Documentation / user-manual.txton commit regex: use regexec_buf() (b7d36ff)
   1Git User Manual
   2===============
   3
   4Git is a fast distributed revision control system.
   5
   6This manual is designed to be readable by someone with basic UNIX
   7command-line skills, but no previous knowledge of Git.
   8
   9<<repositories-and-branches>> and <<exploring-git-history>> explain how
  10to fetch and study a project using git--read these chapters to learn how
  11to build and test a particular version of a software project, search for
  12regressions, and so on.
  13
  14People needing to do actual development will also want to read
  15<<Developing-With-git>> and <<sharing-development>>.
  16
  17Further chapters cover more specialized topics.
  18
  19Comprehensive reference documentation is available through the man
  20pages, or linkgit:git-help[1] command.  For example, for the command
  21`git clone <repo>`, you can either use:
  22
  23------------------------------------------------
  24$ man git-clone
  25------------------------------------------------
  26
  27or:
  28
  29------------------------------------------------
  30$ git help clone
  31------------------------------------------------
  32
  33With the latter, you can use the manual viewer of your choice; see
  34linkgit:git-help[1] for more information.
  35
  36See also <<git-quick-start>> for a brief overview of Git commands,
  37without any explanation.
  38
  39Finally, see <<todo>> for ways that you can help make this manual more
  40complete.
  41
  42
  43[[repositories-and-branches]]
  44Repositories and Branches
  45=========================
  46
  47[[how-to-get-a-git-repository]]
  48How to get a Git repository
  49---------------------------
  50
  51It will be useful to have a Git repository to experiment with as you
  52read this manual.
  53
  54The best way to get one is by using the linkgit:git-clone[1] command to
  55download a copy of an existing repository.  If you don't already have a
  56project in mind, here are some interesting examples:
  57
  58------------------------------------------------
  59        # Git itself (approx. 40MB download):
  60$ git clone git://git.kernel.org/pub/scm/git/git.git
  61        # the Linux kernel (approx. 640MB download):
  62$ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
  63------------------------------------------------
  64
  65The initial clone may be time-consuming for a large project, but you
  66will only need to clone once.
  67
  68The clone command creates a new directory named after the project
  69(`git` or `linux` in the examples above).  After you cd into this
  70directory, you will see that it contains a copy of the project files,
  71called the <<def_working_tree,working tree>>, together with a special
  72top-level directory named `.git`, which contains all the information
  73about the history of the project.
  74
  75[[how-to-check-out]]
  76How to check out a different version of a project
  77-------------------------------------------------
  78
  79Git is best thought of as a tool for storing the history of a collection
  80of files.  It stores the history as a compressed collection of
  81interrelated snapshots of the project's contents.  In Git each such
  82version is called a <<def_commit,commit>>.
  83
  84Those snapshots aren't necessarily all arranged in a single line from
  85oldest to newest; instead, work may simultaneously proceed along
  86parallel lines of development, called <<def_branch,branches>>, which may
  87merge and diverge.
  88
  89A single Git repository can track development on multiple branches.  It
  90does this by keeping a list of <<def_head,heads>> which reference the
  91latest commit on each branch; the linkgit:git-branch[1] command shows
  92you the list of branch heads:
  93
  94------------------------------------------------
  95$ git branch
  96* master
  97------------------------------------------------
  98
  99A freshly cloned repository contains a single branch head, by default
 100named "master", with the working directory initialized to the state of
 101the project referred to by that branch head.
 102
 103Most projects also use <<def_tag,tags>>.  Tags, like heads, are
 104references into the project's history, and can be listed using the
 105linkgit:git-tag[1] command:
 106
 107------------------------------------------------
 108$ git tag -l
 109v2.6.11
 110v2.6.11-tree
 111v2.6.12
 112v2.6.12-rc2
 113v2.6.12-rc3
 114v2.6.12-rc4
 115v2.6.12-rc5
 116v2.6.12-rc6
 117v2.6.13
 118...
 119------------------------------------------------
 120
 121Tags are expected to always point at the same version of a project,
 122while heads are expected to advance as development progresses.
 123
 124Create a new branch head pointing to one of these versions and check it
 125out using linkgit:git-checkout[1]:
 126
 127------------------------------------------------
 128$ git checkout -b new v2.6.13
 129------------------------------------------------
 130
 131The working directory then reflects the contents that the project had
 132when it was tagged v2.6.13, and linkgit:git-branch[1] shows two
 133branches, with an asterisk marking the currently checked-out branch:
 134
 135------------------------------------------------
 136$ git branch
 137  master
 138* new
 139------------------------------------------------
 140
 141If you decide that you'd rather see version 2.6.17, you can modify
 142the current branch to point at v2.6.17 instead, with
 143
 144------------------------------------------------
 145$ git reset --hard v2.6.17
 146------------------------------------------------
 147
 148Note that if the current branch head was your only reference to a
 149particular point in history, then resetting that branch may leave you
 150with no way to find the history it used to point to; so use this command
 151carefully.
