1GITCORE-TUTORIAL(7) Git Manual GITCORE-TUTORIAL(7)
2
6 gitcore-tutorial - A git core tutorial for developers
7
9 git *
10
12 This tutorial explains how to use the "core" git commands to set up and
13 work with a git repository.
14 If you just need to use git as a revision control system you may prefer
16 to start with "A Tutorial Introduction to GIT" (gittutorial(7)) or the
17 GIT User Manual[1].
18 However, an understanding of these low-level tools can be helpful if
20 you want to understand git’s internals.
21 The core git is often called "plumbing", with the prettier user
23 interfaces on top of it called "porcelain". You may not want to use the
24 plumbing directly very often, but it can be good to know what the
25 plumbing does for when the porcelain isn’t flushing.
26 Back when this document was originally written, many porcelain commands
28 were shell scripts. For simplicity, it still uses them as examples to
29 illustrate how plumbing is fit together to form the porcelain commands.
30 The source tree includes some of these scripts in contrib/examples/ for
31 reference. Although these are not implemented as shell scripts anymore,
32 the description of what the plumbing layer commands do is still valid.
33 Note
35 Deeper technical details are often marked as Notes, which you can
36 skip on your first reading.
37
39 Creating a new git repository couldn’t be easier: all git repositories
40 start out empty, and the only thing you need to do is find yourself a
41 subdirectory that you want to use as a working tree - either an empty
42 one for a totally new project, or an existing working tree that you
43 want to import into git.
44 For our first example, we’re going to start a totally new repository
46 from scratch, with no pre-existing files, and we’ll call it
47 git-tutorial. To start up, create a subdirectory for it, change into
48 that subdirectory, and initialize the git infrastructure with git init:
49 $ mkdir git-tutorial
51 $ cd git-tutorial
52 $ git init
53 to which git will reply
56 Initialized empty Git repository in .git/
58 which is just git’s way of saying that you haven’t been doing anything
61 strange, and that it will have created a local .git directory setup for
62 your new project. You will now have a .git directory, and you can
63 inspect that with ls. For your new empty project, it should show you
64 three entries, among other things:
65 · a file called HEAD, that has ref: refs/heads/master in it. This is
67 similar to a symbolic link and points at refs/heads/master relative
68 to the HEAD file.
69 Don’t worry about the fact that the file that the HEAD link points
71 to doesn’t even exist yet — you haven’t created the commit that
72 will start your HEAD development branch yet.
73 · a subdirectory called objects, which will contain all the objects
75 of your project. You should never have any real reason to look at
76 the objects directly, but you might want to know that these objects
77 are what contains all the real data in your repository.
78 · a subdirectory called refs, which contains references to objects.
80 In particular, the refs subdirectory will contain two other
82 subdirectories, named heads and tags respectively. They do exactly what
83 their names imply: they contain references to any number of different
84 heads of development (aka branches), and to any tags that you have
85 created to name specific versions in your repository.
86 One note: the special master head is the default branch, which is why
88 the .git/HEAD file was created points to it even if it doesn’t yet
89 exist. Basically, the HEAD link is supposed to always point to the
90 branch you are working on right now, and you always start out expecting
91 to work on the master branch.
92 However, this is only a convention, and you can name your branches
94 anything you want, and don’t have to ever even have a master branch. A
95 number of the git tools will assume that .git/HEAD is valid, though.
96 Note
98 An object is identified by its 160-bit SHA1 hash, aka object name,
99 and a reference to an object is always the 40-byte hex
100 representation of that SHA1 name. The files in the refs
101 subdirectory are expected to contain these hex references (usually
102 with a final \n at the end), and you should thus expect to see a
103 number of 41-byte files containing these references in these refs
104 subdirectories when you actually start populating your tree.
105 Note
107 An advanced user may want to take a look at gitrepository-layout(5)
108 after finishing this tutorial.
109 You have now created your first git repository. Of course, since it’s
111 empty, that’s not very useful, so let’s start populating it with data.
112
114 We’ll keep this simple and stupid, so we’ll start off with populating a
115 few trivial files just to get a feel for it.
116 Start off with just creating any random files that you want to maintain
118 in your git repository. We’ll start off with a few bad examples, just
119 to get a feel for how this works:
120 $ echo "Hello World" >hello
122 $ echo "Silly example" >example
123 you have now created two files in your working tree (aka working
126 directory), but to actually check in your hard work, you will have to
127 go through two steps:
128 · fill in the index file (aka cache) with the information about your
130 working tree state.
131 · commit that index file as an object.
133 The first step is trivial: when you want to tell git about any changes
135 to your working tree, you use the git update-index program. That
136 program normally just takes a list of filenames you want to update, but
137 to avoid trivial mistakes, it refuses to add new entries to the index
138 (or remove existing ones) unless you explicitly tell it that you’re
139 adding a new entry with the --add flag (or removing an entry with the
140 --remove) flag.
141 So to populate the index with the two files you just created, you can
143 do
144 $ git update-index --add hello example
146 and you have now told git to track those two files.
149 In fact, as you did that, if you now look into your object directory,
151 you’ll notice that git will have added two new objects to the object
152 database. If you did exactly the steps above, you should now be able to
153 do
154 $ ls .git/objects/??/*
156 and see two files:
159 .git/objects/55/7db03de997c86a4a028e1ebd3a1ceb225be238
161 .git/objects/f2/4c74a2e500f5ee1332c86b94199f52b1d1d962
162 which correspond with the objects with names of 557db... and f24c7...
165 respectively.
166 If you want to, you can use git cat-file to look at those objects, but
168 you’ll have to use the object name, not the filename of the object:
169 $ git cat-file -t 557db03de997c86a4a028e1ebd3a1ceb225be238
171 where the -t tells git cat-file to tell you what the "type" of the
174 object is. git will tell you that you have a "blob" object (i.e., just
175 a regular file), and you can see the contents with
176 $ git cat-file blob 557db03
178 which will print out "Hello World". The object 557db03 is nothing more
181 than the contents of your file hello.
182 Note
184 Don’t confuse that object with the file hello itself. The object is
185 literally just those specific contents of the file, and however
186 much you later change the contents in file hello, the object we
187 just looked at will never change. Objects are immutable.
188 Note
190 The second example demonstrates that you can abbreviate the object
191 name to only the first several hexadecimal digits in most places.
192 Anyway, as we mentioned previously, you normally never actually take a
194 look at the objects themselves, and typing long 40-character hex names
195 is not something you’d normally want to do. The above digression was
196 just to show that git update-index did something magical, and actually
197 saved away the contents of your files into the git object database.
