterisk marking the currently checked-out branch:
347
348------------------------------------------------
349$ git branch
350 master
351* new
352------------------------------------------------
353
354If you decide that you'd rather see version 2.6.17, you can modify
355the current branch to point at v2.6.17 instead, with
356
357------------------------------------------------
358$ git reset --hard v2.6.17
359------------------------------------------------
360
81b6c950 361Note that if the current branch head was your only reference to a
d19fbc3c 362particular point in history, then resetting that branch may leave you
81b6c950
BF
363with no way to find the history it used to point to; so use this command
364carefully.
d19fbc3c 365
e34caace 366[[understanding-commits]]
d19fbc3c
BF
367Understanding History: Commits
368------------------------------
369
370Every change in the history of a project is represented by a commit.
371The gitlink:git-show[1] command shows the most recent commit on the
372current branch:
373
374------------------------------------------------
375$ git show
376commit 2b5f6dcce5bf94b9b119e9ed8d537098ec61c3d2
377Author: Jamal Hadi Salim <hadi@cyberus.ca>
378Date: Sat Dec 2 22:22:25 2006 -0800
379
380 [XFRM]: Fix aevent structuring to be more complete.
381
382 aevents can not uniquely identify an SA. We break the ABI with this
383 patch, but consensus is that since it is not yet utilized by any
384 (known) application then it is fine (better do it now than later).
385
386 Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
387 Signed-off-by: David S. Miller <davem@davemloft.net>
388
389diff --git a/Documentation/networking/xfrm_sync.txt b/Documentation/networking/xfrm_sync.txt
390index 8be626f..d7aac9d 100644
391--- a/Documentation/networking/xfrm_sync.txt
392+++ b/Documentation/networking/xfrm_sync.txt
393@@ -47,10 +47,13 @@ aevent_id structure looks like:
394
395 struct xfrm_aevent_id {
396 struct xfrm_usersa_id sa_id;
397+ xfrm_address_t saddr;
398 __u32 flags;
399+ __u32 reqid;
400 };
401...
402------------------------------------------------
403
404As you can see, a commit shows who made the latest change, what they
405did, and why.
406
35121930
BF
407Every commit has a 40-hexdigit id, sometimes called the "object name" or the
408"SHA1 id", shown on the first line of the "git show" output. You can usually
409refer to a commit by a shorter name, such as a tag or a branch name, but this
410longer name can also be useful. Most importantly, it is a globally unique
411name for this commit: so if you tell somebody else the object name (for
412example in email), then you are guaranteed that name will refer to the same
413commit in their repository that it does in yours (assuming their repository
414has that commit at all). Since the object name is computed as a hash over the
415contents of the commit, you are guaranteed that the commit can never change
416without its name also changing.
417
418In fact, in <<git-internals>> we shall see that everything stored in git
419history, including file data and directory contents, is stored in an object
420with a name that is a hash of its contents.
d19fbc3c 421
e34caace 422[[understanding-reachability]]
d19fbc3c
BF
423Understanding history: commits, parents, and reachability
424~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
425
426Every commit (except the very first commit in a project) also has a
427parent commit which shows what happened before this commit.
428Following the chain of parents will eventually take you back to the
429beginning of the project.
430
431However, the commits do not form a simple list; git allows lines of
432development to diverge and then reconverge, and the point where two
433lines of development reconverge is called a "merge". The commit
434representing a merge can therefore have more than one parent, with
435each parent representing the most recent commit on one of the lines
436of development leading to that point.
437
438The best way to see how this works is using the gitlink:gitk[1]
439command; running gitk now on a git repository and looking for merge
440commits will help understand how the git organizes history.
441
442In the following, we say that commit X is "reachable" from commit Y
443if commit X is an ancestor of commit Y. Equivalently, you could say
444that Y is a descendent of X, or that there is a chain of parents
445leading from commit Y to commit X.
446
e34caace 447[[history-diagrams]]
3dff5379
PR
448Understanding history: History diagrams
449~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c
BF
450
451We will sometimes represent git history using diagrams like the one
452below. Commits are shown as "o", and the links between them with
453lines drawn with - / and \. Time goes left to right:
454
1dc71a91
BF
455
456................................................
d19fbc3c
BF
457 o--o--o <-- Branch A
458 /
459 o--o--o <-- master
460 \
461 o--o--o <-- Branch B
1dc71a91 462................................................
d19fbc3c
BF
463
464If we need to talk about a particular commit, the character "o" may
465be replaced with another letter or number.
466
e34caace 467[[what-is-a-branch]]
d19fbc3c
BF
468Understanding history: What is a branch?
469~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
470
81b6c950
BF
471When we need to be precise, we will use the word "branch" to mean a line
472of development, and "branch head" (or just "head") to mean a reference
473to the most recent commit on a branch. In the example above, the branch
474head named "A" is a pointer to one particular commit, but we refer to
475the line of three commits leading up to that point as all being part of
d19fbc3c
BF
476"branch A".
477
81b6c950
BF
478However, when no confusion will result, we often just use the term
479"branch" both for branches and for branch heads.
d19fbc3c 480
e34caace 481[[manipulating-branches]]
d19fbc3c
BF
482Manipulating branches
483---------------------
484
485Creating, deleting, and modifying branches is quick and easy; here's
486a summary of the commands:
487
488git branch::
489 list all branches
490git branch <branch>::
491 create a new branch named <branch>, referencing the same
492 point in history as the current branch
493git branch <branch> <start-point>::
494 create a new branch named <branch>, referencing
495 <start-point>, which may be specified any way you like,
496 including using a branch name or a tag name
497git branch -d <branch>::
498 delete the branch <branch>; if the branch you are deleting
c64415e2
BF
499 points to a commit which is not reachable from the current
500 branch, this command will fail with a warning.
d19fbc3c
BF
501git branch -D <branch>::
502 even if the branch points to a commit not reachable
503 from the current branch, you may know that that commit
504 is still reachable from some other branch or tag. In that
505 case it is safe to use this command to force git to delete
506 the branch.
507git checkout <branch>::
508 make the current branch <branch>, updating the working
509 directory to reflect the version referenced by <branch>
510git checkout -b <new> <start-point>::
511 create a new branch <new> referencing <start-point>, and
512 check it out.
513
72a76c95
BF
514The special symbol "HEAD" can always be used to refer to the current
515branch. In fact, git uses a file named "HEAD" in the .git directory to
516remember which branch is current:
517
518------------------------------------------------
519$ cat .git/HEAD
520ref: refs/heads/master
521------------------------------------------------
522
25d9f3fa 523[[detached-head]]
72a76c95
BF
524Examining an old version without creating a new branch
525------------------------------------------------------
526
527The git-checkout command normally expects a branch head, but will also
528accept an arbitrary commit; for example, you can check out the commit
529referenced by a tag:
530
531------------------------------------------------
532$ git checkout v2.6.17
533Note: moving to "v2.6.17" which isn't a local branch
534If you want to create a new branch from this checkout, you may do so
535(now or later) by using -b with the checkout command again. Example:
536 git checkout -b <new_branch_name>
537HEAD is now at 427abfa... Linux v2.6.17
538------------------------------------------------
539
540The HEAD then refers to the SHA1 of the commit instead of to a branch,
541and git branch shows that you are no longer on a branch:
542
543------------------------------------------------
544$ cat .git/HEAD
545427abfa28afedffadfca9dd8b067eb6d36bac53f
953f3d6f 546$ git branch
72a76c95
BF
547* (no branch)
548 master
549------------------------------------------------
550
551In this case we say that the HEAD is "detached".
552
953f3d6f
BF
553This is an easy way to check out a particular version without having to
554make up a name for the new branch. You can still create a new branch
555(or tag) for this version later if you decide to.
d19fbc3c 556
e34caace 557[[examining-remote-branches]]
d19fbc3c
BF
558Examining branches from a remote repository
559-------------------------------------------
560
561The "master" branch that was created at the time you cloned is a copy
562of the HEAD in the repository that you cloned from. That repository
563may also have had other branches, though, and your local repository
564keeps branches which track each of those remote branches, which you
565can view using the "-r" option to gitlink:git-branch[1]:
566
567------------------------------------------------
568$ git branch -r
569 origin/HEAD
570 origin/html
571 origin/maint
572 origin/man
573 origin/master
574 origin/next
575 origin/pu
576 origin/todo
577------------------------------------------------
578
579You cannot check out these remote-tracking branches, but you can
580examine them on a branch of your own, just as you would a tag:
581
582------------------------------------------------
583$ git checkout -b my-todo-copy origin/todo
584------------------------------------------------
585
586Note that the name "origin" is just the name that git uses by default
587to refer to the repository that you cloned from.
588
589[[how-git-stores-references]]
f60b9642
BF
590Naming branches, tags, and other references
591-------------------------------------------
d19fbc3c
BF
592
593Branches, remote-tracking branches, and tags are all references to
f60b9642
BF
594commits. All references are named with a slash-separated path name
595starting with "refs"; the names we've been using so far are actually
596shorthand:
d19fbc3c 597
f60b9642
BF
598 - The branch "test" is short for "refs/heads/test".
599 - The tag "v2.6.18" is short for "refs/tags/v2.6.18".
600 - "origin/master" is short for "refs/remotes/origin/master".
d19fbc3c 601
f60b9642
BF
602The full name is occasionally useful if, for example, there ever
603exists a tag and a branch with the same name.
d19fbc3c 604
c64415e2
BF
605As another useful shortcut, the "HEAD" of a repository can be referred
606to just using the name of that repository. So, for example, "origin"
607is usually a shortcut for the HEAD branch in the repository "origin".
d19fbc3c
BF
608
609For the complete list of paths which git checks for references, and
f60b9642
BF
610the order it uses to decide which to choose when there are multiple
611references with the same shorthand name, see the "SPECIFYING
612REVISIONS" section of gitlink:git-rev-parse[1].
d19fbc3c
BF
613
614[[Updating-a-repository-with-git-fetch]]
615Updating a repository with git fetch
616------------------------------------
617
618Eventually the developer cloned from will do additional work in her
619repository, creating new commits and advancing the branches to point
620at the new commits.
621
622The command "git fetch", with no arguments, will update all of the
623remote-tracking branches to the latest version found in her
624repository. It will not touch any of your own branches--not even the
625"master" branch that was created for you on clone.
626
e34caace 627[[fetching-branches]]
d5cd5de4
BF
628Fetching branches from other repositories
629-----------------------------------------
630
631You can also track branches from repositories other than the one you
632cloned from, using gitlink:git-remote[1]:
633
634-------------------------------------------------
635$ git remote add linux-nfs git://linux-nfs.org/pub/nfs-2.6.git
04483524 636$ git fetch linux-nfs
d5cd5de4
BF
637* refs/remotes/linux-nfs/master: storing branch 'master' ...
638 commit: bf81b46
639-------------------------------------------------
640
641New remote-tracking branches will be stored under the shorthand name
642that you gave "git remote add", in this case linux-nfs:
643
644-------------------------------------------------
645$ git branch -r
646linux-nfs/master
647origin/master
648-------------------------------------------------
649
650If you run "git fetch <remote>" later, the tracking branches for the
651named <remote> will be updated.
652
653If you examine the file .git/config, you will see that git has added
654a new stanza:
655
656-------------------------------------------------
657$ cat .git/config
658...
659[remote "linux-nfs"]
923642fe
BF
660 url = git://linux-nfs.org/pub/nfs-2.6.git
661 fetch = +refs/heads/*:refs/remotes/linux-nfs/*
d5cd5de4
BF
662...
663-------------------------------------------------
664
fc90c536
BF
665This is what causes git to track the remote's branches; you may modify
666or delete these configuration options by editing .git/config with a
667text editor. (See the "CONFIGURATION FILE" section of
668gitlink:git-config[1] for details.)
d5cd5de4 669
e34caace 670[[exploring-git-history]]
d19fbc3c
BF
671Exploring git history
672=====================
673
674Git is best thought of as a tool for storing the history of a
675collection of files. It does this by storing compressed snapshots of
676the contents of a file heirarchy, together with "commits" which show
677the relationships between these snapshots.
678
679Git provides extremely flexible and fast tools for exploring the
680history of a project.
681
aacd404e 682We start with one specialized tool that is useful for finding the
d19fbc3c
BF
683commit that introduced a bug into a project.
684
e34caace 685[[using-bisect]]
d19fbc3c
BF
686How to use bisect to find a regression
687--------------------------------------
688
689Suppose version 2.6.18 of your project worked, but the version at
690"master" crashes. Sometimes the best way to find the cause of such a
691regression is to perform a brute-force search through the project's
692history to find the particular commit that caused the problem. The
693gitlink:git-bisect[1] command can help you do this:
694
695-------------------------------------------------
696$ git bisect start
697$ git bisect good v2.6.18
698$ git bisect bad master
699Bisecting: 3537 revisions left to test after this
700[65934a9a028b88e83e2b0f8b36618fe503349f8e] BLOCK: Make USB storage depend on SCSI rather than selecting it [try #6]
701-------------------------------------------------
702
703If you run "git branch" at this point, you'll see that git has
704temporarily moved you to a new branch named "bisect". This branch
705points to a commit (with commit id 65934...) that is reachable from
706v2.6.19 but not from v2.6.18. Compile and test it, and see whether
707it crashes. Assume it does crash. Then:
708
709-------------------------------------------------
710$ git bisect bad
711Bisecting: 1769 revisions left to test after this
712[7eff82c8b1511017ae605f0c99ac275a7e21b867] i2c-core: Drop useless bitmaskings
713-------------------------------------------------
714
715checks out an older version. Continue like this, telling git at each
716stage whether the version it gives you is good or bad, and notice
717that the number of revisions left to test is cut approximately in
718half each time.
719
720After about 13 tests (in this case), it will output the commit id of
721the guilty commit. You can then examine the commit with
722gitlink:git-show[1], find out who wrote it, and mail them your bug
723report with the commit id. Finally, run
724
725-------------------------------------------------
726$ git bisect reset
727-------------------------------------------------
728
729to return you to the branch you were on before and delete the
730temporary "bisect" branch.
731
732Note that the version which git-bisect checks out for you at each
733point is just a suggestion, and you're free to try a different
734version if you think it would be a good idea. For example,
735occasionally you may land on a commit that broke something unrelated;
736run
737
738-------------------------------------------------
04483524 739$ git bisect visualize
d19fbc3c
BF
740-------------------------------------------------
741
742which will run gitk and label the commit it chose with a marker that
743says "bisect". Chose a safe-looking commit nearby, note its commit
744id, and check it out with:
745
746-------------------------------------------------
747$ git reset --hard fb47ddb2db...