 152
 153[[understanding-commits]]
 154Understanding History: Commits
 155------------------------------
 156
 157Every change in the history of a project is represented by a commit.
 158The linkgit:git-show[1] command shows the most recent commit on the
 159current branch:
 160
 161------------------------------------------------
 162$ git show
 163commit 17cf781661e6d38f737f15f53ab552f1e95960d7
 164Author: Linus Torvalds <torvalds@ppc970.osdl.org.(none)>
 165Date:   Tue Apr 19 14:11:06 2005 -0700
 166
 167    Remove duplicate getenv(DB_ENVIRONMENT) call
 168
 169    Noted by Tony Luck.
 170
 171diff --git a/init-db.c b/init-db.c
 172index 65898fa..b002dc6 100644
 173--- a/init-db.c
 174+++ b/init-db.c
 175@@ -7,7 +7,7 @@
 176 
 177 int main(int argc, char **argv)
 178 {
 179-       char *sha1_dir = getenv(DB_ENVIRONMENT), *path;
 180+       char *sha1_dir, *path;
 181        int len, i;
 182 
 183        if (mkdir(".git", 0755) < 0) {
 184------------------------------------------------
 185
 186As you can see, a commit shows who made the latest change, what they
 187did, and why.
 188
 189Every commit has a 40-hexdigit id, sometimes called the "object name" or the
 190"SHA-1 id", shown on the first line of the `git show` output.  You can usually
 191refer to a commit by a shorter name, such as a tag or a branch name, but this
 192longer name can also be useful.  Most importantly, it is a globally unique
 193name for this commit: so if you tell somebody else the object name (for
 194example in email), then you are guaranteed that name will refer to the same
 195commit in their repository that it does in yours (assuming their repository
 196has that commit at all).  Since the object name is computed as a hash over the
 197contents of the commit, you are guaranteed that the commit can never change
 198without its name also changing.
 199
 200In fact, in <<git-concepts>> we shall see that everything stored in Git
 201history, including file data and directory contents, is stored in an object
 202with a name that is a hash of its contents.
 203
 204[[understanding-reachability]]
 205Understanding history: commits, parents, and reachability
 206~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 207
 208Every commit (except the very first commit in a project) also has a
 209parent commit which shows what happened before this commit.
 210Following the chain of parents will eventually take you back to the
 211beginning of the project.
 212
 213However, the commits do not form a simple list; Git allows lines of
 214development to diverge and then reconverge, and the point where two
 215lines of development reconverge is called a "merge".  The commit
 216representing a merge can therefore have more than one parent, with
 217each parent representing the most recent commit on one of the lines
 218of development leading to that point.
 219
 220The best way to see how this works is using the linkgit:gitk[1]
 221command; running gitk now on a Git repository and looking for merge
 222commits will help understand how Git organizes history.
 223
 224In the following, we say that commit X is "reachable" from commit Y
 225if commit X is an ancestor of commit Y.  Equivalently, you could say
 226that Y is a descendant of X, or that there is a chain of parents
 227leading from commit Y to commit X.
 228
 229[[history-diagrams]]
 230Understanding history: History diagrams
 231~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 232
 233We will sometimes represent Git history using diagrams like the one
 234below.  Commits are shown as "o", and the links between them with
 235lines drawn with - / and \.  Time goes left to right:
 236
 237
 238................................................
 239         o--o--o <-- Branch A
 240        /
 241 o--o--o <-- master
 242        \
 243         o--o--o <-- Branch B
 244................................................
 245
 246If we need to talk about a particular commit, the character "o" may
 247be replaced with another letter or number.
 248
 249[[what-is-a-branch]]
 250Understanding history: What is a branch?
 251~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 252
 253When we need to be precise, we will use the word "branch" to mean a line
 254of development, and "branch head" (or just "head") to mean a reference
 255to the most recent commit on a branch.  In the example above, the branch
 256head named "A" is a pointer to one particular commit, but we refer to
 257the line of three commits leading up to that point as all being part of
 258"branch A".
 259
 260However, when no confusion will result, we often just use the term
 261"branch" both for branches and for branch heads.