198 Updating the index did something else too: it created a .git/index
200 file. This is the index that describes your current working tree, and
201 something you should be very aware of. Again, you normally never worry
202 about the index file itself, but you should be aware of the fact that
203 you have not actually really "checked in" your files into git so far,
204 you’ve only told git about them.
205 However, since git knows about them, you can now start using some of
207 the most basic git commands to manipulate the files or look at their
208 status.
209 In particular, let’s not even check in the two files into git yet,
211 we’ll start off by adding another line to hello first:
212 $ echo "It's a new day for git" >>hello
214 and you can now, since you told git about the previous state of hello,
217 ask git what has changed in the tree compared to your old index, using
218 the git diff-files command:
219 $ git diff-files
221 Oops. That wasn’t very readable. It just spit out its own internal
224 version of a diff, but that internal version really just tells you that
225 it has noticed that "hello" has been modified, and that the old object
226 contents it had have been replaced with something else.
227 To make it readable, we can tell git diff-files to output the
229 differences as a patch, using the -p flag:
230 $ git diff-files -p
232 diff --git a/hello b/hello
233 index 557db03..263414f 100644
234 --- a/hello
235 +++ b/hello
236 @@ -1 +1,2 @@
237 Hello World
238 +It's a new day for git
239 i.e. the diff of the change we caused by adding another line to hello.
242 In other words, git diff-files always shows us the difference between
244 what is recorded in the index, and what is currently in the working
245 tree. That’s very useful.
246 A common shorthand for git diff-files -p is to just write git diff,
248 which will do the same thing.
249 $ git diff
251 diff --git a/hello b/hello
252 index 557db03..263414f 100644
253 --- a/hello
254 +++ b/hello
255 @@ -1 +1,2 @@
256 Hello World
257 +It's a new day for git
258
261 Now, we want to go to the next stage in git, which is to take the files
262 that git knows about in the index, and commit them as a real tree. We
263 do that in two phases: creating a tree object, and committing that tree
264 object as a commit object together with an explanation of what the tree
265 was all about, along with information of how we came to that state.
266 Creating a tree object is trivial, and is done with git write-tree.
268 There are no options or other input: git write-tree will take the
269 current index state, and write an object that describes that whole
270 index. In other words, we’re now tying together all the different
271 filenames with their contents (and their permissions), and we’re
272 creating the equivalent of a git "directory" object:
273 $ git write-tree
275 and this will just output the name of the resulting tree, in this case
278 (if you have done exactly as I’ve described) it should be
279 8988da15d077d4829fc51d8544c097def6644dbb
281 which is another incomprehensible object name. Again, if you want to,
284 you can use git cat-file -t 8988d... to see that this time the object
285 is not a "blob" object, but a "tree" object (you can also use git
286 cat-file to actually output the raw object contents, but you’ll see
287 mainly a binary mess, so that’s less interesting).
288 However — normally you’d never use git write-tree on its own, because
290 normally you always commit a tree into a commit object using the git
291 commit-tree command. In fact, it’s easier to not actually use git
292 write-tree on its own at all, but to just pass its result in as an
293 argument to git commit-tree.
294 git commit-tree normally takes several arguments — it wants to know
296 what the parent of a commit was, but since this is the first commit
297 ever in this new repository, and it has no parents, we only need to
298 pass in the object name of the tree. However, git commit-tree also
299 wants to get a commit message on its standard input, and it will write
300 out the resulting object name for the commit to its standard output.
301 And this is where we create the .git/refs/heads/master file which is
303 pointed at by HEAD. This file is supposed to contain the reference to
304 the top-of-tree of the master branch, and since that’s exactly what git
305 commit-tree spits out, we can do this all with a sequence of simple
306 shell commands:
307 $ tree=$(git write-tree)
309 $ commit=$(echo 'Initial commit' | git commit-tree $tree)
310 $ git update-ref HEAD $commit
311 In this case this creates a totally new commit that is not related to
314 anything else. Normally you do this only once for a project ever, and
315 all later commits will be parented on top of an earlier commit.
316 Again, normally you’d never actually do this by hand. There is a
318 helpful script called git commit that will do all of this for you. So
319 you could have just written git commit instead, and it would have done
320 the above magic scripting for you.
321
323 Remember how we did the git update-index on file hello and then we
324 changed hello afterward, and could compare the new state of hello with
325 the state we saved in the index file?
326 Further, remember how I said that git write-tree writes the contents of
328 the index file to the tree, and thus what we just committed was in fact
329 the original contents of the file hello, not the new ones. We did that
330 on purpose, to show the difference between the index state, and the
331 state in the working tree, and how they don’t have to match, even when
332 we commit things.
333 As before, if we do git diff-files -p in our git-tutorial project,
335 we’ll still see the same difference we saw last time: the index file
336 hasn’t changed by the act of committing anything. However, now that we
337 have committed something, we can also learn to use a new command: git
338 diff-index.
339 Unlike git diff-files, which showed the difference between the index
341 file and the working tree, git diff-index shows the differences between
342 a committed tree and either the index file or the working tree. In
343 other words, git diff-index wants a tree to be diffed against, and
344 before we did the commit, we couldn’t do that, because we didn’t have
345 anything to diff against.
346 But now we can do
348 $ git diff-index -p HEAD
350 (where -p has the same meaning as it did in git diff-files), and it
353 will show us the same difference, but for a totally different reason.
354 Now we’re comparing the working tree not against the index file, but
355 against the tree we just wrote. It just so happens that those two are
356 obviously the same, so we get the same result.
357 Again, because this is a common operation, you can also just shorthand
359 it with
360 $ git diff HEAD
362 which ends up doing the above for you.
365 In other words, git diff-index normally compares a tree against the
367 working tree, but when given the --cached flag, it is told to instead
368 compare against just the index cache contents, and ignore the current
369 working tree state entirely. Since we just wrote the index file to
370 HEAD, doing git diff-index --cached -p HEAD should thus return an empty
371 set of differences, and that’s exactly what it does.
372 Note
374 git diff-index really always uses the index for its comparisons,
375 and saying that it compares a tree against the working tree is thus
376 not strictly accurate. In particular, the list of files to compare
377 (the "meta-data") always comes from the index file, regardless of
378 whether the --cached flag is used or not. The --cached flag really
379 only determines whether the file contents to be compared come from
380 the working tree or not.
381 This is not hard to understand, as soon as you realize that git
383 simply never knows (or cares) about files that it is not told about
384 explicitly. git will never go looking for files to compare, it
385 expects you to tell it what the files are, and that’s what the
386 index is there for.
387 However, our next step is to commit the change we did, and again, to
389 understand what’s going on, keep in mind the difference between
390 "working tree contents", "index file" and "committed tree". We have
391 changes in the working tree that we want to commit, and we always have
392 to work through the index file, so the first thing we need to do is to
393 update the index cache:
394 $ git update-index hello
396 (note how we didn’t need the --add flag this time, since git knew about
399 the file already).