748-------------------------------------------------
749
750then test, run "bisect good" or "bisect bad" as appropriate, and
751continue.
752
e34caace 753[[naming-commits]]
d19fbc3c
BF
754Naming commits
755--------------
756
757We have seen several ways of naming commits already:
758
d55ae921 759 - 40-hexdigit object name
d19fbc3c
BF
760 - branch name: refers to the commit at the head of the given
761 branch
762 - tag name: refers to the commit pointed to by the given tag
763 (we've seen branches and tags are special cases of
764 <<how-git-stores-references,references>>).
765 - HEAD: refers to the head of the current branch
766
eb6ae7f4 767There are many more; see the "SPECIFYING REVISIONS" section of the
aec053bb 768gitlink:git-rev-parse[1] man page for the complete list of ways to
d19fbc3c
BF
769name revisions. Some examples:
770
771-------------------------------------------------
d55ae921 772$ git show fb47ddb2 # the first few characters of the object name
d19fbc3c
BF
773 # are usually enough to specify it uniquely
774$ git show HEAD^ # the parent of the HEAD commit
775$ git show HEAD^^ # the grandparent
776$ git show HEAD~4 # the great-great-grandparent
777-------------------------------------------------
778
779Recall that merge commits may have more than one parent; by default,
780^ and ~ follow the first parent listed in the commit, but you can
781also choose:
782
783-------------------------------------------------
784$ git show HEAD^1 # show the first parent of HEAD
785$ git show HEAD^2 # show the second parent of HEAD
786-------------------------------------------------
787
788In addition to HEAD, there are several other special names for
789commits:
790
791Merges (to be discussed later), as well as operations such as
792git-reset, which change the currently checked-out commit, generally
793set ORIG_HEAD to the value HEAD had before the current operation.
794
795The git-fetch operation always stores the head of the last fetched
796branch in FETCH_HEAD. For example, if you run git fetch without
797specifying a local branch as the target of the operation
798
799-------------------------------------------------
800$ git fetch git://example.com/proj.git theirbranch
801-------------------------------------------------
802
803the fetched commits will still be available from FETCH_HEAD.
804
805When we discuss merges we'll also see the special name MERGE_HEAD,
806which refers to the other branch that we're merging in to the current
807branch.
808
aec053bb 809The gitlink:git-rev-parse[1] command is a low-level command that is
d55ae921
BF
810occasionally useful for translating some name for a commit to the object
811name for that commit:
aec053bb
BF
812
813-------------------------------------------------
814$ git rev-parse origin
815e05db0fd4f31dde7005f075a84f96b360d05984b
816-------------------------------------------------
817
e34caace 818[[creating-tags]]
d19fbc3c
BF
819Creating tags
820-------------
821
822We can also create a tag to refer to a particular commit; after
823running
824
825-------------------------------------------------
04483524 826$ git tag stable-1 1b2e1d63ff
d19fbc3c
BF
827-------------------------------------------------
828
829You can use stable-1 to refer to the commit 1b2e1d63ff.
830
c64415e2
BF
831This creates a "lightweight" tag. If you would also like to include a
832comment with the tag, and possibly sign it cryptographically, then you
833should create a tag object instead; see the gitlink:git-tag[1] man page
834for details.
d19fbc3c 835
e34caace 836[[browsing-revisions]]
d19fbc3c
BF
837Browsing revisions
838------------------
839
840The gitlink:git-log[1] command can show lists of commits. On its
841own, it shows all commits reachable from the parent commit; but you
842can also make more specific requests:
843
844-------------------------------------------------
845$ git log v2.5.. # commits since (not reachable from) v2.5
846$ git log test..master # commits reachable from master but not test
847$ git log master..test # ...reachable from test but not master
848$ git log master...test # ...reachable from either test or master,
849 # but not both
850$ git log --since="2 weeks ago" # commits from the last 2 weeks
851$ git log Makefile # commits which modify Makefile
852$ git log fs/ # ... which modify any file under fs/
853$ git log -S'foo()' # commits which add or remove any file data
854 # matching the string 'foo()'
855-------------------------------------------------
856
857And of course you can combine all of these; the following finds
858commits since v2.5 which touch the Makefile or any file under fs:
859
860-------------------------------------------------
861$ git log v2.5.. Makefile fs/
862-------------------------------------------------
863
864You can also ask git log to show patches:
865
866-------------------------------------------------
867$ git log -p
868-------------------------------------------------
869
870See the "--pretty" option in the gitlink:git-log[1] man page for more
871display options.
872
873Note that git log starts with the most recent commit and works
874backwards through the parents; however, since git history can contain
3dff5379 875multiple independent lines of development, the particular order that
d19fbc3c
BF
876commits are listed in may be somewhat arbitrary.
877
e34caace 878[[generating-diffs]]
d19fbc3c
BF
879Generating diffs
880----------------
881
882You can generate diffs between any two versions using
883gitlink:git-diff[1]:
884
885-------------------------------------------------
886$ git diff master..test
887-------------------------------------------------
888
889Sometimes what you want instead is a set of patches:
890
891-------------------------------------------------
892$ git format-patch master..test
893-------------------------------------------------
894
895will generate a file with a patch for each commit reachable from test
896but not from master. Note that if master also has commits which are
897not reachable from test, then the combined result of these patches
898will not be the same as the diff produced by the git-diff example.
899
e34caace 900[[viewing-old-file-versions]]
d19fbc3c
BF
901Viewing old file versions
902-------------------------
903
904You can always view an old version of a file by just checking out the
905correct revision first. But sometimes it is more convenient to be
906able to view an old version of a single file without checking
907anything out; this command does that:
908
909-------------------------------------------------
910$ git show v2.5:fs/locks.c
911-------------------------------------------------
912
913Before the colon may be anything that names a commit, and after it
914may be any path to a file tracked by git.
915
e34caace 916[[history-examples]]
aec053bb
BF
917Examples
918--------
919
e34caace 920[[checking-for-equal-branches]]
aec053bb 921Check whether two branches point at the same history
2f99710c 922~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
aec053bb
BF
923
924Suppose you want to check whether two branches point at the same point
925in history.
926
927-------------------------------------------------
928$ git diff origin..master
929-------------------------------------------------
930
69f7ad73
BF
931will tell you whether the contents of the project are the same at the
932two branches; in theory, however, it's possible that the same project
933contents could have been arrived at by two different historical
d55ae921 934routes. You could compare the object names:
aec053bb
BF
935
936-------------------------------------------------
937$ git rev-list origin
938e05db0fd4f31dde7005f075a84f96b360d05984b
939$ git rev-list master
940e05db0fd4f31dde7005f075a84f96b360d05984b
941-------------------------------------------------
942
69f7ad73
BF
943Or you could recall that the ... operator selects all commits
944contained reachable from either one reference or the other but not
945both: so
aec053bb
BF
946
947-------------------------------------------------
948$ git log origin...master
949-------------------------------------------------
950
951will return no commits when the two branches are equal.
952
e34caace 953[[finding-tagged-descendants]]
b181d57f
BF
954Find first tagged version including a given fix
955~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
aec053bb 956
69f7ad73
BF
957Suppose you know that the commit e05db0fd fixed a certain problem.
958You'd like to find the earliest tagged release that contains that
959fix.
960
961Of course, there may be more than one answer--if the history branched
962after commit e05db0fd, then there could be multiple "earliest" tagged
963releases.
964
965You could just visually inspect the commits since e05db0fd:
966
967-------------------------------------------------
968$ gitk e05db0fd..
969-------------------------------------------------
970
b181d57f
BF
971Or you can use gitlink:git-name-rev[1], which will give the commit a
972name based on any tag it finds pointing to one of the commit's
973descendants:
974
975-------------------------------------------------
04483524 976$ git name-rev --tags e05db0fd
b181d57f
BF
977e05db0fd tags/v1.5.0-rc1^0~23
978-------------------------------------------------
979
980The gitlink:git-describe[1] command does the opposite, naming the
981revision using a tag on which the given commit is based:
982
983-------------------------------------------------
984$ git describe e05db0fd
04483524 985v1.5.0-rc0-260-ge05db0f
b181d57f
BF
986-------------------------------------------------
987
988but that may sometimes help you guess which tags might come after the
989given commit.
990
991If you just want to verify whether a given tagged version contains a
992given commit, you could use gitlink:git-merge-base[1]:
993
994-------------------------------------------------
995$ git merge-base e05db0fd v1.5.0-rc1
996e05db0fd4f31dde7005f075a84f96b360d05984b
997-------------------------------------------------
998
999The merge-base command finds a common ancestor of the given commits,
1000and always returns one or the other in the case where one is a
1001descendant of the other; so the above output shows that e05db0fd
1002actually is an ancestor of v1.5.0-rc1.
1003
1004Alternatively, note that
1005
1006-------------------------------------------------
4a7979ca 1007$ git log v1.5.0-rc1..e05db0fd
b181d57f
BF
1008-------------------------------------------------
1009
4a7979ca 1010will produce empty output if and only if v1.5.0-rc1 includes e05db0fd,
b181d57f 1011because it outputs only commits that are not reachable from v1.5.0-rc1.
aec053bb 1012
4a7979ca
BF
1013As yet another alternative, the gitlink:git-show-branch[1] command lists
1014the commits reachable from its arguments with a display on the left-hand
1015side that indicates which arguments that commit is reachable from. So,
1016you can run something like
1017
1018-------------------------------------------------
1019$ git show-branch e05db0fd v1.5.0-rc0 v1.5.0-rc1 v1.5.0-rc2
1020! [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
1021available
1022 ! [v1.5.0-rc0] GIT v1.5.0 preview
1023 ! [v1.5.0-rc1] GIT v1.5.0-rc1
1024 ! [v1.5.0-rc2] GIT v1.5.0-rc2
1025...
1026-------------------------------------------------
1027
1028then search for a line that looks like
1029
1030-------------------------------------------------
1031+ ++ [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
1032available
1033-------------------------------------------------
1034
1035Which shows that e05db0fd is reachable from itself, from v1.5.0-rc1, and
1036from v1.5.0-rc2, but not from v1.5.0-rc0.
1037
1038
e34caace 1039[[Developing-with-git]]
d19fbc3c
BF
1040Developing with git
1041===================
1042
e34caace 1043[[telling-git-your-name]]
d19fbc3c
BF
1044Telling git your name
1045---------------------
1046
1047Before creating any commits, you should introduce yourself to git. The
58c19d1f
BF
1048easiest way to do so is to make sure the following lines appear in a
1049file named .gitconfig in your home directory:
d19fbc3c
BF
1050
1051------------------------------------------------
d19fbc3c
BF
1052[user]
1053 name = Your Name Comes Here
1054 email = you@yourdomain.example.com
d19fbc3c
BF
1055------------------------------------------------
1056
fc90c536
BF
1057(See the "CONFIGURATION FILE" section of gitlink:git-config[1] for
1058details on the configuration file.)
1059
d19fbc3c 1060
e34caace 1061[[creating-a-new-repository]]
d19fbc3c
BF
1062Creating a new repository
1063-------------------------
1064
1065Creating a new repository from scratch is very easy:
1066
1067-------------------------------------------------
1068$ mkdir project
1069$ cd project
f1d2b477 1070$ git init
d19fbc3c
BF
1071-------------------------------------------------
1072
1073If you have some initial content (say, a tarball):
1074
1075-------------------------------------------------
1076$ tar -xzvf project.tar.gz
1077$ cd project
f1d2b477 1078$ git init
d19fbc3c
BF
1079$ git add . # include everything below ./ in the first commit:
1080$ git commit
1081-------------------------------------------------
1082
1083[[how-to-make-a-commit]]
ae25c67a 1084How to make a commit
d19fbc3c
BF
1085--------------------
1086
1087Creating a new commit takes three steps:
1088
1089 1. Making some changes to the working directory using your
1090 favorite editor.
1091 2. Telling git about your changes.
1092 3. Creating the commit using the content you told git about
1093 in step 2.
1094
1095In practice, you can interleave and repeat steps 1 and 2 as many
1096times as you want: in order to keep track of what you want committed
1097at step 3, git maintains a snapshot of the tree's contents in a
1098special staging area called "the index."
1099
01997b4a
BF
1100At the beginning, the content of the index will be identical to
1101that of the HEAD. The command "git diff --cached", which shows
1102the difference between the HEAD and the index, should therefore
1103produce no output at that point.
eb6ae7f4 1104
d19fbc3c
BF
1105Modifying the index is easy:
1106
1107To update the index with the new contents of a modified file, use
1108
1109-------------------------------------------------
1110$ git add path/to/file
1111-------------------------------------------------
1112
1113To add the contents of a new file to the index, use
1114
1115-------------------------------------------------
1116$ git add path/to/file
1117-------------------------------------------------
1118
eb6ae7f4 1119To remove a file from the index and from the working tree,
d19fbc3c
BF
1120
1121-------------------------------------------------
1122$ git rm path/to/file
1123-------------------------------------------------
1124
1125After each step you can verify that
1126
1127-------------------------------------------------
1128$ git diff --cached
1129-------------------------------------------------
1130
1131always shows the difference between the HEAD and the index file--this
1132is what you'd commit if you created the commit now--and that
1133
1134-------------------------------------------------
1135$ git diff
1136-------------------------------------------------
1137
1138shows the difference between the working tree and the index file.
1139
1140Note that "git add" always adds just the current contents of a file
1141to the index; further changes to the same file will be ignored unless
1142you run git-add on the file again.
1143
1144When you're ready, just run
1145
1146-------------------------------------------------
1147$ git commit
1148-------------------------------------------------
1149
1150and git will prompt you for a commit message and then create the new
3dff5379 1151commit. Check to make sure it looks like what you expected with
d19fbc3c
BF
1152
1153-------------------------------------------------
1154$ git show
1155-------------------------------------------------
1156
1157As a special shortcut,
1158
1159-------------------------------------------------
1160$ git commit -a
1161-------------------------------------------------
1162
1163will update the index with any files that you've modified or removed
1164and create a commit, all in one step.
1165
1166A number of commands are useful for keeping track of what you're
1167about to commit:
1168
1169-------------------------------------------------
1170$ git diff --cached # difference between HEAD and the index; what
1171 # would be commited if you ran "commit" now.