 262
 263[[manipulating-branches]]
 264Manipulating branches
 265---------------------
 266
 267Creating, deleting, and modifying branches is quick and easy; here's
 268a summary of the commands:
 269
 270`git branch`::
 271        list all branches.
 272`git branch <branch>`::
 273        create a new branch named `<branch>`, referencing the same
 274        point in history as the current branch.
 275`git branch <branch> <start-point>`::
 276        create a new branch named `<branch>`, referencing
 277        `<start-point>`, which may be specified any way you like,
 278        including using a branch name or a tag name.
 279`git branch -d <branch>`::
 280        delete the branch `<branch>`; if the branch is not fully
 281        merged in its upstream branch or contained in the current branch,
 282        this command will fail with a warning.
 283`git branch -D <branch>`::
 284        delete the branch `<branch>` irrespective of its merged status.
 285`git checkout <branch>`::
 286        make the current branch `<branch>`, updating the working
 287        directory to reflect the version referenced by `<branch>`.
 288`git checkout -b <new> <start-point>`::
 289        create a new branch `<new>` referencing `<start-point>`, and
 290        check it out.
 291
 292The special symbol "HEAD" can always be used to refer to the current
 293branch.  In fact, Git uses a file named `HEAD` in the `.git` directory
 294to remember which branch is current:
 295
 296------------------------------------------------
 297$ cat .git/HEAD
 298ref: refs/heads/master
 299------------------------------------------------
 300
 301[[detached-head]]
 302Examining an old version without creating a new branch
 303------------------------------------------------------
 304
 305The `git checkout` command normally expects a branch head, but will also
 306accept an arbitrary commit; for example, you can check out the commit
 307referenced by a tag:
 308
 309------------------------------------------------
 310$ git checkout v2.6.17
 311Note: checking out 'v2.6.17'.
 312
 313You are in 'detached HEAD' state. You can look around, make experimental
 314changes and commit them, and you can discard any commits you make in this
 315state without impacting any branches by performing another checkout.
 316
 317If you want to create a new branch to retain commits you create, you may
 318do so (now or later) by using -b with the checkout command again. Example:
 319
 320  git checkout -b new_branch_name
 321
 322HEAD is now at 427abfa... Linux v2.6.17
 323------------------------------------------------
 324
 325The HEAD then refers to the SHA-1 of the commit instead of to a branch,
 326and git branch shows that you are no longer on a branch:
 327
 328------------------------------------------------
 329$ cat .git/HEAD
 330427abfa28afedffadfca9dd8b067eb6d36bac53f
 331$ git branch
 332* (detached from v2.6.17)
 333  master
 334------------------------------------------------
 335
 336In this case we say that the HEAD is "detached".
 337
 338This is an easy way to check out a particular version without having to
 339make up a name for the new branch.   You can still create a new branch
 340(or tag) for this version later if you decide to.
 341
 342[[examining-remote-branches]]
 343Examining branches from a remote repository
 344-------------------------------------------
 345
 346The "master" branch that was created at the time you cloned is a copy
 347of the HEAD in the repository that you cloned from.  That repository
 348may also have had other branches, though, and your local repository
 349keeps branches which track each of those remote branches, called
 350remote-tracking branches, which you
 351can view using the `-r` option to linkgit:git-branch[1]:
 352
 353------------------------------------------------
 354$ git branch -r
 355  origin/HEAD
 356  origin/html
 357  origin/maint
 358  origin/man
 359  origin/master
 360  origin/next
 361  origin/pu
 362  origin/todo
 363------------------------------------------------
 364
 365In this example, "origin" is called a remote repository, or "remote"
 366for short. The branches of this repository are called "remote
 367branches" from our point of view. The remote-tracking branches listed
 368above were created based on the remote branches at clone time and will
 369be updated by `git fetch` (hence `git pull`) and `git push`. See
 370<<Updating-a-repository-With-git-fetch>> for details.
 371
 372You might want to build on one of these remote-tracking branches
 373on a branch of your own, just as you would for a tag:
 374
 375------------------------------------------------
 376$ git checkout -b my-todo-copy origin/todo
 377------------------------------------------------
 378
 379You can also check out `origin/todo` directly to examine it or
 380write a one-off patch.  See <<detached-head,detached head>>.
 381
 382Note that the name "origin" is just the name that Git uses by default
 383to refer to the repository that you cloned from.
 384
 385[[how-git-stores-references]]
 386Naming branches, tags, and other references
 387-------------------------------------------
 388
 389Branches, remote-tracking branches, and tags are all references to
 390commits.  All references are named with a slash-separated path name
 391starting with `refs`; the names we've been using so far are actually
 392shorthand:
 393
 394        - The branch `test` is short for `refs/heads/test`.