400 Note what happens to the different git diff-* versions here. After
402 we’ve updated hello in the index, git diff-files -p now shows no
403 differences, but git diff-index -p HEAD still does show that the
404 current state is different from the state we committed. In fact, now
405 git diff-index shows the same difference whether we use the --cached
406 flag or not, since now the index is coherent with the working tree.
407 Now, since we’ve updated hello in the index, we can commit the new
409 version. We could do it by writing the tree by hand again, and
410 committing the tree (this time we’d have to use the -p HEAD flag to
411 tell commit that the HEAD was the parent of the new commit, and that
412 this wasn’t an initial commit any more), but you’ve done that once
413 already, so let’s just use the helpful script this time:
414 $ git commit
416 which starts an editor for you to write the commit message and tells
419 you a bit about what you have done.
420 Write whatever message you want, and all the lines that start with #
422 will be pruned out, and the rest will be used as the commit message for
423 the change. If you decide you don’t want to commit anything after all
424 at this point (you can continue to edit things and update the index),
425 you can just leave an empty message. Otherwise git commit will commit
426 the change for you.
427 You’ve now made your first real git commit. And if you’re interested in
429 looking at what git commit really does, feel free to investigate: it’s
430 a few very simple shell scripts to generate the helpful (?) commit
431 message headers, and a few one-liners that actually do the commit
432 itself (git commit).
433
435 While creating changes is useful, it’s even more useful if you can tell
436 later what changed. The most useful command for this is another of the
437 diff family, namely git diff-tree.
438 git diff-tree can be given two arbitrary trees, and it will tell you
440 the differences between them. Perhaps even more commonly, though, you
441 can give it just a single commit object, and it will figure out the
442 parent of that commit itself, and show the difference directly. Thus,
443 to get the same diff that we’ve already seen several times, we can now
444 do
445 $ git diff-tree -p HEAD
447 (again, -p means to show the difference as a human-readable patch), and
450 it will show what the last commit (in HEAD) actually changed.
451 Note
453 Here is an ASCII art by Jon Loeliger that illustrates how various
454 diff-* commands compare things.
455 diff-tree
457 +----+
458 | |
459 | |
460 V V
461 +-----------+
462 | Object DB |
463 | Backing |
464 | Store |
465 +-----------+
466 ^ ^
467 | |
468 | | diff-index --cached
469 | |
470 diff-index | V
471 | +-----------+
472 | | Index |
473 | | "cache" |
474 | +-----------+
475 | ^
476 | |
477 | | diff-files
478 | |
479 V V
480 +-----------+
481 | Working |
482 | Directory |
483 +-----------+
484 More interestingly, you can also give git diff-tree the --pretty flag,
486 which tells it to also show the commit message and author and date of
487 the commit, and you can tell it to show a whole series of diffs.
488 Alternatively, you can tell it to be "silent", and not show the diffs
489 at all, but just show the actual commit message.
490 In fact, together with the git rev-list program (which generates a list
492 of revisions), git diff-tree ends up being a veritable fount of
493 changes. A trivial (but very useful) script called git whatchanged is
494 included with git which does exactly this, and shows a log of recent
495 activities.
496 To see the whole history of our pitiful little git-tutorial project,
498 you can do
499 $ git log
501 which shows just the log messages, or if we want to see the log
504 together with the associated patches use the more complex (and much
505 more powerful)
506 $ git whatchanged -p
508 and you will see exactly what has changed in the repository over its
511 short history.
512 Note
514 When using the above two commands, the initial commit will be
515 shown. If this is a problem because it is huge, you can hide it by
516 setting the log.showroot configuration variable to false. Having
517 this, you can still show it for each command just adding the --root
518 option, which is a flag for git diff-tree accepted by both
519 commands.
520 With that, you should now be having some inkling of what git does, and
522 can explore on your own.
523 Note
525 Most likely, you are not directly using the core git Plumbing
526 commands, but using Porcelain such as git add, ‘git-rm’ and
527 ‘git-commit’.
528
530 In git, there are two kinds of tags, a "light" one, and an "annotated
531 tag".
532 A "light" tag is technically nothing more than a branch, except we put
534 it in the .git/refs/tags/ subdirectory instead of calling it a head. So
535 the simplest form of tag involves nothing more than
536 $ git tag my-first-tag
538 which just writes the current HEAD into the .git/refs/tags/my-first-tag
541 file, after which point you can then use this symbolic name for that
542 particular state. You can, for example, do
543 $ git diff my-first-tag
545 to diff your current state against that tag which at this point will
548 obviously be an empty diff, but if you continue to develop and commit
549 stuff, you can use your tag as an "anchor-point" to see what has
550 changed since you tagged it.
551 An "annotated tag" is actually a real git object, and contains not only
553 a pointer to the state you want to tag, but also a small tag name and
554 message, along with optionally a PGP signature that says that yes, you
555 really did that tag. You create these annotated tags with either the -a
556 or -s flag to git tag:
557 $ git tag -s <tagname>
559 which will sign the current HEAD (but you can also give it another
562 argument that specifies the thing to tag, e.g., you could have tagged
563 the current mybranch point by using git tag <tagname> mybranch).
564 You normally only do signed tags for major releases or things like
566 that, while the light-weight tags are useful for any marking you want
567 to do — any time you decide that you want to remember a certain point,
568 just create a private tag for it, and you have a nice symbolic name for
569 the state at that point.
570
572 git repositories are normally totally self-sufficient and relocatable.
573 Unlike CVS, for example, there is no separate notion of "repository"
574 and "working tree". A git repository normally is the working tree, with
575 the local git information hidden in the .git subdirectory. There is
576 nothing else. What you see is what you got.
577 Note
579 You can tell git to split the git internal information from the
580 directory that it tracks, but we’ll ignore that for now: it’s not
581 how normal projects work, and it’s really only meant for special
582 uses. So the mental model of "the git information is always tied
583 directly to the working tree that it describes" may not be
584 technically 100% accurate, but it’s a good model for all normal
585 use.
586 This has two implications:
588 · if you grow bored with the tutorial repository you created (or
590 you’ve made a mistake and want to start all over), you can just do
591 simple
592 $ rm -rf git-tutorial
594 and it will be gone. There’s no external repository, and there’s no
596 history outside the project you created.
597 · if you want to move or duplicate a git repository, you can do so.
599 There is git clone command, but if all you want to do is just to
600 create a copy of your repository (with all the full history that
601 went along with it), you can do so with a regular cp -a
602 git-tutorial new-git-tutorial.