1172$ git diff # difference between the index file and your
1173 # working directory; changes that would not
1174 # be included if you ran "commit" now.
c64415e2
BF
1175$ git diff HEAD # difference between HEAD and working tree; what
1176 # would be committed if you ran "commit -a" now.
d19fbc3c
BF
1177$ git status # a brief per-file summary of the above.
1178-------------------------------------------------
1179
e34caace 1180[[creating-good-commit-messages]]
ae25c67a 1181Creating good commit messages
d19fbc3c
BF
1182-----------------------------
1183
1184Though not required, it's a good idea to begin the commit message
1185with a single short (less than 50 character) line summarizing the
1186change, followed by a blank line and then a more thorough
1187description. Tools that turn commits into email, for example, use
1188the first line on the Subject line and the rest of the commit in the
1189body.
1190
e34caace 1191[[how-to-merge]]
ae25c67a 1192How to merge
d19fbc3c
BF
1193------------
1194
1195You can rejoin two diverging branches of development using
1196gitlink:git-merge[1]:
1197
1198-------------------------------------------------
1199$ git merge branchname
1200-------------------------------------------------
1201
1202merges the development in the branch "branchname" into the current
1203branch. If there are conflicts--for example, if the same file is
1204modified in two different ways in the remote branch and the local
1205branch--then you are warned; the output may look something like this:
1206
1207-------------------------------------------------
fabbd8f6
BF
1208$ git merge next
1209 100% (4/4) done
1210Auto-merged file.txt
d19fbc3c
BF
1211CONFLICT (content): Merge conflict in file.txt
1212Automatic merge failed; fix conflicts and then commit the result.
1213-------------------------------------------------
1214
1215Conflict markers are left in the problematic files, and after
1216you resolve the conflicts manually, you can update the index
1217with the contents and run git commit, as you normally would when
1218creating a new file.
1219
1220If you examine the resulting commit using gitk, you will see that it
1221has two parents, one pointing to the top of the current branch, and
1222one to the top of the other branch.
1223
d19fbc3c
BF
1224[[resolving-a-merge]]
1225Resolving a merge
1226-----------------
1227
1228When a merge isn't resolved automatically, git leaves the index and
1229the working tree in a special state that gives you all the
1230information you need to help resolve the merge.
1231
1232Files with conflicts are marked specially in the index, so until you
ef561ac7
BF
1233resolve the problem and update the index, gitlink:git-commit[1] will
1234fail:
d19fbc3c
BF
1235
1236-------------------------------------------------
1237$ git commit
1238file.txt: needs merge
1239-------------------------------------------------
1240
ef561ac7
BF
1241Also, gitlink:git-status[1] will list those files as "unmerged", and the
1242files with conflicts will have conflict markers added, like this:
1243
1244-------------------------------------------------
1245<<<<<<< HEAD:file.txt
1246Hello world
1247=======
1248Goodbye
1249>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1250-------------------------------------------------
1251
1252All you need to do is edit the files to resolve the conflicts, and then
1253
1254-------------------------------------------------
1255$ git add file.txt
1256$ git commit
1257-------------------------------------------------
1258
1259Note that the commit message will already be filled in for you with
1260some information about the merge. Normally you can just use this
1261default message unchanged, but you may add additional commentary of
1262your own if desired.
1263
1264The above is all you need to know to resolve a simple merge. But git
1265also provides more information to help resolve conflicts:
1266
e34caace 1267[[conflict-resolution]]
ef561ac7
BF
1268Getting conflict-resolution help during a merge
1269~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d19fbc3c
BF
1270
1271All of the changes that git was able to merge automatically are
1272already added to the index file, so gitlink:git-diff[1] shows only
ef561ac7 1273the conflicts. It uses an unusual syntax:
d19fbc3c
BF
1274
1275-------------------------------------------------
1276$ git diff
1277diff --cc file.txt
1278index 802992c,2b60207..0000000
1279--- a/file.txt
1280+++ b/file.txt
1281@@@ -1,1 -1,1 +1,5 @@@
1282++<<<<<<< HEAD:file.txt
1283 +Hello world
1284++=======
1285+ Goodbye
1286++>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1287-------------------------------------------------
1288
1289Recall that the commit which will be commited after we resolve this
1290conflict will have two parents instead of the usual one: one parent
1291will be HEAD, the tip of the current branch; the other will be the
1292tip of the other branch, which is stored temporarily in MERGE_HEAD.
1293
ef561ac7
BF
1294During the merge, the index holds three versions of each file. Each of
1295these three "file stages" represents a different version of the file:
1296
1297-------------------------------------------------
1298$ git show :1:file.txt # the file in a common ancestor of both branches
1299$ git show :2:file.txt # the version from HEAD, but including any
1300 # nonconflicting changes from MERGE_HEAD
1301$ git show :3:file.txt # the version from MERGE_HEAD, but including any
1302 # nonconflicting changes from HEAD.
1303-------------------------------------------------
1304
1305Since the stage 2 and stage 3 versions have already been updated with
1306nonconflicting changes, the only remaining differences between them are
1307the important ones; thus gitlink:git-diff[1] can use the information in
1308the index to show only those conflicts.
1309
1310The diff above shows the differences between the working-tree version of
1311file.txt and the stage 2 and stage 3 versions. So instead of preceding
1312each line by a single "+" or "-", it now uses two columns: the first
1313column is used for differences between the first parent and the working
1314directory copy, and the second for differences between the second parent
1315and the working directory copy. (See the "COMBINED DIFF FORMAT" section
1316of gitlink:git-diff-files[1] for a details of the format.)
1317
1318After resolving the conflict in the obvious way (but before updating the
1319index), the diff will look like:
d19fbc3c
BF
1320
1321-------------------------------------------------
1322$ git diff
1323diff --cc file.txt
1324index 802992c,2b60207..0000000
1325--- a/file.txt
1326+++ b/file.txt
1327@@@ -1,1 -1,1 +1,1 @@@
1328- Hello world
1329 -Goodbye
1330++Goodbye world
1331-------------------------------------------------
1332
1333This shows that our resolved version deleted "Hello world" from the
1334first parent, deleted "Goodbye" from the second parent, and added
1335"Goodbye world", which was previously absent from both.
1336
ef561ac7
BF
1337Some special diff options allow diffing the working directory against
1338any of these stages:
1339
1340-------------------------------------------------
1341$ git diff -1 file.txt # diff against stage 1
1342$ git diff --base file.txt # same as the above
1343$ git diff -2 file.txt # diff against stage 2
1344$ git diff --ours file.txt # same as the above
1345$ git diff -3 file.txt # diff against stage 3
1346$ git diff --theirs file.txt # same as the above.
1347-------------------------------------------------
1348
1349The gitlink:git-log[1] and gitk[1] commands also provide special help
1350for merges:
d19fbc3c
BF
1351
1352-------------------------------------------------
1353$ git log --merge
ef561ac7 1354$ gitk --merge
d19fbc3c
BF
1355-------------------------------------------------
1356
ef561ac7
BF
1357These will display all commits which exist only on HEAD or on
1358MERGE_HEAD, and which touch an unmerged file.
d19fbc3c 1359
61d72564 1360You may also use gitlink:git-mergetool[1], which lets you merge the
c64415e2
BF
1361unmerged files using external tools such as emacs or kdiff3.
1362
ef561ac7 1363Each time you resolve the conflicts in a file and update the index:
d19fbc3c
BF
1364
1365-------------------------------------------------
1366$ git add file.txt
d19fbc3c
BF
1367-------------------------------------------------
1368
ef561ac7
BF
1369the different stages of that file will be "collapsed", after which
1370git-diff will (by default) no longer show diffs for that file.
d19fbc3c
BF
1371
1372[[undoing-a-merge]]
ae25c67a 1373Undoing a merge
d19fbc3c
BF
1374---------------
1375
1376If you get stuck and decide to just give up and throw the whole mess
1377away, you can always return to the pre-merge state with
1378
1379-------------------------------------------------
1380$ git reset --hard HEAD
1381-------------------------------------------------
1382
1383Or, if you've already commited the merge that you want to throw away,
1384
1385-------------------------------------------------
1c73bb0e 1386$ git reset --hard ORIG_HEAD
d19fbc3c
BF
1387-------------------------------------------------
1388
1389However, this last command can be dangerous in some cases--never
1390throw away a commit you have already committed if that commit may
1391itself have been merged into another branch, as doing so may confuse
1392further merges.
1393
e34caace 1394[[fast-forwards]]
d19fbc3c
BF
1395Fast-forward merges
1396-------------------
1397
1398There is one special case not mentioned above, which is treated
1399differently. Normally, a merge results in a merge commit, with two
1400parents, one pointing at each of the two lines of development that
1401were merged.
1402
59723040
BF
1403However, if the current branch is a descendant of the other--so every
1404commit present in the one is already contained in the other--then git
1405just performs a "fast forward"; the head of the current branch is moved
1406forward to point at the head of the merged-in branch, without any new
1407commits being created.
d19fbc3c 1408
e34caace 1409[[fixing-mistakes]]
b684f830
BF
1410Fixing mistakes
1411---------------
1412
1413If you've messed up the working tree, but haven't yet committed your
1414mistake, you can return the entire working tree to the last committed
1415state with
1416
1417-------------------------------------------------
1418$ git reset --hard HEAD
1419-------------------------------------------------
1420
1421If you make a commit that you later wish you hadn't, there are two
1422fundamentally different ways to fix the problem:
1423
1424 1. You can create a new commit that undoes whatever was done
1425 by the previous commit. This is the correct thing if your
1426 mistake has already been made public.
1427
1428 2. You can go back and modify the old commit. You should
1429 never do this if you have already made the history public;
1430 git does not normally expect the "history" of a project to
1431 change, and cannot correctly perform repeated merges from
1432 a branch that has had its history changed.
1433
e34caace 1434[[reverting-a-commit]]
b684f830
BF
1435Fixing a mistake with a new commit
1436~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1437
1438Creating a new commit that reverts an earlier change is very easy;
1439just pass the gitlink:git-revert[1] command a reference to the bad
1440commit; for example, to revert the most recent commit:
1441
1442-------------------------------------------------
1443$ git revert HEAD
1444-------------------------------------------------
1445
1446This will create a new commit which undoes the change in HEAD. You
1447will be given a chance to edit the commit message for the new commit.
1448
1449You can also revert an earlier change, for example, the next-to-last:
1450
1451-------------------------------------------------
1452$ git revert HEAD^
1453-------------------------------------------------
1454
1455In this case git will attempt to undo the old change while leaving
1456intact any changes made since then. If more recent changes overlap
1457with the changes to be reverted, then you will be asked to fix
1458conflicts manually, just as in the case of <<resolving-a-merge,
1459resolving a merge>>.
1460
365aa199 1461[[fixing-a-mistake-by-editing-history]]
b684f830
BF
1462Fixing a mistake by editing history
1463~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1464
1465If the problematic commit is the most recent commit, and you have not
1466yet made that commit public, then you may just
1467<<undoing-a-merge,destroy it using git-reset>>.
1468
1469Alternatively, you
1470can edit the working directory and update the index to fix your
1471mistake, just as if you were going to <<how-to-make-a-commit,create a
1472new commit>>, then run
1473
1474-------------------------------------------------
1475$ git commit --amend
1476-------------------------------------------------
1477
1478which will replace the old commit by a new commit incorporating your
1479changes, giving you a chance to edit the old commit message first.
1480
1481Again, you should never do this to a commit that may already have
1482been merged into another branch; use gitlink:git-revert[1] instead in
1483that case.
1484
1485It is also possible to edit commits further back in the history, but
1486this is an advanced topic to be left for
1487<<cleaning-up-history,another chapter>>.
1488
e34caace 1489[[checkout-of-path]]
b684f830
BF
1490Checking out an old version of a file
1491~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1492
1493In the process of undoing a previous bad change, you may find it
1494useful to check out an older version of a particular file using
1495gitlink:git-checkout[1]. We've used git checkout before to switch
1496branches, but it has quite different behavior if it is given a path
1497name: the command
1498
1499-------------------------------------------------
1500$ git checkout HEAD^ path/to/file
1501-------------------------------------------------
1502
1503replaces path/to/file by the contents it had in the commit HEAD^, and
1504also updates the index to match. It does not change branches.
1505
1506If you just want to look at an old version of the file, without
1507modifying the working directory, you can do that with
1508gitlink:git-show[1]:
1509
1510-------------------------------------------------
ed4eb0d8 1511$ git show HEAD^:path/to/file
b684f830
BF
1512-------------------------------------------------
1513
1514which will display the given version of the file.
1515
e34caace 1516[[ensuring-good-performance]]
d19fbc3c
BF
1517Ensuring good performance
1518-------------------------
1519
1520On large repositories, git depends on compression to keep the history
1521information from taking up to much space on disk or in memory.
1522
1523This compression is not performed automatically. Therefore you
17217090 1524should occasionally run gitlink:git-gc[1]:
d19fbc3c
BF
1525
1526-------------------------------------------------
1527$ git gc
1528-------------------------------------------------
1529
17217090
BF
1530to recompress the archive. This can be very time-consuming, so
1531you may prefer to run git-gc when you are not doing other work.
d19fbc3c 1532
e34caace
BF
1533
1534[[ensuring-reliability]]
11e016a3
BF
1535Ensuring reliability
1536--------------------
1537
e34caace 1538[[checking-for-corruption]]
11e016a3
BF
1539Checking the repository for corruption
1540~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1541
1191ee18
BF
1542The gitlink:git-fsck[1] command runs a number of self-consistency checks
1543on the repository, and reports on any problems. This may take some
21dcb3b7
BF
1544time. The most common warning by far is about "dangling" objects:
1545
1546-------------------------------------------------
04e50e94 1547$ git fsck
21dcb3b7
BF
1548dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1549dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1550dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1551dangling blob 218761f9d90712d37a9c5e36f406f92202db07eb
1552dangling commit bf093535a34a4d35731aa2bd90fe6b176302f14f
1553dangling commit 8e4bec7f2ddaa268bef999853c25755452100f8e
1554dangling tree d50bb86186bf27b681d25af89d3b5b68382e4085
1555dangling tree b24c2473f1fd3d91352a624795be026d64c8841f
1556...
1557-------------------------------------------------
1558
59723040
BF
1559Dangling objects are not a problem. At worst they may take up a little
1560extra disk space. They can sometimes provide a last-resort method of
1561recovery lost work--see <<dangling-objects>> for details. However, if
1562you want, you may remove them with gitlink:git-prune[1] or the --prune
1191ee18 1563option to gitlink:git-gc[1]:
21dcb3b7
BF
1564
1565-------------------------------------------------
1566$ git gc --prune
1567-------------------------------------------------
1568
1191ee18
BF
1569This may be time-consuming. Unlike most other git operations (including
1570git-gc when run without any options), it is not safe to prune while
1571other git operations are in progress in the same repository.