 395        - The tag `v2.6.18` is short for `refs/tags/v2.6.18`.
 396        - `origin/master` is short for `refs/remotes/origin/master`.
 397
 398The full name is occasionally useful if, for example, there ever
 399exists a tag and a branch with the same name.
 400
 401(Newly created refs are actually stored in the `.git/refs` directory,
 402under the path given by their name.  However, for efficiency reasons
 403they may also be packed together in a single file; see
 404linkgit:git-pack-refs[1]).
 405
 406As another useful shortcut, the "HEAD" of a repository can be referred
 407to just using the name of that repository.  So, for example, "origin"
 408is usually a shortcut for the HEAD branch in the repository "origin".
 409
 410For the complete list of paths which Git checks for references, and
 411the order it uses to decide which to choose when there are multiple
 412references with the same shorthand name, see the "SPECIFYING
 413REVISIONS" section of linkgit:gitrevisions[7].
 414
 415[[Updating-a-repository-With-git-fetch]]
 416Updating a repository with git fetch
 417------------------------------------
 418
 419After you clone a repository and commit a few changes of your own, you
 420may wish to check the original repository for updates.
 421
 422The `git-fetch` command, with no arguments, will update all of the
 423remote-tracking branches to the latest version found in the original
 424repository.  It will not touch any of your own branches--not even the
 425"master" branch that was created for you on clone.
 426
 427[[fetching-branches]]
 428Fetching branches from other repositories
 429-----------------------------------------
 430
 431You can also track branches from repositories other than the one you
 432cloned from, using linkgit:git-remote[1]:
 433
 434-------------------------------------------------
 435$ git remote add staging git://git.kernel.org/.../gregkh/staging.git
 436$ git fetch staging
 437...
 438From git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging
 439 * [new branch]      master     -> staging/master
 440 * [new branch]      staging-linus -> staging/staging-linus
 441 * [new branch]      staging-next -> staging/staging-next
 442-------------------------------------------------
 443
 444New remote-tracking branches will be stored under the shorthand name
 445that you gave `git remote add`, in this case `staging`:
 446
 447-------------------------------------------------
 448$ git branch -r
 449  origin/HEAD -> origin/master
 450  origin/master
 451  staging/master
 452  staging/staging-linus
 453  staging/staging-next
 454-------------------------------------------------
 455
 456If you run `git fetch <remote>` later, the remote-tracking branches
 457for the named `<remote>` will be updated.
 458
 459If you examine the file `.git/config`, you will see that Git has added
 460a new stanza:
 461
 462-------------------------------------------------
 463$ cat .git/config
 464...
 465[remote "staging"]
 466        url = git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging.git
 467        fetch = +refs/heads/*:refs/remotes/staging/*
 468...
 469-------------------------------------------------
 470
 471This is what causes Git to track the remote's branches; you may modify
 472or delete these configuration options by editing `.git/config` with a
 473text editor.  (See the "CONFIGURATION FILE" section of
 474linkgit:git-config[1] for details.)
 475
 476[[exploring-git-history]]
 477Exploring Git history
 478=====================
 479
 480Git is best thought of as a tool for storing the history of a
 481collection of files.  It does this by storing compressed snapshots of
 482the contents of a file hierarchy, together with "commits" which show
 483the relationships between these snapshots.
 484
 485Git provides extremely flexible and fast tools for exploring the
 486history of a project.
 487
 488We start with one specialized tool that is useful for finding the
 489commit that introduced a bug into a project.
 490
 491[[using-bisect]]
 492How to use bisect to find a regression
 493--------------------------------------
 494
 495Suppose version 2.6.18 of your project worked, but the version at
 496"master" crashes.  Sometimes the best way to find the cause of such a
 497regression is to perform a brute-force search through the project's
 498history to find the particular commit that caused the problem.  The
 499linkgit:git-bisect[1] command can help you do this:
 500
 501-------------------------------------------------
 502$ git bisect start
 503$ git bisect good v2.6.18
 504$ git bisect bad master
 505Bisecting: 3537 revisions left to test after this
 506[65934a9a028b88e83e2b0f8b36618fe503349f8e] BLOCK: Make USB storage depend on SCSI rather than selecting it [try #6]
 507-------------------------------------------------
 508
 509If you run `git branch` at this point, you'll see that Git has
 510temporarily moved you in "(no branch)". HEAD is now detached from any
 511branch and points directly to a commit (with commit id 65934...) that
 512is reachable from "master" but not from v2.6.18. Compile and test it,
 513and see whether it crashes. Assume it does crash. Then:
 514
 515-------------------------------------------------
 516$ git bisect bad
 517Bisecting: 1769 revisions left to test after this
 518[7eff82c8b1511017ae605f0c99ac275a7e21b867] i2c-core: Drop useless bitmaskings
 519-------------------------------------------------
 520
 521checks out an older version.  Continue like this, telling Git at each
 522stage whether the version it gives you is good or bad, and notice
 523that the number of revisions left to test is cut approximately in
 524half each time.