603 Note that when you’ve moved or copied a git repository, your git
605 index file (which caches various information, notably some of the
606 "stat" information for the files involved) will likely need to be
607 refreshed. So after you do a cp -a to create a new copy, you’ll
608 want to do
609 $ git update-index --refresh
611 in the new repository to make sure that the index file is
613 up-to-date.
614 Note that the second point is true even across machines. You can
616 duplicate a remote git repository with any regular copy mechanism, be
617 it scp, rsync or wget.
618 When copying a remote repository, you’ll want to at a minimum update
620 the index cache when you do this, and especially with other peoples'
621 repositories you often want to make sure that the index cache is in
622 some known state (you don’t know what they’ve done and not yet checked
623 in), so usually you’ll precede the git update-index with a
624 $ git read-tree --reset HEAD
626 $ git update-index --refresh
627 which will force a total index re-build from the tree pointed to by
630 HEAD. It resets the index contents to HEAD, and then the git
631 update-index makes sure to match up all index entries with the
632 checked-out files. If the original repository had uncommitted changes
633 in its working tree, git update-index --refresh notices them and tells
634 you they need to be updated.
635 The above can also be written as simply
637 $ git reset
639 and in fact a lot of the common git command combinations can be
642 scripted with the git xyz interfaces. You can learn things by just
643 looking at what the various git scripts do. For example, git reset used
644 to be the above two lines implemented in git reset, but some things
645 like git status and git commit are slightly more complex scripts around
646 the basic git commands.
647 Many (most?) public remote repositories will not contain any of the
649 checked out files or even an index file, and will only contain the
650 actual core git files. Such a repository usually doesn’t even have the
651 .git subdirectory, but has all the git files directly in the
652 repository.
653 To create your own local live copy of such a "raw" git repository,
655 you’d first create your own subdirectory for the project, and then copy
656 the raw repository contents into the .git directory. For example, to
657 create your own copy of the git repository, you’d do the following
658 $ mkdir my-git
660 $ cd my-git
661 $ rsync -rL rsync://rsync.kernel.org/pub/scm/git/git.git/ .git
662 followed by
665 $ git read-tree HEAD
667 to populate the index. However, now you have populated the index, and
670 you have all the git internal files, but you will notice that you don’t
671 actually have any of the working tree files to work on. To get those,
672 you’d check them out with
673 $ git checkout-index -u -a
675 where the -u flag means that you want the checkout to keep the index
678 up-to-date (so that you don’t have to refresh it afterward), and the -a
679 flag means "check out all files" (if you have a stale copy or an older
680 version of a checked out tree you may also need to add the -f flag
681 first, to tell git checkout-index to force overwriting of any old
682 files).
683 Again, this can all be simplified with
685 $ git clone rsync://rsync.kernel.org/pub/scm/git/git.git/ my-git
687 $ cd my-git
688 $ git checkout
689 which will end up doing all of the above for you.
692 You have now successfully copied somebody else’s (mine) remote
694 repository, and checked it out.
695
697 Branches in git are really nothing more than pointers into the git
698 object database from within the .git/refs/ subdirectory, and as we
699 already discussed, the HEAD branch is nothing but a symlink to one of
700 these object pointers.
701 You can at any time create a new branch by just picking an arbitrary
703 point in the project history, and just writing the SHA1 name of that
704 object into a file under .git/refs/heads/. You can use any filename you
705 want (and indeed, subdirectories), but the convention is that the
706 "normal" branch is called master. That’s just a convention, though, and
707 nothing enforces it.
708 To show that as an example, let’s go back to the git-tutorial
710 repository we used earlier, and create a branch in it. You do that by
711 simply just saying that you want to check out a new branch:
712 $ git checkout -b mybranch
714 will create a new branch based at the current HEAD position, and switch
717 to it.
718 Note
720 If you make the decision to start your new branch at some other
721 point in the history than the current HEAD, you can do so by just
722 telling git checkout what the base of the checkout would be. In
723 other words, if you have an earlier tag or branch, you’d just do
724 $ git checkout -b mybranch earlier-commit
726 and it would create the new branch mybranch at the earlier commit,
729 and check out the state at that time.
730 You can always just jump back to your original master branch by doing
732 $ git checkout master
734 (or any other branch-name, for that matter) and if you forget which
737 branch you happen to be on, a simple
738 $ cat .git/HEAD
740 will tell you where it’s pointing. To get the list of branches you
743 have, you can say
744 $ git branch
746 which used to be nothing more than a simple script around ls
749 .git/refs/heads. There will be an asterisk in front of the branch you
750 are currently on.
751 Sometimes you may wish to create a new branch without actually checking
753 it out and switching to it. If so, just use the command
754 $ git branch <branchname> [startingpoint]
756 which will simply create the branch, but will not do anything further.
759 You can then later — once you decide that you want to actually develop
760 on that branch — switch to that branch with a regular git checkout with
761 the branchname as the argument.
762
764 One of the ideas of having a branch is that you do some (possibly
765 experimental) work in it, and eventually merge it back to the main
766 branch. So assuming you created the above mybranch that started out
767 being the same as the original master branch, let’s make sure we’re in
768 that branch, and do some work there.
769 $ git checkout mybranch
771 $ echo "Work, work, work" >>hello
772 $ git commit -m "Some work." -i hello
773 Here, we just added another line to hello, and we used a shorthand for
776 doing both git update-index hello and git commit by just giving the
777 filename directly to git commit, with an -i flag (it tells git to
778 include that file in addition to what you have done to the index file
779 so far when making the commit). The -m flag is to give the commit log
780 message from the command line.
781 Now, to make it a bit more interesting, let’s assume that somebody else
783 does some work in the original branch, and simulate that by going back
784 to the master branch, and editing the same file differently there:
785 $ git checkout master
787 Here, take a moment to look at the contents of hello, and notice how
790 they don’t contain the work we just did in mybranch — because that work
791 hasn’t happened in the master branch at all. Then do
792 $ echo "Play, play, play" >>hello
794 $ echo "Lots of fun" >>example
795 $ git commit -m "Some fun." -i hello example
796 since the master branch is obviously in a much better mood.
799 Now, you’ve got two branches, and you decide that you want to merge the
801 work done. Before we do that, let’s introduce a cool graphical tool
802 that helps you view what’s going on:
803 $ gitk --all
805 will show you graphically both of your branches (that’s what the --all
808 means: normally it will just show you your current HEAD) and their
809 histories. You can also see exactly how they came to be from a common
810 source.
811 Anyway, let’s exit gitk (^Q or the File menu), and decide that we want
813 to merge the work we did on the mybranch branch into the master branch
814 (which is currently our HEAD too). To do that, there’s a nice script
815 called git merge, which wants to know which branches you want to
816 resolve and what the merge is all about:
817 $ git merge -m "Merge work in mybranch" mybranch
819 where the first argument is going to be used as the commit message if
822 the merge can be resolved automatically.