21dcb3b7 1572
e34caace 1573[[recovering-lost-changes]]
11e016a3
BF
1574Recovering lost changes
1575~~~~~~~~~~~~~~~~~~~~~~~
1576
e34caace 1577[[reflogs]]
559e4d7a
BF
1578Reflogs
1579^^^^^^^
1580
1581Say you modify a branch with gitlink:git-reset[1] --hard, and then
1582realize that the branch was the only reference you had to that point in
1583history.
1584
1585Fortunately, git also keeps a log, called a "reflog", of all the
1586previous values of each branch. So in this case you can still find the
1587old history using, for example,
1588
1589-------------------------------------------------
1590$ git log master@{1}
1591-------------------------------------------------
1592
1593This lists the commits reachable from the previous version of the head.
1594This syntax can be used to with any git command that accepts a commit,
1595not just with git log. Some other examples:
1596
1597-------------------------------------------------
1598$ git show master@{2} # See where the branch pointed 2,
1599$ git show master@{3} # 3, ... changes ago.
1600$ gitk master@{yesterday} # See where it pointed yesterday,
1601$ gitk master@{"1 week ago"} # ... or last week
953f3d6f
BF
1602$ git log --walk-reflogs master # show reflog entries for master
1603-------------------------------------------------
1604
1605A separate reflog is kept for the HEAD, so
1606
1607-------------------------------------------------
1608$ git show HEAD@{"1 week ago"}
559e4d7a
BF
1609-------------------------------------------------
1610
953f3d6f
BF
1611will show what HEAD pointed to one week ago, not what the current branch
1612pointed to one week ago. This allows you to see the history of what
1613you've checked out.
1614
559e4d7a 1615The reflogs are kept by default for 30 days, after which they may be
036be17e 1616pruned. See gitlink:git-reflog[1] and gitlink:git-gc[1] to learn
559e4d7a
BF
1617how to control this pruning, and see the "SPECIFYING REVISIONS"
1618section of gitlink:git-rev-parse[1] for details.
1619
1620Note that the reflog history is very different from normal git history.
1621While normal history is shared by every repository that works on the
1622same project, the reflog history is not shared: it tells you only about
1623how the branches in your local repository have changed over time.
1624
59723040 1625[[dangling-object-recovery]]
559e4d7a
BF
1626Examining dangling objects
1627^^^^^^^^^^^^^^^^^^^^^^^^^^
1628
59723040
BF
1629In some situations the reflog may not be able to save you. For example,
1630suppose you delete a branch, then realize you need the history it
1631contained. The reflog is also deleted; however, if you have not yet
1632pruned the repository, then you may still be able to find the lost
1633commits in the dangling objects that git-fsck reports. See
1634<<dangling-objects>> for the details.
559e4d7a
BF
1635
1636-------------------------------------------------
1637$ git fsck
1638dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1639dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1640dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1641...
1642-------------------------------------------------
1643
aacd404e 1644You can examine
559e4d7a
BF
1645one of those dangling commits with, for example,
1646
1647------------------------------------------------
1648$ gitk 7281251ddd --not --all
1649------------------------------------------------
1650
1651which does what it sounds like: it says that you want to see the commit
1652history that is described by the dangling commit(s), but not the
1653history that is described by all your existing branches and tags. Thus
1654you get exactly the history reachable from that commit that is lost.
1655(And notice that it might not be just one commit: we only report the
1656"tip of the line" as being dangling, but there might be a whole deep
79c96c57 1657and complex commit history that was dropped.)
559e4d7a
BF
1658
1659If you decide you want the history back, you can always create a new
1660reference pointing to it, for example, a new branch:
1661
1662------------------------------------------------
1663$ git branch recovered-branch 7281251ddd
1664------------------------------------------------
1665
59723040
BF
1666Other types of dangling objects (blobs and trees) are also possible, and
1667dangling objects can arise in other situations.
1668
11e016a3 1669
e34caace 1670[[sharing-development]]
d19fbc3c 1671Sharing development with others
b684f830 1672===============================
d19fbc3c
BF
1673
1674[[getting-updates-with-git-pull]]
1675Getting updates with git pull
b684f830 1676-----------------------------
d19fbc3c
BF
1677
1678After you clone a repository and make a few changes of your own, you
1679may wish to check the original repository for updates and merge them
1680into your own work.
1681
1682We have already seen <<Updating-a-repository-with-git-fetch,how to
1683keep remote tracking branches up to date>> with gitlink:git-fetch[1],
1684and how to merge two branches. So you can merge in changes from the
1685original repository's master branch with:
1686
1687-------------------------------------------------
1688$ git fetch
1689$ git merge origin/master
1690-------------------------------------------------
1691
1692However, the gitlink:git-pull[1] command provides a way to do this in
1693one step:
1694
1695-------------------------------------------------
1696$ git pull origin master
1697-------------------------------------------------
1698
1699In fact, "origin" is normally the default repository to pull from,
1700and the default branch is normally the HEAD of the remote repository,
1701so often you can accomplish the above with just
1702
1703-------------------------------------------------
1704$ git pull
1705-------------------------------------------------
1706
c64415e2
BF
1707See the descriptions of the branch.<name>.remote and branch.<name>.merge
1708options in gitlink:git-config[1] to learn how to control these defaults
1709depending on the current branch. Also note that the --track option to
1710gitlink:git-branch[1] and gitlink:git-checkout[1] can be used to
1711automatically set the default remote branch to pull from at the time
1712that a branch is created:
1713
1714-------------------------------------------------
1715$ git checkout --track -b origin/maint maint
1716-------------------------------------------------
d19fbc3c
BF
1717
1718In addition to saving you keystrokes, "git pull" also helps you by
1719producing a default commit message documenting the branch and
1720repository that you pulled from.
1721
1722(But note that no such commit will be created in the case of a
1723<<fast-forwards,fast forward>>; instead, your branch will just be
79c96c57 1724updated to point to the latest commit from the upstream branch.)
d19fbc3c 1725
1191ee18
BF
1726The git-pull command can also be given "." as the "remote" repository,
1727in which case it just merges in a branch from the current repository; so
4c63ff45
BF
1728the commands
1729
1730-------------------------------------------------
1731$ git pull . branch
1732$ git merge branch
1733-------------------------------------------------
1734
1735are roughly equivalent. The former is actually very commonly used.
1736
e34caace 1737[[submitting-patches]]
d19fbc3c 1738Submitting patches to a project
b684f830 1739-------------------------------
d19fbc3c
BF
1740
1741If you just have a few changes, the simplest way to submit them may
1742just be to send them as patches in email:
1743
036be17e 1744First, use gitlink:git-format-patch[1]; for example:
d19fbc3c
BF
1745
1746-------------------------------------------------
eb6ae7f4 1747$ git format-patch origin
d19fbc3c
BF
1748-------------------------------------------------
1749
1750will produce a numbered series of files in the current directory, one
1751for each patch in the current branch but not in origin/HEAD.
1752
1753You can then import these into your mail client and send them by
1754hand. However, if you have a lot to send at once, you may prefer to
1755use the gitlink:git-send-email[1] script to automate the process.
1756Consult the mailing list for your project first to determine how they
1757prefer such patches be handled.
1758
e34caace 1759[[importing-patches]]
d19fbc3c 1760Importing patches to a project
b684f830 1761------------------------------
d19fbc3c
BF
1762
1763Git also provides a tool called gitlink:git-am[1] (am stands for
1764"apply mailbox"), for importing such an emailed series of patches.
1765Just save all of the patch-containing messages, in order, into a
1766single mailbox file, say "patches.mbox", then run
1767
1768-------------------------------------------------
eb6ae7f4 1769$ git am -3 patches.mbox
d19fbc3c
BF
1770-------------------------------------------------
1771
1772Git will apply each patch in order; if any conflicts are found, it
1773will stop, and you can fix the conflicts as described in
01997b4a
BF
1774"<<resolving-a-merge,Resolving a merge>>". (The "-3" option tells
1775git to perform a merge; if you would prefer it just to abort and
1776leave your tree and index untouched, you may omit that option.)
1777
1778Once the index is updated with the results of the conflict
1779resolution, instead of creating a new commit, just run
d19fbc3c
BF
1780
1781-------------------------------------------------
1782$ git am --resolved
1783-------------------------------------------------
1784
1785and git will create the commit for you and continue applying the
1786remaining patches from the mailbox.
1787
1788The final result will be a series of commits, one for each patch in
1789the original mailbox, with authorship and commit log message each
1790taken from the message containing each patch.
1791
1792[[setting-up-a-public-repository]]
1793Setting up a public repository
b684f830 1794------------------------------
d19fbc3c
BF
1795
1796Another way to submit changes to a project is to simply tell the
1797maintainer of that project to pull from your repository, exactly as
1798you did in the section "<<getting-updates-with-git-pull, Getting
1799updates with git pull>>".
1800
1801If you and maintainer both have accounts on the same machine, then
1802then you can just pull changes from each other's repositories
79c96c57
MC
1803directly; note that all of the commands (gitlink:git-clone[1],
1804git-fetch[1], git-pull[1], etc.) that accept a URL as an argument
21f13ee2 1805will also accept a local directory name; so, for example, you can
d19fbc3c
BF
1806use
1807
1808-------------------------------------------------
1809$ git clone /path/to/repository
1810$ git pull /path/to/other/repository
1811-------------------------------------------------
1812
1813If this sort of setup is inconvenient or impossible, another (more
1814common) option is to set up a public repository on a public server.
1815This also allows you to cleanly separate private work in progress
1816from publicly visible work.
1817
1818You will continue to do your day-to-day work in your personal
1819repository, but periodically "push" changes from your personal
1820repository into your public repository, allowing other developers to
1821pull from that repository. So the flow of changes, in a situation
1822where there is one other developer with a public repository, looks
1823like this:
1824
1825 you push
1826 your personal repo ------------------> your public repo
1827 ^ |
1828 | |
1829 | you pull | they pull
1830 | |
1831 | |
1832 | they push V
1833 their public repo <------------------- their repo
1834
1835Now, assume your personal repository is in the directory ~/proj. We
1836first create a new clone of the repository:
1837
1838-------------------------------------------------
52c80037 1839$ git clone --bare ~/proj proj.git
d19fbc3c
BF
1840-------------------------------------------------
1841
52c80037
BF
1842The resulting directory proj.git contains a "bare" git repository--it is
1843just the contents of the ".git" directory, without a checked-out copy of
1844a working directory.
d19fbc3c 1845
c64415e2 1846Next, copy proj.git to the server where you plan to host the
d19fbc3c
BF
1847public repository. You can use scp, rsync, or whatever is most
1848convenient.
1849
1850If somebody else maintains the public server, they may already have
1851set up a git service for you, and you may skip to the section
1852"<<pushing-changes-to-a-public-repository,Pushing changes to a public
1853repository>>", below.
1854
1855Otherwise, the following sections explain how to export your newly
1856created public repository:
1857
1858[[exporting-via-http]]
1859Exporting a git repository via http
b684f830 1860-----------------------------------
d19fbc3c
BF
1861
1862The git protocol gives better performance and reliability, but on a
1863host with a web server set up, http exports may be simpler to set up.
1864
1865All you need to do is place the newly created bare git repository in
1866a directory that is exported by the web server, and make some
1867adjustments to give web clients some extra information they need:
1868
1869-------------------------------------------------
1870$ mv proj.git /home/you/public_html/proj.git
1871$ cd proj.git
c64415e2 1872$ git --bare update-server-info
d19fbc3c
BF
1873$ chmod a+x hooks/post-update
1874-------------------------------------------------
1875
1876(For an explanation of the last two lines, see
1877gitlink:git-update-server-info[1], and the documentation
1878link:hooks.txt[Hooks used by git].)
1879
1880Advertise the url of proj.git. Anybody else should then be able to
1881clone or pull from that url, for example with a commandline like:
1882
1883-------------------------------------------------
1884$ git clone http://yourserver.com/~you/proj.git
1885-------------------------------------------------
1886
1887(See also
1888link:howto/setup-git-server-over-http.txt[setup-git-server-over-http]
1889for a slightly more sophisticated setup using WebDAV which also
1890allows pushing over http.)
1891
1892[[exporting-via-git]]
1893Exporting a git repository via the git protocol
b684f830 1894-----------------------------------------------
d19fbc3c
BF
1895
1896This is the preferred method.
1897
1898For now, we refer you to the gitlink:git-daemon[1] man page for
1899instructions. (See especially the examples section.)
1900
1901[[pushing-changes-to-a-public-repository]]
1902Pushing changes to a public repository
b684f830 1903--------------------------------------
d19fbc3c
BF
1904
1905Note that the two techniques outline above (exporting via
1906<<exporting-via-http,http>> or <<exporting-via-git,git>>) allow other
1907maintainers to fetch your latest changes, but they do not allow write
1908access, which you will need to update the public repository with the
1909latest changes created in your private repository.
1910
1911The simplest way to do this is using gitlink:git-push[1] and ssh; to
1912update the remote branch named "master" with the latest state of your
1913branch named "master", run
1914
1915-------------------------------------------------
1916$ git push ssh://yourserver.com/~you/proj.git master:master
1917-------------------------------------------------
1918
1919or just
1920
1921-------------------------------------------------
1922$ git push ssh://yourserver.com/~you/proj.git master
1923-------------------------------------------------
1924
1925As with git-fetch, git-push will complain if this does not result in
1926a <<fast-forwards,fast forward>>. Normally this is a sign of
1927something wrong. However, if you are sure you know what you're
1928doing, you may force git-push to perform the update anyway by
1929proceeding the branch name by a plus sign:
1930
1931-------------------------------------------------
1932$ git push ssh://yourserver.com/~you/proj.git +master
1933-------------------------------------------------
1934
1935As with git-fetch, you may also set up configuration options to
1936save typing; so, for example, after
1937
1938-------------------------------------------------
c64415e2 1939$ cat >>.git/config <<EOF
d19fbc3c
BF
1940[remote "public-repo"]
1941 url = ssh://yourserver.com/~you/proj.git
1942EOF
1943-------------------------------------------------
1944
1945you should be able to perform the above push with just
1946
1947-------------------------------------------------
1948$ git push public-repo master
1949-------------------------------------------------
1950
1951See the explanations of the remote.<name>.url, branch.<name>.remote,
9d13bda3 1952and remote.<name>.push options in gitlink:git-config[1] for
d19fbc3c
BF
1953details.