 525
 526After about 13 tests (in this case), it will output the commit id of
 527the guilty commit.  You can then examine the commit with
 528linkgit:git-show[1], find out who wrote it, and mail them your bug
 529report with the commit id.  Finally, run
 530
 531-------------------------------------------------
 532$ git bisect reset
 533-------------------------------------------------
 534
 535to return you to the branch you were on before.
 536
 537Note that the version which `git bisect` checks out for you at each
 538point is just a suggestion, and you're free to try a different
 539version if you think it would be a good idea.  For example,
 540occasionally you may land on a commit that broke something unrelated;
 541run
 542
 543-------------------------------------------------
 544$ git bisect visualize
 545-------------------------------------------------
 546
 547which will run gitk and label the commit it chose with a marker that
 548says "bisect".  Choose a safe-looking commit nearby, note its commit
 549id, and check it out with:
 550
 551-------------------------------------------------
 552$ git reset --hard fb47ddb2db...
 553-------------------------------------------------
 554
 555then test, run `bisect good` or `bisect bad` as appropriate, and
 556continue.
 557
 558Instead of `git bisect visualize` and then `git reset --hard
 559fb47ddb2db...`, you might just want to tell Git that you want to skip
 560the current commit:
 561
 562-------------------------------------------------
 563$ git bisect skip
 564-------------------------------------------------
 565
 566In this case, though, Git may not eventually be able to tell the first
 567bad one between some first skipped commits and a later bad commit.
 568
 569There are also ways to automate the bisecting process if you have a
 570test script that can tell a good from a bad commit. See
 571linkgit:git-bisect[1] for more information about this and other `git
 572bisect` features.
 573
 574[[naming-commits]]
 575Naming commits
 576--------------
 577
 578We have seen several ways of naming commits already:
 579
 580        - 40-hexdigit object name
 581        - branch name: refers to the commit at the head of the given
 582          branch
 583        - tag name: refers to the commit pointed to by the given tag
 584          (we've seen branches and tags are special cases of
 585          <<how-git-stores-references,references>>).
 586        - HEAD: refers to the head of the current branch
 587
 588There are many more; see the "SPECIFYING REVISIONS" section of the
 589linkgit:gitrevisions[7] man page for the complete list of ways to
 590name revisions.  Some examples:
 591
 592-------------------------------------------------
 593$ git show fb47ddb2 # the first few characters of the object name
 594                    # are usually enough to specify it uniquely
 595$ git show HEAD^    # the parent of the HEAD commit
 596$ git show HEAD^^   # the grandparent
 597$ git show HEAD~4   # the great-great-grandparent
 598-------------------------------------------------
 599
 600Recall that merge commits may have more than one parent; by default,
 601`^` and `~` follow the first parent listed in the commit, but you can
 602also choose:
 603
 604-------------------------------------------------
 605$ git show HEAD^1   # show the first parent of HEAD
 606$ git show HEAD^2   # show the second parent of HEAD
 607-------------------------------------------------
 608
 609In addition to HEAD, there are several other special names for
 610commits:
 611
 612Merges (to be discussed later), as well as operations such as
 613`git reset`, which change the currently checked-out commit, generally
 614set ORIG_HEAD to the value HEAD had before the current operation.
 615
 616The `git fetch` operation always stores the head of the last fetched
 617branch in FETCH_HEAD.  For example, if you run `git fetch` without
 618specifying a local branch as the target of the operation
 619
 620-------------------------------------------------
 621$ git fetch git://example.com/proj.git theirbranch
 622-------------------------------------------------
 623
 624the fetched commits will still be available from FETCH_HEAD.
 625
 626When we discuss merges we'll also see the special name MERGE_HEAD,
 627which refers to the other branch that we're merging in to the current
 628branch.
 629
 630The linkgit:git-rev-parse[1] command is a low-level command that is
 631occasionally useful for translating some name for a commit to the object
 632name for that commit:
 633
 634-------------------------------------------------
 635$ git rev-parse origin
 636e05db0fd4f31dde7005f075a84f96b360d05984b
 637-------------------------------------------------
 638
 639[[creating-tags]]
 640Creating tags
 641-------------
 642
 643We can also create a tag to refer to a particular commit; after
 644running
 645
 646-------------------------------------------------
 647$ git tag stable-1 1b2e1d63ff
 648-------------------------------------------------
 649
 650You can use `stable-1` to refer to the commit 1b2e1d63ff.