823 Now, in this case we’ve intentionally created a situation where the
825 merge will need to be fixed up by hand, though, so git will do as much
826 of it as it can automatically (which in this case is just merge the
827 example file, which had no differences in the mybranch branch), and
828 say:
829 Auto-merging hello
831 CONFLICT (content): Merge conflict in hello
832 Automatic merge failed; fix conflicts and then commit the result.
833 It tells you that it did an "Automatic merge", which failed due to
836 conflicts in hello.
837 Not to worry. It left the (trivial) conflict in hello in the same form
839 you should already be well used to if you’ve ever used CVS, so let’s
840 just open hello in our editor (whatever that may be), and fix it up
841 somehow. I’d suggest just making it so that hello contains all four
842 lines:
843 Hello World
845 It's a new day for git
846 Play, play, play
847 Work, work, work
848 and once you’re happy with your manual merge, just do a
851 $ git commit -i hello
853 which will very loudly warn you that you’re now committing a merge
856 (which is correct, so never mind), and you can write a small merge
857 message about your adventures in git merge-land.
858 After you’re done, start up gitk --all to see graphically what the
860 history looks like. Notice that mybranch still exists, and you can
861 switch to it, and continue to work with it if you want to. The mybranch
862 branch will not contain the merge, but next time you merge it from the
863 master branch, git will know how you merged it, so you’ll not have to
864 do that merge again.
865 Another useful tool, especially if you do not always work in X-Window
867 environment, is git show-branch.
868 $ git show-branch --topo-order --more=1 master mybranch
870 * [master] Merge work in mybranch
871 ! [mybranch] Some work.
872 --
873 - [master] Merge work in mybranch
874 *+ [mybranch] Some work.
875 * [master^] Some fun.
876 The first two lines indicate that it is showing the two branches and
879 the first line of the commit log message from their top-of-the-tree
880 commits, you are currently on master branch (notice the asterisk *
881 character), and the first column for the later output lines is used to
882 show commits contained in the master branch, and the second column for
883 the mybranch branch. Three commits are shown along with their log
884 messages. All of them have non blank characters in the first column (*
885 shows an ordinary commit on the current branch, - is a merge commit),
886 which means they are now part of the master branch. Only the "Some
887 work" commit has the plus + character in the second column, because
888 mybranch has not been merged to incorporate these commits from the
889 master branch. The string inside brackets before the commit log message
890 is a short name you can use to name the commit. In the above example,
891 master and mybranch are branch heads. master^ is the first parent of
892 master branch head. Please see gitrevisions(7) if you want to see more
893 complex cases.
894 Note
896 Without the --more=1 option, git show-branch would not output the
897 [master^] commit, as [mybranch] commit is a common ancestor of both
898 master and mybranch tips. Please see git-show-branch(1) for
899 details.
900 Note
902 If there were more commits on the master branch after the merge,
903 the merge commit itself would not be shown by git show-branch by
904 default. You would need to provide --sparse option to make the
905 merge commit visible in this case.
906 Now, let’s pretend you are the one who did all the work in mybranch,
908 and the fruit of your hard work has finally been merged to the master
909 branch. Let’s go back to mybranch, and run git merge to get the
910 "upstream changes" back to your branch.
911 $ git checkout mybranch
913 $ git merge -m "Merge upstream changes." master
914 This outputs something like this (the actual commit object names would
917 be different)
918 Updating from ae3a2da... to a80b4aa....
920 Fast-forward (no commit created; -m option ignored)
921 example | 1 +
922 hello | 1 +
923 2 files changed, 2 insertions(+), 0 deletions(-)
924 Because your branch did not contain anything more than what had already
927 been merged into the master branch, the merge operation did not
928 actually do a merge. Instead, it just updated the top of the tree of
929 your branch to that of the master branch. This is often called
930 fast-forward merge.
931 You can run gitk --all again to see how the commit ancestry looks like,
933 or run show-branch, which tells you this.
934 $ git show-branch master mybranch
936 ! [master] Merge work in mybranch
937 * [mybranch] Merge work in mybranch
938 --
939 -- [master] Merge work in mybranch
940
943 It’s usually much more common that you merge with somebody else than
944 merging with your own branches, so it’s worth pointing out that git
945 makes that very easy too, and in fact, it’s not that different from
946 doing a git merge. In fact, a remote merge ends up being nothing more
947 than "fetch the work from a remote repository into a temporary tag"
948 followed by a git merge.
949 Fetching from a remote repository is done by, unsurprisingly, git
951 fetch:
952 $ git fetch <remote-repository>
954 One of the following transports can be used to name the repository to
957 download from:
958 Rsync
960 rsync://remote.machine/path/to/repo.git/
962 Rsync transport is usable for both uploading and downloading, but
964 is completely unaware of what git does, and can produce unexpected
965 results when you download from the public repository while the
966 repository owner is uploading into it via rsync transport. Most
967 notably, it could update the files under refs/ which holds the
968 object name of the topmost commits before uploading the files in
969 objects/ — the downloader would obtain head commit object name
970 while that object itself is still not available in the repository.
971 For this reason, it is considered deprecated.
972 SSH
974 remote.machine:/path/to/repo.git/ or
976 ssh://remote.machine/path/to/repo.git/
978 This transport can be used for both uploading and downloading, and
980 requires you to have a log-in privilege over ssh to the remote
981 machine. It finds out the set of objects the other side lacks by
982 exchanging the head commits both ends have and transfers (close to)
983 minimum set of objects. It is by far the most efficient way to
984 exchange git objects between repositories.
985 Local directory
987 /path/to/repo.git/
989 This transport is the same as SSH transport but uses sh to run both
991 ends on the local machine instead of running other end on the
992 remote machine via ssh.
993 git Native
995 git://remote.machine/path/to/repo.git/
997 This transport was designed for anonymous downloading. Like SSH
999 transport, it finds out the set of objects the downstream side
1000 lacks and transfers (close to) minimum set of objects.
1001 HTTP(S)
1003 http://remote.machine/path/to/repo.git/
1005 Downloader from http and https URL first obtains the topmost commit
1007 object name from the remote site by looking at the specified
1008 refname under repo.git/refs/ directory, and then tries to obtain
1009 the commit object by downloading from repo.git/objects/xx/xxx...
1010 using the object name of that commit object. Then it reads the
1011 commit object to find out its parent commits and the associate tree
1012 object; it repeats this process until it gets all the necessary
1013 objects. Because of this behavior, they are sometimes also called
1014 commit walkers.