1954
e34caace 1955[[setting-up-a-shared-repository]]
d19fbc3c 1956Setting up a shared repository
b684f830 1957------------------------------
d19fbc3c
BF
1958
1959Another way to collaborate is by using a model similar to that
1960commonly used in CVS, where several developers with special rights
1961all push to and pull from a single shared repository. See
1962link:cvs-migration.txt[git for CVS users] for instructions on how to
1963set this up.
1964
e34caace 1965[[setting-up-gitweb]]
b684f830
BF
1966Allow web browsing of a repository
1967----------------------------------
d19fbc3c 1968
a8cd1402
BF
1969The gitweb cgi script provides users an easy way to browse your
1970project's files and history without having to install git; see the file
04483524 1971gitweb/INSTALL in the git source tree for instructions on setting it up.
d19fbc3c 1972
e34caace 1973[[sharing-development-examples]]
b684f830
BF
1974Examples
1975--------
d19fbc3c 1976
b684f830 1977TODO: topic branches, typical roles as in everyday.txt, ?
d19fbc3c 1978
d19fbc3c 1979
d19fbc3c 1980[[cleaning-up-history]]
4c63ff45
BF
1981Rewriting history and maintaining patch series
1982==============================================
1983
1984Normally commits are only added to a project, never taken away or
1985replaced. Git is designed with this assumption, and violating it will
1986cause git's merge machinery (for example) to do the wrong thing.
1987
1988However, there is a situation in which it can be useful to violate this
1989assumption.
1990
e34caace 1991[[patch-series]]
4c63ff45
BF
1992Creating the perfect patch series
1993---------------------------------
1994
1995Suppose you are a contributor to a large project, and you want to add a
1996complicated feature, and to present it to the other developers in a way
1997that makes it easy for them to read your changes, verify that they are
1998correct, and understand why you made each change.
1999
b181d57f 2000If you present all of your changes as a single patch (or commit), they
79c96c57 2001may find that it is too much to digest all at once.
4c63ff45
BF
2002
2003If you present them with the entire history of your work, complete with
2004mistakes, corrections, and dead ends, they may be overwhelmed.
2005
2006So the ideal is usually to produce a series of patches such that:
2007
2008 1. Each patch can be applied in order.
2009
2010 2. Each patch includes a single logical change, together with a
2011 message explaining the change.
2012
2013 3. No patch introduces a regression: after applying any initial
2014 part of the series, the resulting project still compiles and
2015 works, and has no bugs that it didn't have before.
2016
2017 4. The complete series produces the same end result as your own
2018 (probably much messier!) development process did.
2019
b181d57f
BF
2020We will introduce some tools that can help you do this, explain how to
2021use them, and then explain some of the problems that can arise because
2022you are rewriting history.
4c63ff45 2023
e34caace 2024[[using-git-rebase]]
4c63ff45
BF
2025Keeping a patch series up to date using git-rebase
2026--------------------------------------------------
2027
79c96c57
MC
2028Suppose that you create a branch "mywork" on a remote-tracking branch
2029"origin", and create some commits on top of it:
4c63ff45
BF
2030
2031-------------------------------------------------
2032$ git checkout -b mywork origin
2033$ vi file.txt
2034$ git commit
2035$ vi otherfile.txt
2036$ git commit
2037...
2038-------------------------------------------------
2039
2040You have performed no merges into mywork, so it is just a simple linear
2041sequence of patches on top of "origin":
2042
1dc71a91 2043................................................
4c63ff45
BF
2044 o--o--o <-- origin
2045 \
2046 o--o--o <-- mywork
1dc71a91 2047................................................
4c63ff45
BF
2048
2049Some more interesting work has been done in the upstream project, and
2050"origin" has advanced:
2051
1dc71a91 2052................................................
4c63ff45
BF
2053 o--o--O--o--o--o <-- origin
2054 \
2055 a--b--c <-- mywork
1dc71a91 2056................................................
4c63ff45
BF
2057
2058At this point, you could use "pull" to merge your changes back in;
2059the result would create a new merge commit, like this:
2060
1dc71a91 2061................................................
4c63ff45
BF
2062 o--o--O--o--o--o <-- origin
2063 \ \
2064 a--b--c--m <-- mywork
1dc71a91 2065................................................
4c63ff45
BF
2066
2067However, if you prefer to keep the history in mywork a simple series of
2068commits without any merges, you may instead choose to use
2069gitlink:git-rebase[1]:
2070
2071-------------------------------------------------
2072$ git checkout mywork
2073$ git rebase origin
2074-------------------------------------------------
2075
b181d57f
BF
2076This will remove each of your commits from mywork, temporarily saving
2077them as patches (in a directory named ".dotest"), update mywork to
2078point at the latest version of origin, then apply each of the saved
2079patches to the new mywork. The result will look like:
4c63ff45
BF
2080
2081
1dc71a91 2082................................................
4c63ff45
BF
2083 o--o--O--o--o--o <-- origin
2084 \
2085 a'--b'--c' <-- mywork
1dc71a91 2086................................................
4c63ff45 2087
b181d57f
BF
2088In the process, it may discover conflicts. In that case it will stop
2089and allow you to fix the conflicts; after fixing conflicts, use "git
2090add" to update the index with those contents, and then, instead of
2091running git-commit, just run
4c63ff45
BF
2092
2093-------------------------------------------------
2094$ git rebase --continue
2095-------------------------------------------------
2096
2097and git will continue applying the rest of the patches.
2098
2099At any point you may use the --abort option to abort this process and
2100return mywork to the state it had before you started the rebase:
2101
2102-------------------------------------------------
2103$ git rebase --abort
2104-------------------------------------------------
2105
e34caace 2106[[modifying-one-commit]]
365aa199
BF
2107Modifying a single commit
2108-------------------------
2109
2110We saw in <<fixing-a-mistake-by-editing-history>> that you can replace the
2111most recent commit using
2112
2113-------------------------------------------------
2114$ git commit --amend
2115-------------------------------------------------
2116
2117which will replace the old commit by a new commit incorporating your
2118changes, giving you a chance to edit the old commit message first.
2119
2120You can also use a combination of this and gitlink:git-rebase[1] to edit
2121commits further back in your history. First, tag the problematic commit with
2122
2123-------------------------------------------------
2124$ git tag bad mywork~5
2125-------------------------------------------------
2126
2127(Either gitk or git-log may be useful for finding the commit.)
2128
25d9f3fa
BF
2129Then check out that commit, edit it, and rebase the rest of the series
2130on top of it (note that we could check out the commit on a temporary
2131branch, but instead we're using a <<detached-head,detached head>>):
365aa199
BF
2132
2133-------------------------------------------------
25d9f3fa 2134$ git checkout bad
365aa199
BF
2135$ # make changes here and update the index
2136$ git commit --amend
25d9f3fa 2137$ git rebase --onto HEAD bad mywork
365aa199
BF
2138-------------------------------------------------
2139
25d9f3fa
BF
2140When you're done, you'll be left with mywork checked out, with the top
2141patches on mywork reapplied on top of your modified commit. You can
365aa199
BF
2142then clean up with
2143
2144-------------------------------------------------
365aa199
BF
2145$ git tag -d bad
2146-------------------------------------------------
2147
2148Note that the immutable nature of git history means that you haven't really
2149"modified" existing commits; instead, you have replaced the old commits with
2150new commits having new object names.
2151
e34caace 2152[[reordering-patch-series]]
4c63ff45
BF
2153Reordering or selecting from a patch series
2154-------------------------------------------
2155
b181d57f
BF
2156Given one existing commit, the gitlink:git-cherry-pick[1] command
2157allows you to apply the change introduced by that commit and create a
2158new commit that records it. So, for example, if "mywork" points to a
2159series of patches on top of "origin", you might do something like:
2160
2161-------------------------------------------------
2162$ git checkout -b mywork-new origin
2163$ gitk origin..mywork &
2164-------------------------------------------------
2165
2166And browse through the list of patches in the mywork branch using gitk,
2167applying them (possibly in a different order) to mywork-new using
2168cherry-pick, and possibly modifying them as you go using commit
2169--amend.
2170
2171Another technique is to use git-format-patch to create a series of
2172patches, then reset the state to before the patches:
4c63ff45 2173
b181d57f
BF
2174-------------------------------------------------
2175$ git format-patch origin
2176$ git reset --hard origin
2177-------------------------------------------------
4c63ff45 2178
b181d57f
BF
2179Then modify, reorder, or eliminate patches as preferred before applying
2180them again with gitlink:git-am[1].
4c63ff45 2181
e34caace 2182[[patch-series-tools]]
4c63ff45
BF
2183Other tools
2184-----------
2185
b181d57f 2186There are numerous other tools, such as stgit, which exist for the
79c96c57 2187purpose of maintaining a patch series. These are outside of the scope of
b181d57f 2188this manual.
4c63ff45 2189
e34caace 2190[[problems-with-rewriting-history]]
4c63ff45
BF
2191Problems with rewriting history
2192-------------------------------
2193
b181d57f
BF
2194The primary problem with rewriting the history of a branch has to do
2195with merging. Suppose somebody fetches your branch and merges it into
2196their branch, with a result something like this:
2197
1dc71a91 2198................................................
b181d57f
BF
2199 o--o--O--o--o--o <-- origin
2200 \ \
2201 t--t--t--m <-- their branch:
1dc71a91 2202................................................
b181d57f
BF
2203
2204Then suppose you modify the last three commits:
2205
1dc71a91 2206................................................
b181d57f
BF
2207 o--o--o <-- new head of origin
2208 /
2209 o--o--O--o--o--o <-- old head of origin
1dc71a91 2210................................................
b181d57f
BF
2211
2212If we examined all this history together in one repository, it will
2213look like:
2214
1dc71a91 2215................................................
b181d57f
BF
2216 o--o--o <-- new head of origin
2217 /
2218 o--o--O--o--o--o <-- old head of origin
2219 \ \
2220 t--t--t--m <-- their branch:
1dc71a91 2221................................................
b181d57f
BF
2222
2223Git has no way of knowing that the new head is an updated version of
2224the old head; it treats this situation exactly the same as it would if
2225two developers had independently done the work on the old and new heads
2226in parallel. At this point, if someone attempts to merge the new head
2227in to their branch, git will attempt to merge together the two (old and
2228new) lines of development, instead of trying to replace the old by the
2229new. The results are likely to be unexpected.
2230
2231You may still choose to publish branches whose history is rewritten,
2232and it may be useful for others to be able to fetch those branches in
2233order to examine or test them, but they should not attempt to pull such
2234branches into their own work.
2235
2236For true distributed development that supports proper merging,
2237published branches should never be rewritten.
2238
e34caace 2239[[advanced-branch-management]]
b181d57f
BF
2240Advanced branch management
2241==========================
4c63ff45 2242
e34caace 2243[[fetching-individual-branches]]
b181d57f
BF
2244Fetching individual branches
2245----------------------------
2246
2247Instead of using gitlink:git-remote[1], you can also choose just
2248to update one branch at a time, and to store it locally under an
2249arbitrary name:
2250
2251-------------------------------------------------
2252$ git fetch origin todo:my-todo-work
2253-------------------------------------------------
2254
2255The first argument, "origin", just tells git to fetch from the
2256repository you originally cloned from. The second argument tells git
2257to fetch the branch named "todo" from the remote repository, and to
2258store it locally under the name refs/heads/my-todo-work.
2259
2260You can also fetch branches from other repositories; so
2261
2262-------------------------------------------------
2263$ git fetch git://example.com/proj.git master:example-master
2264-------------------------------------------------
2265
2266will create a new branch named "example-master" and store in it the
2267branch named "master" from the repository at the given URL. If you
2268already have a branch named example-master, it will attempt to
59723040
BF
2269<<fast-forwards,fast-forward>> to the commit given by example.com's
2270master branch. In more detail:
b181d57f 2271
59723040
BF
2272[[fetch-fast-forwards]]
2273git fetch and fast-forwards
2274---------------------------
b181d57f
BF
2275
2276In the previous example, when updating an existing branch, "git
2277fetch" checks to make sure that the most recent commit on the remote
2278branch is a descendant of the most recent commit on your copy of the
2279branch before updating your copy of the branch to point at the new
59723040 2280commit. Git calls this process a <<fast-forwards,fast forward>>.
b181d57f
BF
2281
2282A fast forward looks something like this:
2283
1dc71a91 2284................................................
b181d57f
BF
2285 o--o--o--o <-- old head of the branch
2286 \
2287 o--o--o <-- new head of the branch
1dc71a91 2288................................................
b181d57f
BF
2289
2290
2291In some cases it is possible that the new head will *not* actually be
2292a descendant of the old head. For example, the developer may have
2293realized she made a serious mistake, and decided to backtrack,
2294resulting in a situation like:
2295
1dc71a91 2296................................................
b181d57f
BF
2297 o--o--o--o--a--b <-- old head of the branch
2298 \
2299 o--o--o <-- new head of the branch
1dc71a91 2300................................................
b181d57f
BF
2301
2302In this case, "git fetch" will fail, and print out a warning.
2303
2304In that case, you can still force git to update to the new head, as
2305described in the following section. However, note that in the
2306situation above this may mean losing the commits labeled "a" and "b",
2307unless you've already created a reference of your own pointing to
2308them.