 651
 652This creates a "lightweight" tag.  If you would also like to include a
 653comment with the tag, and possibly sign it cryptographically, then you
 654should create a tag object instead; see the linkgit:git-tag[1] man page
 655for details.
 656
 657[[browsing-revisions]]
 658Browsing revisions
 659------------------
 660
 661The linkgit:git-log[1] command can show lists of commits.  On its
 662own, it shows all commits reachable from the parent commit; but you
 663can also make more specific requests:
 664
 665-------------------------------------------------
 666$ git log v2.5..        # commits since (not reachable from) v2.5
 667$ git log test..master  # commits reachable from master but not test
 668$ git log master..test  # ...reachable from test but not master
 669$ git log master...test # ...reachable from either test or master,
 670                        #    but not both
 671$ git log --since="2 weeks ago" # commits from the last 2 weeks
 672$ git log Makefile      # commits which modify Makefile
 673$ git log fs/           # ... which modify any file under fs/
 674$ git log -S'foo()'     # commits which add or remove any file data
 675                        # matching the string 'foo()'
 676-------------------------------------------------
 677
 678And of course you can combine all of these; the following finds
 679commits since v2.5 which touch the `Makefile` or any file under `fs`:
 680
 681-------------------------------------------------
 682$ git log v2.5.. Makefile fs/
 683-------------------------------------------------
 684
 685You can also ask git log to show patches:
 686
 687-------------------------------------------------
 688$ git log -p
 689-------------------------------------------------
 690
 691See the `--pretty` option in the linkgit:git-log[1] man page for more
 692display options.
 693
 694Note that git log starts with the most recent commit and works
 695backwards through the parents; however, since Git history can contain
 696multiple independent lines of development, the particular order that
 697commits are listed in may be somewhat arbitrary.
 698
 699[[generating-diffs]]
 700Generating diffs
 701----------------
 702
 703You can generate diffs between any two versions using
 704linkgit:git-diff[1]:
 705
 706-------------------------------------------------
 707$ git diff master..test
 708-------------------------------------------------
 709
 710That will produce the diff between the tips of the two branches.  If
 711you'd prefer to find the diff from their common ancestor to test, you
 712can use three dots instead of two:
 713
 714-------------------------------------------------
 715$ git diff master...test
 716-------------------------------------------------
 717
 718Sometimes what you want instead is a set of patches; for this you can
 719use linkgit:git-format-patch[1]:
 720
 721-------------------------------------------------
 722$ git format-patch master..test
 723-------------------------------------------------
 724
 725will generate a file with a patch for each commit reachable from test
 726but not from master.
 727
 728[[viewing-old-file-versions]]
 729Viewing old file versions
 730-------------------------
 731
 732You can always view an old version of a file by just checking out the
 733correct revision first.  But sometimes it is more convenient to be
 734able to view an old version of a single file without checking
 735anything out; this command does that:
 736
 737-------------------------------------------------
 738$ git show v2.5:fs/locks.c
 739-------------------------------------------------
 740
 741Before the colon may be anything that names a commit, and after it
 742may be any path to a file tracked by Git.
 743
 744[[history-examples]]
 745Examples
 746--------
 747
 748[[counting-commits-on-a-branch]]
 749Counting the number of commits on a branch
 750~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 751
 752Suppose you want to know how many commits you've made on `mybranch`
 753since it diverged from `origin`:
 754
 755-------------------------------------------------
 756$ git log --pretty=oneline origin..mybranch | wc -l
 757-------------------------------------------------
 758
 759Alternatively, you may often see this sort of thing done with the
 760lower-level command linkgit:git-rev-list[1], which just lists the SHA-1's
 761of all the given commits:
 762
 763-------------------------------------------------
 764$ git rev-list origin..mybranch | wc -l
 765-------------------------------------------------
 766
 767[[checking-for-equal-branches]]
 768Check whether two branches point at the same history
 769~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 770
 771Suppose you want to check whether two branches point at the same point
 772in history.