1015 The commit walkers are sometimes also called dumb transports,
1017 because they do not require any git aware smart server like git
1018 Native transport does. Any stock HTTP server that does not even
1019 support directory index would suffice. But you must prepare your
1020 repository with git update-server-info to help dumb transport
1021 downloaders.
1022 Once you fetch from the remote repository, you merge that with your
1024 current branch.
1025 However — it’s such a common thing to fetch and then immediately merge,
1027 that it’s called git pull, and you can simply do
1028 $ git pull <remote-repository>
1030 and optionally give a branch-name for the remote end as a second
1033 argument.
1034 Note
1036 You could do without using any branches at all, by keeping as many
1037 local repositories as you would like to have branches, and merging
1038 between them with git pull, just like you merge between branches.
1039 The advantage of this approach is that it lets you keep a set of
1040 files for each branch checked out and you may find it easier to
1041 switch back and forth if you juggle multiple lines of development
1042 simultaneously. Of course, you will pay the price of more disk
1043 usage to hold multiple working trees, but disk space is cheap these
1044 days.
1045 It is likely that you will be pulling from the same remote repository
1047 from time to time. As a short hand, you can store the remote repository
1048 URL in the local repository’s config file like this:
1049 $ git config remote.linus.url http://www.kernel.org/pub/scm/git/git.git/
1051 and use the "linus" keyword with git pull instead of the full URL.
1054 Examples.
1056 1. git pull linus
1058 2. git pull linus tag v0.99.1
1060 the above are equivalent to:
1062 1. git pull http://www.kernel.org/pub/scm/git/git.git/ HEAD
1064 2. git pull http://www.kernel.org/pub/scm/git/git.git/ tag v0.99.1
1066
1068 We said this tutorial shows what plumbing does to help you cope with
1069 the porcelain that isn’t flushing, but we so far did not talk about how
1070 the merge really works. If you are following this tutorial the first
1071 time, I’d suggest to skip to "Publishing your work" section and come
1072 back here later.
1073 OK, still with me? To give us an example to look at, let’s go back to
1075 the earlier repository with "hello" and "example" file, and bring
1076 ourselves back to the pre-merge state:
1077 $ git show-branch --more=2 master mybranch
1079 ! [master] Merge work in mybranch
1080 * [mybranch] Merge work in mybranch
1081 --
1082 -- [master] Merge work in mybranch
1083 +* [master^2] Some work.
1084 +* [master^] Some fun.
1085 Remember, before running git merge, our master head was at "Some fun."
1088 commit, while our mybranch head was at "Some work." commit.
1089 $ git checkout mybranch
1091 $ git reset --hard master^2
1092 $ git checkout master
1093 $ git reset --hard master^
1094 After rewinding, the commit structure should look like this:
1097 $ git show-branch
1099 * [master] Some fun.
1100 ! [mybranch] Some work.
1101 --
1102 * [master] Some fun.
1103 + [mybranch] Some work.
1104 *+ [master^] Initial commit
1105 Now we are ready to experiment with the merge by hand.
1108 git merge command, when merging two branches, uses 3-way merge
1110 algorithm. First, it finds the common ancestor between them. The
1111 command it uses is git merge-base:
1112 $ mb=$(git merge-base HEAD mybranch)
1114 The command writes the commit object name of the common ancestor to the
1117 standard output, so we captured its output to a variable, because we
1118 will be using it in the next step. By the way, the common ancestor
1119 commit is the "Initial commit" commit in this case. You can tell it by:
1120 $ git name-rev --name-only --tags $mb
1122 my-first-tag
1123 After finding out a common ancestor commit, the second step is this:
1126 $ git read-tree -m -u $mb HEAD mybranch
1128 This is the same git read-tree command we have already seen, but it
1131 takes three trees, unlike previous examples. This reads the contents of
1132 each tree into different stage in the index file (the first tree goes
1133 to stage 1, the second to stage 2, etc.). After reading three trees
1134 into three stages, the paths that are the same in all three stages are
1135 collapsed into stage 0. Also paths that are the same in two of three
1136 stages are collapsed into stage 0, taking the SHA1 from either stage 2
1137 or stage 3, whichever is different from stage 1 (i.e. only one side
1138 changed from the common ancestor).
1139 After collapsing operation, paths that are different in three trees are
1141 left in non-zero stages. At this point, you can inspect the index file
1142 with this command:
1143 $ git ls-files --stage
1145 100644 7f8b141b65fdcee47321e399a2598a235a032422 0 example
1146 100644 557db03de997c86a4a028e1ebd3a1ceb225be238 1 hello
1147 100644 ba42a2a96e3027f3333e13ede4ccf4498c3ae942 2 hello
1148 100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello
1149 In our example of only two files, we did not have unchanged files so
1152 only example resulted in collapsing. But in real-life large projects,
1153 when only a small number of files change in one commit, this collapsing
1154 tends to trivially merge most of the paths fairly quickly, leaving only
1155 a handful of real changes in non-zero stages.
1156 To look at only non-zero stages, use --unmerged flag:
1158 $ git ls-files --unmerged
1160 100644 557db03de997c86a4a028e1ebd3a1ceb225be238 1 hello
1161 100644 ba42a2a96e3027f3333e13ede4ccf4498c3ae942 2 hello
1162 100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello
1163 The next step of merging is to merge these three versions of the file,
1166 using 3-way merge. This is done by giving git merge-one-file command as
1167 one of the arguments to git merge-index command:
1168 $ git merge-index git-merge-one-file hello
1170 Auto-merging hello
1171 ERROR: Merge conflict in hello
1172 fatal: merge program failed
1173 git merge-one-file script is called with parameters to describe those
1176 three versions, and is responsible to leave the merge results in the
1177 working tree. It is a fairly straightforward shell script, and
1178 eventually calls merge program from RCS suite to perform a file-level
1179 3-way merge. In this case, merge detects conflicts, and the merge
1180 result with conflict marks is left in the working tree.. This can be
1181 seen if you run ls-files --stage again at this point:
1182 $ git ls-files --stage
1184 100644 7f8b141b65fdcee47321e399a2598a235a032422 0 example
1185 100644 557db03de997c86a4a028e1ebd3a1ceb225be238 1 hello
1186 100644 ba42a2a96e3027f3333e13ede4ccf4498c3ae942 2 hello
1187 100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello
1188 This is the state of the index file and the working file after git
1191 merge returns control back to you, leaving the conflicting merge for
1192 you to resolve. Notice that the path hello is still unmerged, and what
1193 you see with git diff at this point is differences since stage 2 (i.e.
1194 your version).
1195
1197 So, we can use somebody else’s work from a remote repository, but how
1198 can you prepare a repository to let other people pull from it?