2309
e34caace 2310[[forcing-fetch]]
b181d57f
BF
2311Forcing git fetch to do non-fast-forward updates
2312------------------------------------------------
2313
2314If git fetch fails because the new head of a branch is not a
2315descendant of the old head, you may force the update with:
2316
2317-------------------------------------------------
2318$ git fetch git://example.com/proj.git +master:refs/remotes/example/master
2319-------------------------------------------------
2320
c64415e2
BF
2321Note the addition of the "+" sign. Alternatively, you can use the "-f"
2322flag to force updates of all the fetched branches, as in:
2323
2324-------------------------------------------------
2325$ git fetch -f origin
2326-------------------------------------------------
2327
2328Be aware that commits that the old version of example/master pointed at
2329may be lost, as we saw in the previous section.
b181d57f 2330
e34caace 2331[[remote-branch-configuration]]
b181d57f
BF
2332Configuring remote branches
2333---------------------------
2334
2335We saw above that "origin" is just a shortcut to refer to the
79c96c57 2336repository that you originally cloned from. This information is
b181d57f 2337stored in git configuration variables, which you can see using
9d13bda3 2338gitlink:git-config[1]:
b181d57f
BF
2339
2340-------------------------------------------------
9d13bda3 2341$ git config -l
b181d57f
BF
2342core.repositoryformatversion=0
2343core.filemode=true
2344core.logallrefupdates=true
2345remote.origin.url=git://git.kernel.org/pub/scm/git/git.git
2346remote.origin.fetch=+refs/heads/*:refs/remotes/origin/*
2347branch.master.remote=origin
2348branch.master.merge=refs/heads/master
2349-------------------------------------------------
2350
2351If there are other repositories that you also use frequently, you can
2352create similar configuration options to save typing; for example,
2353after
2354
2355-------------------------------------------------
9d13bda3 2356$ git config remote.example.url git://example.com/proj.git
b181d57f
BF
2357-------------------------------------------------
2358
2359then the following two commands will do the same thing:
2360
2361-------------------------------------------------
2362$ git fetch git://example.com/proj.git master:refs/remotes/example/master
2363$ git fetch example master:refs/remotes/example/master
2364-------------------------------------------------
2365
2366Even better, if you add one more option:
2367
2368-------------------------------------------------
9d13bda3 2369$ git config remote.example.fetch master:refs/remotes/example/master
b181d57f
BF
2370-------------------------------------------------
2371
2372then the following commands will all do the same thing:
2373
2374-------------------------------------------------
52c80037
BF
2375$ git fetch git://example.com/proj.git master:refs/remotes/example/master
2376$ git fetch example master:refs/remotes/example/master
b181d57f
BF
2377$ git fetch example
2378-------------------------------------------------
2379
2380You can also add a "+" to force the update each time:
2381
2382-------------------------------------------------
9d13bda3 2383$ git config remote.example.fetch +master:ref/remotes/example/master
b181d57f
BF
2384-------------------------------------------------
2385
2386Don't do this unless you're sure you won't mind "git fetch" possibly
2387throwing away commits on mybranch.
2388
2389Also note that all of the above configuration can be performed by
2390directly editing the file .git/config instead of using
9d13bda3 2391gitlink:git-config[1].
b181d57f 2392
9d13bda3 2393See gitlink:git-config[1] for more details on the configuration
b181d57f 2394options mentioned above.
d19fbc3c 2395
d19fbc3c 2396
35121930 2397[[git-internals]]
d19fbc3c
BF
2398Git internals
2399=============
2400
a536b08b
BF
2401Git depends on two fundamental abstractions: the "object database", and
2402the "current directory cache" aka "index".
b181d57f 2403
e34caace 2404[[the-object-database]]
b181d57f
BF
2405The Object Database
2406-------------------
2407
2408The object database is literally just a content-addressable collection
2409of objects. All objects are named by their content, which is
2410approximated by the SHA1 hash of the object itself. Objects may refer
2411to other objects (by referencing their SHA1 hash), and so you can
2412build up a hierarchy of objects.
2413
c64415e2 2414All objects have a statically determined "type" which is
b181d57f
BF
2415determined at object creation time, and which identifies the format of
2416the object (i.e. how it is used, and how it can refer to other
2417objects). There are currently four different object types: "blob",
a536b08b 2418"tree", "commit", and "tag".
b181d57f 2419
a536b08b
BF
2420A <<def_blob_object,"blob" object>> cannot refer to any other object,
2421and is, as the name implies, a pure storage object containing some
2422user data. It is used to actually store the file data, i.e. a blob
2423object is associated with some particular version of some file.
b181d57f 2424
a536b08b
BF
2425A <<def_tree_object,"tree" object>> is an object that ties one or more
2426"blob" objects into a directory structure. In addition, a tree object
2427can refer to other tree objects, thus creating a directory hierarchy.
b181d57f 2428
a536b08b
BF
2429A <<def_commit_object,"commit" object>> ties such directory hierarchies
2430together into a <<def_DAG,directed acyclic graph>> of revisions - each
2431"commit" is associated with exactly one tree (the directory hierarchy at
2432the time of the commit). In addition, a "commit" refers to one or more
2433"parent" commit objects that describe the history of how we arrived at
2434that directory hierarchy.
b181d57f
BF
2435
2436As a special case, a commit object with no parents is called the "root"
c64415e2 2437commit, and is the point of an initial project commit. Each project
b181d57f
BF
2438must have at least one root, and while you can tie several different
2439root objects together into one project by creating a commit object which
2440has two or more separate roots as its ultimate parents, that's probably
2441just going to confuse people. So aim for the notion of "one root object
2442per project", even if git itself does not enforce that.
2443
a536b08b
BF
2444A <<def_tag_object,"tag" object>> symbolically identifies and can be
2445used to sign other objects. It contains the identifier and type of
2446another object, a symbolic name (of course!) and, optionally, a
2447signature.
b181d57f
BF
2448
2449Regardless of object type, all objects share the following
2450characteristics: they are all deflated with zlib, and have a header
2451that not only specifies their type, but also provides size information
2452about the data in the object. It's worth noting that the SHA1 hash
2453that is used to name the object is the hash of the original data
2454plus this header, so `sha1sum` 'file' does not match the object name
2455for 'file'.
2456(Historical note: in the dawn of the age of git the hash
2457was the sha1 of the 'compressed' object.)
2458
2459As a result, the general consistency of an object can always be tested
2460independently of the contents or the type of the object: all objects can
2461be validated by verifying that (a) their hashes match the content of the
2462file and (b) the object successfully inflates to a stream of bytes that
2463forms a sequence of <ascii type without space> + <space> + <ascii decimal
2464size> + <byte\0> + <binary object data>.
2465
2466The structured objects can further have their structure and
2467connectivity to other objects verified. This is generally done with
04e50e94 2468the `git-fsck` program, which generates a full dependency graph
b181d57f
BF
2469of all objects, and verifies their internal consistency (in addition
2470to just verifying their superficial consistency through the hash).
2471
2472The object types in some more detail:
2473
e34caace 2474[[blob-object]]
b181d57f
BF
2475Blob Object
2476-----------
2477
2478A "blob" object is nothing but a binary blob of data, and doesn't
2479refer to anything else. There is no signature or any other
2480verification of the data, so while the object is consistent (it 'is'
2481indexed by its sha1 hash, so the data itself is certainly correct), it
2482has absolutely no other attributes. No name associations, no
2483permissions. It is purely a blob of data (i.e. normally "file
2484contents").
2485
2486In particular, since the blob is entirely defined by its data, if two
2487files in a directory tree (or in multiple different versions of the
2488repository) have the same contents, they will share the same blob
2489object. The object is totally independent of its location in the
2490directory tree, and renaming a file does not change the object that
2491file is associated with in any way.
2492
2493A blob is typically created when gitlink:git-update-index[1]
2494is run, and its data can be accessed by gitlink:git-cat-file[1].
2495
e34caace 2496[[tree-object]]
b181d57f
BF
2497Tree Object
2498-----------
2499
2500The next hierarchical object type is the "tree" object. A tree object
2501is a list of mode/name/blob data, sorted by name. Alternatively, the
2502mode data may specify a directory mode, in which case instead of
2503naming a blob, that name is associated with another TREE object.
2504
2505Like the "blob" object, a tree object is uniquely determined by the
2506set contents, and so two separate but identical trees will always
2507share the exact same object. This is true at all levels, i.e. it's
2508true for a "leaf" tree (which does not refer to any other trees, only
2509blobs) as well as for a whole subdirectory.
2510
2511For that reason a "tree" object is just a pure data abstraction: it
2512has no history, no signatures, no verification of validity, except
2513that since the contents are again protected by the hash itself, we can
2514trust that the tree is immutable and its contents never change.
2515
2516So you can trust the contents of a tree to be valid, the same way you
2517can trust the contents of a blob, but you don't know where those
2518contents 'came' from.
2519
2520Side note on trees: since a "tree" object is a sorted list of
2521"filename+content", you can create a diff between two trees without
2522actually having to unpack two trees. Just ignore all common parts,
2523and your diff will look right. In other words, you can effectively
2524(and efficiently) tell the difference between any two random trees by
2525O(n) where "n" is the size of the difference, rather than the size of
2526the tree.
2527
2528Side note 2 on trees: since the name of a "blob" depends entirely and
2529exclusively on its contents (i.e. there are no names or permissions
2530involved), you can see trivial renames or permission changes by
2531noticing that the blob stayed the same. However, renames with data
2532changes need a smarter "diff" implementation.
2533
2534A tree is created with gitlink:git-write-tree[1] and
2535its data can be accessed by gitlink:git-ls-tree[1].
2536Two trees can be compared with gitlink:git-diff-tree[1].
2537
e34caace 2538[[commit-object]]
b181d57f
BF
2539Commit Object
2540-------------
2541
2542The "commit" object is an object that introduces the notion of
2543history into the picture. In contrast to the other objects, it
2544doesn't just describe the physical state of a tree, it describes how
2545we got there, and why.
2546
2547A "commit" is defined by the tree-object that it results in, the
2548parent commits (zero, one or more) that led up to that point, and a
2549comment on what happened. Again, a commit is not trusted per se:
2550the contents are well-defined and "safe" due to the cryptographically
2551strong signatures at all levels, but there is no reason to believe
2552that the tree is "good" or that the merge information makes sense.
2553The parents do not have to actually have any relationship with the
2554result, for example.
2555
c64415e2 2556Note on commits: unlike some SCM's, commits do not contain
b181d57f
BF
2557rename information or file mode change information. All of that is
2558implicit in the trees involved (the result tree, and the result trees
2559of the parents), and describing that makes no sense in this idiotic
2560file manager.
2561
2562A commit is created with gitlink:git-commit-tree[1] and
2563its data can be accessed by gitlink:git-cat-file[1].
2564
e34caace 2565[[trust]]
b181d57f
BF
2566Trust
2567-----
2568
2569An aside on the notion of "trust". Trust is really outside the scope
2570of "git", but it's worth noting a few things. First off, since
2571everything is hashed with SHA1, you 'can' trust that an object is
2572intact and has not been messed with by external sources. So the name
2573of an object uniquely identifies a known state - just not a state that
2574you may want to trust.
2575
2576Furthermore, since the SHA1 signature of a commit refers to the
2577SHA1 signatures of the tree it is associated with and the signatures
2578of the parent, a single named commit specifies uniquely a whole set
2579of history, with full contents. You can't later fake any step of the
2580way once you have the name of a commit.
2581
2582So to introduce some real trust in the system, the only thing you need
2583to do is to digitally sign just 'one' special note, which includes the
2584name of a top-level commit. Your digital signature shows others
2585that you trust that commit, and the immutability of the history of
2586commits tells others that they can trust the whole history.
2587
2588In other words, you can easily validate a whole archive by just
2589sending out a single email that tells the people the name (SHA1 hash)
2590of the top commit, and digitally sign that email using something
2591like GPG/PGP.
2592
2593To assist in this, git also provides the tag object...
2594
e34caace 2595[[tag-object]]
b181d57f
BF
2596Tag Object
2597----------
2598
2599Git provides the "tag" object to simplify creating, managing and
2600exchanging symbolic and signed tokens. The "tag" object at its
2601simplest simply symbolically identifies another object by containing
2602the sha1, type and symbolic name.
2603
2604However it can optionally contain additional signature information
2605(which git doesn't care about as long as there's less than 8k of
2606it). This can then be verified externally to git.
2607
2608Note that despite the tag features, "git" itself only handles content
2609integrity; the trust framework (and signature provision and
2610verification) has to come from outside.
2611
2612A tag is created with gitlink:git-mktag[1],
2613its data can be accessed by gitlink:git-cat-file[1],
2614and the signature can be verified by
2615gitlink:git-verify-tag[1].
2616
2617
e34caace 2618[[the-index]]
b181d57f
BF
2619The "index" aka "Current Directory Cache"
2620-----------------------------------------
2621
2622The index is a simple binary file, which contains an efficient
c64415e2 2623representation of the contents of a virtual directory. It
b181d57f
BF
2624does so by a simple array that associates a set of names, dates,
2625permissions and content (aka "blob") objects together. The cache is
2626always kept ordered by name, and names are unique (with a few very
2627specific rules) at any point in time, but the cache has no long-term
2628meaning, and can be partially updated at any time.
2629
2630In particular, the index certainly does not need to be consistent with
2631the current directory contents (in fact, most operations will depend on
2632different ways to make the index 'not' be consistent with the directory
2633hierarchy), but it has three very important attributes:
2634
2635'(a) it can re-generate the full state it caches (not just the
2636directory structure: it contains pointers to the "blob" objects so
2637that it can regenerate the data too)'
2638
2639As a special case, there is a clear and unambiguous one-way mapping
2640from a current directory cache to a "tree object", which can be
2641efficiently created from just the current directory cache without
2642actually looking at any other data. So a directory cache at any one
2643time uniquely specifies one and only one "tree" object (but has
2644additional data to make it easy to match up that tree object with what
2645has happened in the directory)
2646
2647'(b) it has efficient methods for finding inconsistencies between that
2648cached state ("tree object waiting to be instantiated") and the
2649current state.'
2650
2651'(c) it can additionally efficiently represent information about merge
2652conflicts between different tree objects, allowing each pathname to be
2653associated with sufficient information about the trees involved that
2654you can create a three-way merge between them.'
2655
79c96c57 2656Those are the ONLY three things that the directory cache does. It's a
b181d57f
BF
2657cache, and the normal operation is to re-generate it completely from a
2658known tree object, or update/compare it with a live tree that is being
2659developed. If you blow the directory cache away entirely, you generally
2660haven't lost any information as long as you have the name of the tree
2661that it described.
2662
2663At the same time, the index is at the same time also the
2664staging area for creating new trees, and creating a new tree always
2665involves a controlled modification of the index file. In particular,
2666the index file can have the representation of an intermediate tree that
2667has not yet been instantiated. So the index can be thought of as a
2668write-back cache, which can contain dirty information that has not yet
2669been written back to the backing store.
2670
2671
2672
e34caace 2673[[the-workflow]]
b181d57f
BF
2674The Workflow
2675------------
2676
2677Generally, all "git" operations work on the index file. Some operations
2678work *purely* on the index file (showing the current state of the
2679index), but most operations move data to and from the index file. Either
2680from the database or from the working directory. Thus there are four
2681main combinations:
2682
e34caace 2683[[working-directory-to-index]]
b181d57f
BF
2684working directory -> index
2685~~~~~~~~~~~~~~~~~~~~~~~~~~
2686
2687You update the index with information from the working directory with
2688the gitlink:git-update-index[1] command. You
2689generally update the index information by just specifying the filename
2690you want to update, like so:
2691
2692-------------------------------------------------
2693$ git-update-index filename
2694-------------------------------------------------
2695
2696but to avoid common mistakes with filename globbing etc, the command
2697will not normally add totally new entries or remove old entries,
2698i.e. it will normally just update existing cache entries.