 773
 774-------------------------------------------------
 775$ git diff origin..master
 776-------------------------------------------------
 777
 778will tell you whether the contents of the project are the same at the
 779two branches; in theory, however, it's possible that the same project
 780contents could have been arrived at by two different historical
 781routes.  You could compare the object names:
 782
 783-------------------------------------------------
 784$ git rev-list origin
 785e05db0fd4f31dde7005f075a84f96b360d05984b
 786$ git rev-list master
 787e05db0fd4f31dde7005f075a84f96b360d05984b
 788-------------------------------------------------
 789
 790Or you could recall that the `...` operator selects all commits
 791reachable from either one reference or the other but not
 792both; so
 793
 794-------------------------------------------------
 795$ git log origin...master
 796-------------------------------------------------
 797
 798will return no commits when the two branches are equal.
 799
 800[[finding-tagged-descendants]]
 801Find first tagged version including a given fix
 802~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 803
 804Suppose you know that the commit e05db0fd fixed a certain problem.
 805You'd like to find the earliest tagged release that contains that
 806fix.
 807
 808Of course, there may be more than one answer--if the history branched
 809after commit e05db0fd, then there could be multiple "earliest" tagged
 810releases.
 811
 812You could just visually inspect the commits since e05db0fd:
 813
 814-------------------------------------------------
 815$ gitk e05db0fd..
 816-------------------------------------------------
 817
 818or you can use linkgit:git-name-rev[1], which will give the commit a
 819name based on any tag it finds pointing to one of the commit's
 820descendants:
 821
 822-------------------------------------------------
 823$ git name-rev --tags e05db0fd
 824e05db0fd tags/v1.5.0-rc1^0~23
 825-------------------------------------------------
 826
 827The linkgit:git-describe[1] command does the opposite, naming the
 828revision using a tag on which the given commit is based:
 829
 830-------------------------------------------------
 831$ git describe e05db0fd
 832v1.5.0-rc0-260-ge05db0f
 833-------------------------------------------------
 834
 835but that may sometimes help you guess which tags might come after the
 836given commit.
 837
 838If you just want to verify whether a given tagged version contains a
 839given commit, you could use linkgit:git-merge-base[1]:
 840
 841-------------------------------------------------
 842$ git merge-base e05db0fd v1.5.0-rc1
 843e05db0fd4f31dde7005f075a84f96b360d05984b
 844-------------------------------------------------
 845
 846The merge-base command finds a common ancestor of the given commits,
 847and always returns one or the other in the case where one is a
 848descendant of the other; so the above output shows that e05db0fd
 849actually is an ancestor of v1.5.0-rc1.
 850
 851Alternatively, note that
 852
 853-------------------------------------------------
 854$ git log v1.5.0-rc1..e05db0fd
 855-------------------------------------------------
 856
 857will produce empty output if and only if v1.5.0-rc1 includes e05db0fd,
 858because it outputs only commits that are not reachable from v1.5.0-rc1.
 859
 860As yet another alternative, the linkgit:git-show-branch[1] command lists
 861the commits reachable from its arguments with a display on the left-hand
 862side that indicates which arguments that commit is reachable from.
 863So, if you run something like
 864
 865-------------------------------------------------
 866$ git show-branch e05db0fd v1.5.0-rc0 v1.5.0-rc1 v1.5.0-rc2
 867! [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
 868available
 869 ! [v1.5.0-rc0] GIT v1.5.0 preview
 870  ! [v1.5.0-rc1] GIT v1.5.0-rc1
 871   ! [v1.5.0-rc2] GIT v1.5.0-rc2
 872...
 873-------------------------------------------------
 874
 875then a line like
 876
 877-------------------------------------------------
 878+ ++ [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
 879available
 880-------------------------------------------------
 881
 882shows that e05db0fd is reachable from itself, from v1.5.0-rc1,
 883and from v1.5.0-rc2, and not from v1.5.0-rc0.
 884
 885[[showing-commits-unique-to-a-branch]]
 886Showing commits unique to a given branch
 887~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 888
 889Suppose you would like to see all the commits reachable from the branch
 890head named `master` but not from any other head in your repository.