1199 You do your real work in your working tree that has your primary
1201 repository hanging under it as its .git subdirectory. You could make
1202 that repository accessible remotely and ask people to pull from it, but
1203 in practice that is not the way things are usually done. A recommended
1204 way is to have a public repository, make it reachable by other people,
1205 and when the changes you made in your primary working tree are in good
1206 shape, update the public repository from it. This is often called
1207 pushing.
1208 Note
1210 This public repository could further be mirrored, and that is how
1211 git repositories at kernel.org are managed.
1212 Publishing the changes from your local (private) repository to your
1214 remote (public) repository requires a write privilege on the remote
1215 machine. You need to have an SSH account there to run a single command,
1216 git-receive-pack.
1217 First, you need to create an empty repository on the remote machine
1219 that will house your public repository. This empty repository will be
1220 populated and be kept up-to-date by pushing into it later. Obviously,
1221 this repository creation needs to be done only once.
1222 Note
1224 git push uses a pair of commands, git send-pack on your local
1225 machine, and git-receive-pack on the remote machine. The
1226 communication between the two over the network internally uses an
1227 SSH connection.
1228 Your private repository’s git directory is usually .git, but your
1230 public repository is often named after the project name, i.e.
1231 <project>.git. Let’s create such a public repository for project
1232 my-git. After logging into the remote machine, create an empty
1233 directory:
1234 $ mkdir my-git.git
1236 Then, make that directory into a git repository by running git init,
1239 but this time, since its name is not the usual .git, we do things
1240 slightly differently:
1241 $ GIT_DIR=my-git.git git init
1243 Make sure this directory is available for others you want your changes
1246 to be pulled via the transport of your choice. Also you need to make
1247 sure that you have the git-receive-pack program on the $PATH.
1248 Note
1250 Many installations of sshd do not invoke your shell as the login
1251 shell when you directly run programs; what this means is that if
1252 your login shell is bash, only .bashrc is read and not
1253 .bash_profile. As a workaround, make sure .bashrc sets up $PATH so
1254 that you can run git-receive-pack program.
1255 Note
1257 If you plan to publish this repository to be accessed over http,
1258 you should do mv my-git.git/hooks/post-update.sample
1259 my-git.git/hooks/post-update at this point. This makes sure that
1260 every time you push into this repository, git update-server-info is
1261 run.
1262 Your "public repository" is now ready to accept your changes. Come back
1264 to the machine you have your private repository. From there, run this
1265 command:
1266 $ git push <public-host>:/path/to/my-git.git master
1268 This synchronizes your public repository to match the named branch head
1271 (i.e. master in this case) and objects reachable from them in your
1272 current repository.
1273 As a real example, this is how I update my public git repository.
1275 Kernel.org mirror network takes care of the propagation to other
1276 publicly visible machines:
1277 $ git push master.kernel.org:/pub/scm/git/git.git/
1279
1282 Earlier, we saw that one file under .git/objects/??/ directory is
1283 stored for each git object you create. This representation is efficient
1284 to create atomically and safely, but not so convenient to transport
1285 over the network. Since git objects are immutable once they are
1286 created, there is a way to optimize the storage by "packing them
1287 together". The command
1288 $ git repack
1290 will do it for you. If you followed the tutorial examples, you would
1293 have accumulated about 17 objects in .git/objects/??/ directories by
1294 now. git repack tells you how many objects it packed, and stores the
1295 packed file in .git/objects/pack directory.
1296 Note
1298 You will see two files, pack-*.pack and pack-*.idx, in
1299 .git/objects/pack directory. They are closely related to each
1300 other, and if you ever copy them by hand to a different repository
1301 for whatever reason, you should make sure you copy them together.
1302 The former holds all the data from the objects in the pack, and the
1303 latter holds the index for random access.
1304 If you are paranoid, running git verify-pack command would detect if
1306 you have a corrupt pack, but do not worry too much. Our programs are
1307 always perfect ;-).
1308 Once you have packed objects, you do not need to leave the unpacked
1310 objects that are contained in the pack file anymore.
1311 $ git prune-packed
1313 would remove them for you.
1316 You can try running find .git/objects -type f before and after you run
1318 git prune-packed if you are curious. Also git count-objects would tell
1319 you how many unpacked objects are in your repository and how much space
1320 they are consuming.
1321 Note
1323 git pull is slightly cumbersome for HTTP transport, as a packed
1324 repository may contain relatively few objects in a relatively large
1325 pack. If you expect many HTTP pulls from your public repository you
1326 might want to repack & prune often, or never.
1327 If you run git repack again at this point, it will say "Nothing new to
1329 pack.". Once you continue your development and accumulate the changes,
1330 running git repack again will create a new pack, that contains objects
1331 created since you packed your repository the last time. We recommend
1332 that you pack your project soon after the initial import (unless you
1333 are starting your project from scratch), and then run git repack every
1334 once in a while, depending on how active your project is.
1335 When a repository is synchronized via git push and git pull objects
1337 packed in the source repository are usually stored unpacked in the
1338 destination, unless rsync transport is used. While this allows you to
1339 use different packing strategies on both ends, it also means you may
1340 need to repack both repositories every once in a while.
1341
1343 Although git is a truly distributed system, it is often convenient to
1344 organize your project with an informal hierarchy of developers. Linux
1345 kernel development is run this way. There is a nice illustration (page
1346 17, "Merges to Mainline") in Randy Dunlap’s presentation[2].
1347 It should be stressed that this hierarchy is purely informal. There is
1349 nothing fundamental in git that enforces the "chain of patch flow" this
1350 hierarchy implies. You do not have to pull from only one remote
1351 repository.
1352 A recommended workflow for a "project lead" goes like this:
1354 1. Prepare your primary repository on your local machine. Your work is
1356 done there.
1357 2. Prepare a public repository accessible to others.
1359 If other people are pulling from your repository over dumb
1361 transport protocols (HTTP), you need to keep this repository dumb
1362 transport friendly. After git init,
1363 $GIT_DIR/hooks/post-update.sample copied from the standard
1364 templates would contain a call to git update-server-info but you
1365 need to manually enable the hook with mv post-update.sample
1366 post-update. This makes sure git update-server-info keeps the
1367 necessary files up-to-date.
1368 3. Push into the public repository from your primary repository.
1370 4. git repack the public repository. This establishes a big pack that
1372 contains the initial set of objects as the baseline, and possibly
1373 git prune if the transport used for pulling from your repository
1374 supports packed repositories.
1375 5. Keep working in your primary repository. Your changes include
1377 modifications of your own, patches you receive via e-mails, and
1378 merges resulting from pulling the "public" repositories of your
1379 "subsystem maintainers".
1380 You can repack this private repository whenever you feel like.
1382 6. Push your changes to the public repository, and announce it to the
1384 public.