2699
2700To tell git that yes, you really do realize that certain files no
2701longer exist, or that new files should be added, you
2702should use the `--remove` and `--add` flags respectively.
2703
2704NOTE! A `--remove` flag does 'not' mean that subsequent filenames will
2705necessarily be removed: if the files still exist in your directory
2706structure, the index will be updated with their new status, not
2707removed. The only thing `--remove` means is that update-cache will be
2708considering a removed file to be a valid thing, and if the file really
2709does not exist any more, it will update the index accordingly.
2710
2711As a special case, you can also do `git-update-index --refresh`, which
2712will refresh the "stat" information of each index to match the current
2713stat information. It will 'not' update the object status itself, and
2714it will only update the fields that are used to quickly test whether
2715an object still matches its old backing store object.
2716
e34caace 2717[[index-to-object-database]]
b181d57f
BF
2718index -> object database
2719~~~~~~~~~~~~~~~~~~~~~~~~
2720
2721You write your current index file to a "tree" object with the program
2722
2723-------------------------------------------------
2724$ git-write-tree
2725-------------------------------------------------
2726
2727that doesn't come with any options - it will just write out the
2728current index into the set of tree objects that describe that state,
2729and it will return the name of the resulting top-level tree. You can
2730use that tree to re-generate the index at any time by going in the
2731other direction:
2732
e34caace 2733[[object-database-to-index]]
b181d57f
BF
2734object database -> index
2735~~~~~~~~~~~~~~~~~~~~~~~~
2736
2737You read a "tree" file from the object database, and use that to
2738populate (and overwrite - don't do this if your index contains any
2739unsaved state that you might want to restore later!) your current
2740index. Normal operation is just
2741
2742-------------------------------------------------
2743$ git-read-tree <sha1 of tree>
2744-------------------------------------------------
2745
2746and your index file will now be equivalent to the tree that you saved
2747earlier. However, that is only your 'index' file: your working
2748directory contents have not been modified.
2749
e34caace 2750[[index-to-working-directory]]
b181d57f
BF
2751index -> working directory
2752~~~~~~~~~~~~~~~~~~~~~~~~~~
2753
2754You update your working directory from the index by "checking out"
2755files. This is not a very common operation, since normally you'd just
2756keep your files updated, and rather than write to your working
2757directory, you'd tell the index files about the changes in your
2758working directory (i.e. `git-update-index`).
2759
2760However, if you decide to jump to a new version, or check out somebody
2761else's version, or just restore a previous tree, you'd populate your
2762index file with read-tree, and then you need to check out the result
2763with
2764
2765-------------------------------------------------
2766$ git-checkout-index filename
2767-------------------------------------------------
2768
2769or, if you want to check out all of the index, use `-a`.
2770
2771NOTE! git-checkout-index normally refuses to overwrite old files, so
2772if you have an old version of the tree already checked out, you will
2773need to use the "-f" flag ('before' the "-a" flag or the filename) to
2774'force' the checkout.
2775
2776
2777Finally, there are a few odds and ends which are not purely moving
2778from one representation to the other:
2779
e34caace 2780[[tying-it-all-together]]
b181d57f
BF
2781Tying it all together
2782~~~~~~~~~~~~~~~~~~~~~
2783
2784To commit a tree you have instantiated with "git-write-tree", you'd
2785create a "commit" object that refers to that tree and the history
2786behind it - most notably the "parent" commits that preceded it in
2787history.
2788
2789Normally a "commit" has one parent: the previous state of the tree
2790before a certain change was made. However, sometimes it can have two
2791or more parent commits, in which case we call it a "merge", due to the
2792fact that such a commit brings together ("merges") two or more
2793previous states represented by other commits.
2794
2795In other words, while a "tree" represents a particular directory state
2796of a working directory, a "commit" represents that state in "time",
2797and explains how we got there.
2798
2799You create a commit object by giving it the tree that describes the
2800state at the time of the commit, and a list of parents:
2801
2802-------------------------------------------------
2803$ git-commit-tree <tree> -p <parent> [-p <parent2> ..]
2804-------------------------------------------------
2805
2806and then giving the reason for the commit on stdin (either through
2807redirection from a pipe or file, or by just typing it at the tty).
2808
2809git-commit-tree will return the name of the object that represents
2810that commit, and you should save it away for later use. Normally,
2811you'd commit a new `HEAD` state, and while git doesn't care where you
2812save the note about that state, in practice we tend to just write the
2813result to the file pointed at by `.git/HEAD`, so that we can always see
2814what the last committed state was.
2815
2816Here is an ASCII art by Jon Loeliger that illustrates how
2817various pieces fit together.
2818
2819------------
2820
2821 commit-tree
2822 commit obj
2823 +----+
2824 | |
2825 | |
2826 V V
2827 +-----------+
2828 | Object DB |
2829 | Backing |
2830 | Store |
2831 +-----------+
2832 ^
2833 write-tree | |
2834 tree obj | |
2835 | | read-tree
2836 | | tree obj
2837 V
2838 +-----------+
2839 | Index |
2840 | "cache" |
2841 +-----------+
2842 update-index ^
2843 blob obj | |
2844 | |
2845 checkout-index -u | | checkout-index
2846 stat | | blob obj
2847 V
2848 +-----------+
2849 | Working |
2850 | Directory |
2851 +-----------+
2852
2853------------
2854
2855
e34caace 2856[[examining-the-data]]
b181d57f
BF
2857Examining the data
2858------------------
2859
2860You can examine the data represented in the object database and the
2861index with various helper tools. For every object, you can use
2862gitlink:git-cat-file[1] to examine details about the
2863object:
2864
2865-------------------------------------------------
2866$ git-cat-file -t <objectname>
2867-------------------------------------------------
2868
2869shows the type of the object, and once you have the type (which is
2870usually implicit in where you find the object), you can use
2871
2872-------------------------------------------------
2873$ git-cat-file blob|tree|commit|tag <objectname>
2874-------------------------------------------------
2875
2876to show its contents. NOTE! Trees have binary content, and as a result
2877there is a special helper for showing that content, called
2878`git-ls-tree`, which turns the binary content into a more easily
2879readable form.
2880
2881It's especially instructive to look at "commit" objects, since those
2882tend to be small and fairly self-explanatory. In particular, if you
2883follow the convention of having the top commit name in `.git/HEAD`,
2884you can do
2885
2886-------------------------------------------------
2887$ git-cat-file commit HEAD
2888-------------------------------------------------
2889
2890to see what the top commit was.
2891
e34caace 2892[[merging-multiple-trees]]
b181d57f 2893Merging multiple trees
d19fbc3c
BF
2894----------------------
2895
b181d57f
BF
2896Git helps you do a three-way merge, which you can expand to n-way by
2897repeating the merge procedure arbitrary times until you finally
2898"commit" the state. The normal situation is that you'd only do one
2899three-way merge (two parents), and commit it, but if you like to, you
2900can do multiple parents in one go.
2901
2902To do a three-way merge, you need the two sets of "commit" objects
2903that you want to merge, use those to find the closest common parent (a
2904third "commit" object), and then use those commit objects to find the
2905state of the directory ("tree" object) at these points.
2906
2907To get the "base" for the merge, you first look up the common parent
2908of two commits with
2909
2910-------------------------------------------------
2911$ git-merge-base <commit1> <commit2>
2912-------------------------------------------------
2913
2914which will return you the commit they are both based on. You should
2915now look up the "tree" objects of those commits, which you can easily
2916do with (for example)
2917
2918-------------------------------------------------
2919$ git-cat-file commit <commitname> | head -1
2920-------------------------------------------------
2921
2922since the tree object information is always the first line in a commit
2923object.
2924
1191ee18 2925Once you know the three trees you are going to merge (the one "original"
c64415e2 2926tree, aka the common tree, and the two "result" trees, aka the branches
1191ee18
BF
2927you want to merge), you do a "merge" read into the index. This will
2928complain if it has to throw away your old index contents, so you should
b181d57f 2929make sure that you've committed those - in fact you would normally
1191ee18
BF
2930always do a merge against your last commit (which should thus match what
2931you have in your current index anyway).
b181d57f
BF
2932
2933To do the merge, do
2934
2935-------------------------------------------------
2936$ git-read-tree -m -u <origtree> <yourtree> <targettree>
2937-------------------------------------------------
2938
2939which will do all trivial merge operations for you directly in the
2940index file, and you can just write the result out with
2941`git-write-tree`.
2942
2943
e34caace 2944[[merging-multiple-trees-2]]
b181d57f
BF
2945Merging multiple trees, continued
2946---------------------------------
2947
2948Sadly, many merges aren't trivial. If there are files that have
2949been added.moved or removed, or if both branches have modified the
2950same file, you will be left with an index tree that contains "merge
2951entries" in it. Such an index tree can 'NOT' be written out to a tree
2952object, and you will have to resolve any such merge clashes using
2953other tools before you can write out the result.
2954
2955You can examine such index state with `git-ls-files --unmerged`
2956command. An example:
2957
2958------------------------------------------------
2959$ git-read-tree -m $orig HEAD $target
2960$ git-ls-files --unmerged
2961100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1 hello.c
2962100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2 hello.c
2963100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello.c
2964------------------------------------------------
2965
2966Each line of the `git-ls-files --unmerged` output begins with
2967the blob mode bits, blob SHA1, 'stage number', and the
2968filename. The 'stage number' is git's way to say which tree it
2969came from: stage 1 corresponds to `$orig` tree, stage 2 `HEAD`
2970tree, and stage3 `$target` tree.
2971
2972Earlier we said that trivial merges are done inside
2973`git-read-tree -m`. For example, if the file did not change
2974from `$orig` to `HEAD` nor `$target`, or if the file changed
2975from `$orig` to `HEAD` and `$orig` to `$target` the same way,
2976obviously the final outcome is what is in `HEAD`. What the
2977above example shows is that file `hello.c` was changed from
2978`$orig` to `HEAD` and `$orig` to `$target` in a different way.
2979You could resolve this by running your favorite 3-way merge
c64415e2
BF
2980program, e.g. `diff3`, `merge`, or git's own merge-file, on
2981the blob objects from these three stages yourself, like this:
b181d57f
BF
2982
2983------------------------------------------------
2984$ git-cat-file blob 263414f... >hello.c~1
2985$ git-cat-file blob 06fa6a2... >hello.c~2
2986$ git-cat-file blob cc44c73... >hello.c~3
c64415e2 2987$ git merge-file hello.c~2 hello.c~1 hello.c~3
b181d57f
BF
2988------------------------------------------------
2989
2990This would leave the merge result in `hello.c~2` file, along
2991with conflict markers if there are conflicts. After verifying
2992the merge result makes sense, you can tell git what the final
2993merge result for this file is by:
2994
2995-------------------------------------------------
2996$ mv -f hello.c~2 hello.c
2997$ git-update-index hello.c
2998-------------------------------------------------
2999
3000When a path is in unmerged state, running `git-update-index` for
3001that path tells git to mark the path resolved.
3002
3003The above is the description of a git merge at the lowest level,
3004to help you understand what conceptually happens under the hood.
3005In practice, nobody, not even git itself, uses three `git-cat-file`
3006for this. There is `git-merge-index` program that extracts the
3007stages to temporary files and calls a "merge" script on it:
3008
3009-------------------------------------------------
3010$ git-merge-index git-merge-one-file hello.c
3011-------------------------------------------------
3012
207dfa07 3013and that is what higher level `git merge -s resolve` is implemented with.
b181d57f 3014
e34caace 3015[[pack-files]]
b181d57f
BF
3016How git stores objects efficiently: pack files
3017----------------------------------------------
3018
3019We've seen how git stores each object in a file named after the
3020object's SHA1 hash.
3021
3022Unfortunately this system becomes inefficient once a project has a
3023lot of objects. Try this on an old project:
3024
3025------------------------------------------------
3026$ git count-objects
30276930 objects, 47620 kilobytes
3028------------------------------------------------
3029
3030The first number is the number of objects which are kept in
3031individual files. The second is the amount of space taken up by
3032those "loose" objects.
3033
3034You can save space and make git faster by moving these loose objects in
3035to a "pack file", which stores a group of objects in an efficient
3036compressed format; the details of how pack files are formatted can be
3037found in link:technical/pack-format.txt[technical/pack-format.txt].
3038
3039To put the loose objects into a pack, just run git repack:
3040
3041------------------------------------------------
3042$ git repack
3043Generating pack...
3044Done counting 6020 objects.
3045Deltifying 6020 objects.
3046 100% (6020/6020) done
3047Writing 6020 objects.
3048 100% (6020/6020) done
3049Total 6020, written 6020 (delta 4070), reused 0 (delta 0)
3050Pack pack-3e54ad29d5b2e05838c75df582c65257b8d08e1c created.
3051------------------------------------------------
3052
3053You can then run
3054
3055------------------------------------------------
3056$ git prune
3057------------------------------------------------
3058
3059to remove any of the "loose" objects that are now contained in the
3060pack. This will also remove any unreferenced objects (which may be
3061created when, for example, you use "git reset" to remove a commit).
3062You can verify that the loose objects are gone by looking at the
3063.git/objects directory or by running
3064
3065------------------------------------------------
3066$ git count-objects
30670 objects, 0 kilobytes
3068------------------------------------------------
3069
3070Although the object files are gone, any commands that refer to those
3071objects will work exactly as they did before.
3072
3073The gitlink:git-gc[1] command performs packing, pruning, and more for
3074you, so is normally the only high-level command you need.
d19fbc3c 3075
59723040 3076[[dangling-objects]]
21dcb3b7 3077Dangling objects
61b41790 3078----------------
21dcb3b7 3079
04e50e94 3080The gitlink:git-fsck[1] command will sometimes complain about dangling
21dcb3b7
BF
3081objects. They are not a problem.
3082
1191ee18
BF
3083The most common cause of dangling objects is that you've rebased a
3084branch, or you have pulled from somebody else who rebased a branch--see
3085<<cleaning-up-history>>. In that case, the old head of the original
59723040
BF
3086branch still exists, as does everything it pointed to. The branch
3087pointer itself just doesn't, since you replaced it with another one.