1385 7. Every once in a while, git repack the public repository. Go back to
1387 step 5. and continue working.
1388 A recommended work cycle for a "subsystem maintainer" who works on that
1390 project and has an own "public repository" goes like this:
1391 1. Prepare your work repository, by git clone the public repository of
1393 the "project lead". The URL used for the initial cloning is stored
1394 in the remote.origin.url configuration variable.
1395 2. Prepare a public repository accessible to others, just like the
1397 "project lead" person does.
1398 3. Copy over the packed files from "project lead" public repository to
1400 your public repository, unless the "project lead" repository lives
1401 on the same machine as yours. In the latter case, you can use
1402 objects/info/alternates file to point at the repository you are
1403 borrowing from.
1404 4. Push into the public repository from your primary repository. Run
1406 git repack, and possibly git prune if the transport used for
1407 pulling from your repository supports packed repositories.
1408 5. Keep working in your primary repository. Your changes include
1410 modifications of your own, patches you receive via e-mails, and
1411 merges resulting from pulling the "public" repositories of your
1412 "project lead" and possibly your "sub-subsystem maintainers".
1413 You can repack this private repository whenever you feel like.
1415 6. Push your changes to your public repository, and ask your "project
1417 lead" and possibly your "sub-subsystem maintainers" to pull from
1418 it.
1419 7. Every once in a while, git repack the public repository. Go back to
1421 step 5. and continue working.
1422 A recommended work cycle for an "individual developer" who does not
1424 have a "public" repository is somewhat different. It goes like this:
1425 1. Prepare your work repository, by git clone the public repository of
1427 the "project lead" (or a "subsystem maintainer", if you work on a
1428 subsystem). The URL used for the initial cloning is stored in the
1429 remote.origin.url configuration variable.
1430 2. Do your work in your repository on master branch.
1432 3. Run git fetch origin from the public repository of your upstream
1434 every once in a while. This does only the first half of git pull
1435 but does not merge. The head of the public repository is stored in
1436 .git/refs/remotes/origin/master.
1437 4. Use git cherry origin to see which ones of your patches were
1439 accepted, and/or use git rebase origin to port your unmerged
1440 changes forward to the updated upstream.
1441 5. Use git format-patch origin to prepare patches for e-mail
1443 submission to your upstream and send it out. Go back to step 2. and
1444 continue.
1445
1447 If you are coming from CVS background, the style of cooperation
1448 suggested in the previous section may be new to you. You do not have to
1449 worry. git supports "shared public repository" style of cooperation you
1450 are probably more familiar with as well.
1451 See gitcvs-migration(7) for the details.
1453
1455 It is likely that you will be working on more than one thing at a time.
1456 It is easy to manage those more-or-less independent tasks using
1457 branches with git.
1458 We have already seen how branches work previously, with "fun and work"
1460 example using two branches. The idea is the same if there are more than
1461 two branches. Let’s say you started out from "master" head, and have
1462 some new code in the "master" branch, and two independent fixes in the
1463 "commit-fix" and "diff-fix" branches:
1464 $ git show-branch
1466 ! [commit-fix] Fix commit message normalization.
1467 ! [diff-fix] Fix rename detection.
1468 * [master] Release candidate #1
1469 ---
1470 + [diff-fix] Fix rename detection.
1471 + [diff-fix~1] Better common substring algorithm.
1472 + [commit-fix] Fix commit message normalization.
1473 * [master] Release candidate #1
1474 ++* [diff-fix~2] Pretty-print messages.
1475 Both fixes are tested well, and at this point, you want to merge in
1478 both of them. You could merge in diff-fix first and then commit-fix
1479 next, like this:
1480 $ git merge -m "Merge fix in diff-fix" diff-fix
1482 $ git merge -m "Merge fix in commit-fix" commit-fix
1483 Which would result in:
1486 $ git show-branch
1488 ! [commit-fix] Fix commit message normalization.
1489 ! [diff-fix] Fix rename detection.
1490 * [master] Merge fix in commit-fix
1491 ---
1492 - [master] Merge fix in commit-fix
1493 + * [commit-fix] Fix commit message normalization.
1494 - [master~1] Merge fix in diff-fix
1495 +* [diff-fix] Fix rename detection.
1496 +* [diff-fix~1] Better common substring algorithm.
1497 * [master~2] Release candidate #1
1498 ++* [master~3] Pretty-print messages.
1499 However, there is no particular reason to merge in one branch first and
1502 the other next, when what you have are a set of truly independent
1503 changes (if the order mattered, then they are not independent by
1504 definition). You could instead merge those two branches into the
1505 current branch at once. First let’s undo what we just did and start
1506 over. We would want to get the master branch before these two merges by
1507 resetting it to master~2:
1508 $ git reset --hard master~2
1510 You can make sure git show-branch matches the state before those two
1513 git merge you just did. Then, instead of running two git merge commands
1514 in a row, you would merge these two branch heads (this is known as
1515 making an Octopus):
1516 $ git merge commit-fix diff-fix
1518 $ git show-branch
1519 ! [commit-fix] Fix commit message normalization.
1520 ! [diff-fix] Fix rename detection.
1521 * [master] Octopus merge of branches 'diff-fix' and 'commit-fix'
1522 ---
1523 - [master] Octopus merge of branches 'diff-fix' and 'commit-fix'
1524 + * [commit-fix] Fix commit message normalization.
1525 +* [diff-fix] Fix rename detection.
1526 +* [diff-fix~1] Better common substring algorithm.
1527 * [master~1] Release candidate #1
1528 ++* [master~2] Pretty-print messages.
1529 Note that you should not do Octopus because you can. An octopus is a
1532 valid thing to do and often makes it easier to view the commit history
1533 if you are merging more than two independent changes at the same time.
1534 However, if you have merge conflicts with any of the branches you are
1535 merging in and need to hand resolve, that is an indication that the
1536 development happened in those branches were not independent after all,
1537 and you should merge two at a time, documenting how you resolved the
1538 conflicts, and the reason why you preferred changes made in one side
1539 over the other. Otherwise it would make the project history harder to
1540 follow, not easier.
1541
1543 gittutorial(7), gittutorial-2(7), gitcvs-migration(7), git-help(1),
1544 Everyday git[3], The Git User’s Manual[1]
1545
1547 Part of the git(1) suite.
1548
1550 1. the GIT User Manual
1551 file:///usr/share/doc/git-1.7.4.4/user-manual.html
1552 2. Randy Dunlap’s presentation
1554 http://www.xenotime.net/linux/mentor/linux-mentoring-2006.pdf
1555 3. Everyday git
1557 file:///usr/share/doc/git-1.7.4.4/everyday.html
1558Git 1.7.4.4 04/11/2011 GITCORE-TUTORIAL(7)