1191ee18 3088
59723040 3089There are also other situations that cause dangling objects. For
1191ee18
BF
3090example, a "dangling blob" may arise because you did a "git add" of a
3091file, but then, before you actually committed it and made it part of the
3092bigger picture, you changed something else in that file and committed
3093that *updated* thing - the old state that you added originally ends up
3094not being pointed to by any commit or tree, so it's now a dangling blob
3095object.
3096
3097Similarly, when the "recursive" merge strategy runs, and finds that
3098there are criss-cross merges and thus more than one merge base (which is
3099fairly unusual, but it does happen), it will generate one temporary
3100midway tree (or possibly even more, if you had lots of criss-crossing
3101merges and more than two merge bases) as a temporary internal merge
3102base, and again, those are real objects, but the end result will not end
3103up pointing to them, so they end up "dangling" in your repository.
3104
3105Generally, dangling objects aren't anything to worry about. They can
3106even be very useful: if you screw something up, the dangling objects can
3107be how you recover your old tree (say, you did a rebase, and realized
3108that you really didn't want to - you can look at what dangling objects
3109you have, and decide to reset your head to some old dangling state).
21dcb3b7 3110
59723040 3111For commits, you can just use:
21dcb3b7
BF
3112
3113------------------------------------------------
3114$ gitk <dangling-commit-sha-goes-here> --not --all
3115------------------------------------------------
3116
59723040
BF
3117This asks for all the history reachable from the given commit but not
3118from any branch, tag, or other reference. If you decide it's something
3119you want, you can always create a new reference to it, e.g.,
3120
3121------------------------------------------------
3122$ git branch recovered-branch <dangling-commit-sha-goes-here>
3123------------------------------------------------
3124
3125For blobs and trees, you can't do the same, but you can still examine
3126them. You can just do
21dcb3b7
BF
3127
3128------------------------------------------------
3129$ git show <dangling-blob/tree-sha-goes-here>
3130------------------------------------------------
3131
1191ee18
BF
3132to show what the contents of the blob were (or, for a tree, basically
3133what the "ls" for that directory was), and that may give you some idea
3134of what the operation was that left that dangling object.
21dcb3b7 3135
1191ee18
BF
3136Usually, dangling blobs and trees aren't very interesting. They're
3137almost always the result of either being a half-way mergebase (the blob
3138will often even have the conflict markers from a merge in it, if you
3139have had conflicting merges that you fixed up by hand), or simply
3140because you interrupted a "git fetch" with ^C or something like that,
3141leaving _some_ of the new objects in the object database, but just
3142dangling and useless.
21dcb3b7
BF
3143
3144Anyway, once you are sure that you're not interested in any dangling
3145state, you can just prune all unreachable objects:
3146
3147------------------------------------------------
3148$ git prune
3149------------------------------------------------
3150
1191ee18
BF
3151and they'll be gone. But you should only run "git prune" on a quiescent
3152repository - it's kind of like doing a filesystem fsck recovery: you
3153don't want to do that while the filesystem is mounted.
21dcb3b7 3154
04e50e94
BF
3155(The same is true of "git-fsck" itself, btw - but since
3156git-fsck never actually *changes* the repository, it just reports
3157on what it found, git-fsck itself is never "dangerous" to run.
21dcb3b7
BF
3158Running it while somebody is actually changing the repository can cause
3159confusing and scary messages, but it won't actually do anything bad. In
3160contrast, running "git prune" while somebody is actively changing the
3161repository is a *BAD* idea).
3162
126640af
JS
3163[[birdview-on-the-source-code]]
3164A birdview on Git's source code
3165-----------------------------
3166
3167While Git's source code is quite elegant, it is not always easy for
3168new developers to find their way through it. A good idea is to look
3169at the contents of the initial commit:
3170_e83c5163316f89bfbde7d9ab23ca2e25604af290_ (also known as _v0.99~954_).
3171
3172Tip: you can see what files are in there with
3173
3174----------------------------------------------------
3175$ git show e83c5163316f89bfbde7d9ab23ca2e25604af290:
3176----------------------------------------------------
3177
3178and look at those files with something like
3179
3180-----------------------------------------------------------
3181$ git show e83c5163316f89bfbde7d9ab23ca2e25604af290:cache.h
3182-----------------------------------------------------------
3183
3184Be sure to read the README in that revision _after_ you are familiar with
3185the terminology (<<glossary>>), since the terminology has changed a little
3186since then. For example, we call the things "commits" now, which are
3187described in that README as "changesets".
3188
3189Actually a lot of the structure as it is now can be explained by that
3190initial commit.
3191
3192For example, we do not call it "cache" any more, but "index", however, the
3193file is still called `cache.h`. Remark: Not much reason to change it now,
3194especially since there is no good single name for it anyway, because it is
3195basically _the_ header file which is included by _all_ of Git's C sources.
3196
3197If you grasp the ideas in that initial commit (it is really small and you
3198can get into it really fast, and it will help you recognize things in the
3199much larger code base we have now), you should go on skimming `cache.h`,
3200`object.h` and `commit.h` in the current version.
3201
3202In the early days, Git (in the tradition of UNIX) was a bunch of programs
3203which were extremely simple, and which you used in scripts, piping the
3204output of one into another. This turned out to be good for initial
3205development, since it was easier to test new things. However, recently
3206many of these parts have become builtins, and some of the core has been
3207"libified", i.e. put into libgit.a for performance, portability reasons,
3208and to avoid code duplication.
3209
3210By now, you know what the index is (and find the corresponding data
3211structures in `cache.h`), and that there are just a couple of object types
3212(blobs, trees, commits and tags) which inherit their common structure from
3213`struct object`, which is their first member (and thus, you can cast e.g.
3214`(struct object *)commit` to achieve the _same_ as `&commit->object`, i.e.
3215get at the object name and flags).
3216
3217Now is a good point to take a break to let this information sink in.
3218
3219Next step: get familiar with the object naming. Read <<naming-commits>>.
3220There are quite a few ways to name an object (and not only revisions!).
3221All of these are handled in `sha1_name.c`. Just have a quick look at
3222the function `get_sha1()`. A lot of the special handling is done by
3223functions like `get_sha1_basic()` or the likes.
3224
3225This is just to get you into the groove for the most libified part of Git:
3226the revision walker.
3227
3228Basically, the initial version of `git log` was a shell script:
3229
3230----------------------------------------------------------------
3231$ git-rev-list --pretty $(git-rev-parse --default HEAD "$@") | \
3232 LESS=-S ${PAGER:-less}
3233----------------------------------------------------------------
3234
3235What does this mean?
3236
3237`git-rev-list` is the original version of the revision walker, which
3238_always_ printed a list of revisions to stdout. It is still functional,
3239and needs to, since most new Git programs start out as scripts using
3240`git-rev-list`.
3241
3242`git-rev-parse` is not as important any more; it was only used to filter out
3243options that were relevant for the different plumbing commands that were
3244called by the script.
3245
3246Most of what `git-rev-list` did is contained in `revision.c` and
3247`revision.h`. It wraps the options in a struct named `rev_info`, which
3248controls how and what revisions are walked, and more.
3249
3250The original job of `git-rev-parse` is now taken by the function
3251`setup_revisions()`, which parses the revisions and the common command line
3252options for the revision walker. This information is stored in the struct
3253`rev_info` for later consumption. You can do your own command line option
3254parsing after calling `setup_revisions()`. After that, you have to call
3255`prepare_revision_walk()` for initialization, and then you can get the
3256commits one by one with the function `get_revision()`.
3257
3258If you are interested in more details of the revision walking process,
3259just have a look at the first implementation of `cmd_log()`; call
3260`git-show v1.3.0~155^2~4` and scroll down to that function (note that you
3261no longer need to call `setup_pager()` directly).
3262
3263Nowadays, `git log` is a builtin, which means that it is _contained_ in the
3264command `git`. The source side of a builtin is
3265
3266- a function called `cmd_<bla>`, typically defined in `builtin-<bla>.c`,
3267 and declared in `builtin.h`,
3268
3269- an entry in the `commands[]` array in `git.c`, and
3270
3271- an entry in `BUILTIN_OBJECTS` in the `Makefile`.
3272
3273Sometimes, more than one builtin is contained in one source file. For
3274example, `cmd_whatchanged()` and `cmd_log()` both reside in `builtin-log.c`,
3275since they share quite a bit of code. In that case, the commands which are
3276_not_ named like the `.c` file in which they live have to be listed in
3277`BUILT_INS` in the `Makefile`.
3278
3279`git log` looks more complicated in C than it does in the original script,
3280but that allows for a much greater flexibility and performance.
3281
3282Here again it is a good point to take a pause.
3283
3284Lesson three is: study the code. Really, it is the best way to learn about
3285the organization of Git (after you know the basic concepts).
3286
3287So, think about something which you are interested in, say, "how can I
3288access a blob just knowing the object name of it?". The first step is to
3289find a Git command with which you can do it. In this example, it is either
3290`git show` or `git cat-file`.
3291
3292For the sake of clarity, let's stay with `git cat-file`, because it
3293
3294- is plumbing, and
3295
3296- was around even in the initial commit (it literally went only through
3297 some 20 revisions as `cat-file.c`, was renamed to `builtin-cat-file.c`
3298 when made a builtin, and then saw less than 10 versions).
3299
3300So, look into `builtin-cat-file.c`, search for `cmd_cat_file()` and look what
3301it does.
3302
3303------------------------------------------------------------------
3304 git_config(git_default_config);
3305 if (argc != 3)
3306 usage("git-cat-file [-t|-s|-e|-p|<type>] <sha1>");
3307 if (get_sha1(argv[2], sha1))
3308 die("Not a valid object name %s", argv[2]);
3309------------------------------------------------------------------
3310
3311Let's skip over the obvious details; the only really interesting part
3312here is the call to `get_sha1()`. It tries to interpret `argv[2]` as an
3313object name, and if it refers to an object which is present in the current
3314repository, it writes the resulting SHA-1 into the variable `sha1`.
3315
3316Two things are interesting here:
3317
3318- `get_sha1()` returns 0 on _success_. This might surprise some new
3319 Git hackers, but there is a long tradition in UNIX to return different
3320 negative numbers in case of different errors -- and 0 on success.
3321
3322- the variable `sha1` in the function signature of `get_sha1()` is `unsigned
3323 char *`, but is actually expected to be a pointer to `unsigned
3324 char[20]`. This variable will contain the 160-bit SHA-1 of the given
3325 commit. Note that whenever a SHA-1 is passed as "unsigned char *", it
3326 is the binary representation, as opposed to the ASCII representation in
3327 hex characters, which is passed as "char *".
3328
3329You will see both of these things throughout the code.
3330
3331Now, for the meat:
3332
3333-----------------------------------------------------------------------------
3334 case 0:
3335 buf = read_object_with_reference(sha1, argv[1], &size, NULL);
3336-----------------------------------------------------------------------------
3337
3338This is how you read a blob (actually, not only a blob, but any type of
3339object). To know how the function `read_object_with_reference()` actually
3340works, find the source code for it (something like `git grep
3341read_object_with | grep ":[a-z]"` in the git repository), and read
3342the source.
3343
3344To find out how the result can be used, just read on in `cmd_cat_file()`:
3345
3346-----------------------------------
3347 write_or_die(1, buf, size);
3348-----------------------------------
3349
3350Sometimes, you do not know where to look for a feature. In many such cases,
3351it helps to search through the output of `git log`, and then `git show` the
3352corresponding commit.
3353
3354Example: If you know that there was some test case for `git bundle`, but
3355do not remember where it was (yes, you _could_ `git grep bundle t/`, but that
3356does not illustrate the point!):
3357
3358------------------------
3359$ git log --no-merges t/
3360------------------------
3361
3362In the pager (`less`), just search for "bundle", go a few lines back,
3363and see that it is in commit 18449ab0... Now just copy this object name,
3364and paste it into the command line
3365
3366-------------------
3367$ git show 18449ab0
3368-------------------
3369
3370Voila.
3371
3372Another example: Find out what to do in order to make some script a
3373builtin:
3374
3375-------------------------------------------------
3376$ git log --no-merges --diff-filter=A builtin-*.c
3377-------------------------------------------------
3378
3379You see, Git is actually the best tool to find out about the source of Git
3380itself!
3381
e34caace 3382[[glossary]]
d19fbc3c
BF
3383include::glossary.txt[]
3384
e34caace 3385[[todo]]
6bd9b682
BF
3386Notes and todo list for this manual
3387===================================
3388
3389This is a work in progress.
3390
3391The basic requirements:
2f99710c
BF
3392 - It must be readable in order, from beginning to end, by
3393 someone intelligent with a basic grasp of the unix
3394 commandline, but without any special knowledge of git. If
3395 necessary, any other prerequisites should be specifically
3396 mentioned as they arise.
3397 - Whenever possible, section headings should clearly describe
3398 the task they explain how to do, in language that requires
3399 no more knowledge than necessary: for example, "importing
3400 patches into a project" rather than "the git-am command"
6bd9b682 3401
d5cd5de4
BF
3402Think about how to create a clear chapter dependency graph that will
3403allow people to get to important topics without necessarily reading
3404everything in between.
d19fbc3c 3405
aacd404e
MC
3406Say something about .gitignore.
3407
d19fbc3c
BF
3408Scan Documentation/ for other stuff left out; in particular:
3409 howto's
d19fbc3c
BF
3410 some of technical/?
3411 hooks
0b375ab0 3412 list of commands in gitlink:git[1]
d19fbc3c
BF
3413
3414Scan email archives for other stuff left out
3415
3416Scan man pages to see if any assume more background than this manual
3417provides.
3418
2f99710c 3419Simplify beginning by suggesting disconnected head instead of
b181d57f 3420temporary branch creation?
d19fbc3c 3421
2f99710c
BF
3422Add more good examples. Entire sections of just cookbook examples
3423might be a good idea; maybe make an "advanced examples" section a
3424standard end-of-chapter section?
d19fbc3c
BF
3425
3426Include cross-references to the glossary, where appropriate.
3427
9a241220
BF
3428Document shallow clones? See draft 1.5.0 release notes for some
3429documentation.
3430
3dff5379 3431Add a section on working with other version control systems, including
9a241220
BF
3432CVS, Subversion, and just imports of series of release tarballs.
3433
a8cd1402 3434More details on gitweb?
0b375ab0
BF
3435
3436Write a chapter on using plumbing and writing scripts.