 891
 892We can list all the heads in this repository with
 893linkgit:git-show-ref[1]:
 894
 895-------------------------------------------------
 896$ git show-ref --heads
 897bf62196b5e363d73353a9dcf094c59595f3153b7 refs/heads/core-tutorial
 898db768d5504c1bb46f63ee9d6e1772bd047e05bf9 refs/heads/maint
 899a07157ac624b2524a059a3414e99f6f44bebc1e7 refs/heads/master
 90024dbc180ea14dc1aebe09f14c8ecf32010690627 refs/heads/tutorial-2
 9011e87486ae06626c2f31eaa63d26fc0fd646c8af2 refs/heads/tutorial-fixes
 902-------------------------------------------------
 903
 904We can get just the branch-head names, and remove `master`, with
 905the help of the standard utilities cut and grep:
 906
 907-------------------------------------------------
 908$ git show-ref --heads | cut -d' ' -f2 | grep -v '^refs/heads/master'
 909refs/heads/core-tutorial
 910refs/heads/maint
 911refs/heads/tutorial-2
 912refs/heads/tutorial-fixes
 913-------------------------------------------------
 914
 915And then we can ask to see all the commits reachable from master
 916but not from these other heads:
 917
 918-------------------------------------------------
 919$ gitk master --not $( git show-ref --heads | cut -d' ' -f2 |
 920                                grep -v '^refs/heads/master' )
 921-------------------------------------------------
 922
 923Obviously, endless variations are possible; for example, to see all
 924commits reachable from some head but not from any tag in the repository:
 925
 926-------------------------------------------------
 927$ gitk $( git show-ref --heads ) --not  $( git show-ref --tags )
 928-------------------------------------------------
 929
 930(See linkgit:gitrevisions[7] for explanations of commit-selecting
 931syntax such as `--not`.)
 932
 933[[making-a-release]]
 934Creating a changelog and tarball for a software release
 935~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 936
 937The linkgit:git-archive[1] command can create a tar or zip archive from
 938any version of a project; for example:
 939
 940-------------------------------------------------
 941$ git archive -o latest.tar.gz --prefix=project/ HEAD
 942-------------------------------------------------
 943
 944will use HEAD to produce a gzipped tar archive in which each filename
 945is preceded by `project/`.  The output file format is inferred from
 946the output file extension if possible, see linkgit:git-archive[1] for
 947details.
 948
 949Versions of Git older than 1.7.7 don't know about the `tar.gz` format,
 950you'll need to use gzip explicitly:
 951
 952-------------------------------------------------
 953$ git archive --format=tar --prefix=project/ HEAD | gzip >latest.tar.gz
 954-------------------------------------------------
 955
 956If you're releasing a new version of a software project, you may want
 957to simultaneously make a changelog to include in the release
 958announcement.
 959
 960Linus Torvalds, for example, makes new kernel releases by tagging them,
 961then running:
 962
 963-------------------------------------------------
 964$ release-script 2.6.12 2.6.13-rc6 2.6.13-rc7
 965-------------------------------------------------
 966
 967where release-script is a shell script that looks like:
 968
 969-------------------------------------------------
 970#!/bin/sh
 971stable="$1"
 972last="$2"
 973new="$3"
 974echo "# git tag v$new"
 975echo "git archive --prefix=linux-$new/ v$new | gzip -9 > ../linux-$new.tar.gz"
 976echo "git diff v$stable v$new | gzip -9 > ../patch-$new.gz"
 977echo "git log --no-merges v$new ^v$last > ../ChangeLog-$new"
 978echo "git shortlog --no-merges v$new ^v$last > ../ShortLog"
 979echo "git diff --stat --summary -M v$last v$new > ../diffstat-$new"
 980-------------------------------------------------
 981
 982and then he just cut-and-pastes the output commands after verifying that
 983they look OK.
 984
 985[[Finding-commits-With-given-Content]]
 986Finding commits referencing a file with given content
 987~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 988
 989Somebody hands you a copy of a file, and asks which commits modified a
 990file such that it contained the given content either before or after the
 991commit.  You can find out with this:
 992
 993-------------------------------------------------
 994$  git log --raw --abbrev=40 --pretty=oneline |
 995        grep -B 1 `git hash-object filename`
 996-------------------------------------------------
 997
 998Figuring out why this works is left as an exercise to the (advanced)
 999student.  The linkgit:git-log[1], linkgit:git-diff-tree[1], and
1000linkgit:git-hash-object[1] man pages may prove helpful.
1001
1002[[Developing-With-git]]
1003Developing with Git
1004===================
1005
1006[[telling-git-your-name]]
1007Telling Git your name
1008---------------------
1009
1010Before creating any commits, you should introduce yourself to Git.
1011The easiest way to do so is to use linkgit:git-config[1]:
1012
1013------------------------------------------------
1014$ git config --global user.name 'Your Name Comes Here'
1015$ git config --global user.email 'you@yourdomain.example.com'
1016------------------------------------------------
1017
1018Which will add the following to a file named `.gitconfig` in your
1019home directory:
1020
1021------------------------------------------------
1022[user]
1023        name = Your Name Comes Here
1024        email = you@yourdomain.example.com
1025------------------------------------------------
1026
1027See the "CONFIGURATION FILE" section of linkgit:git-config[1] for
1028details on the configuration file.  The file is plain text, so you can
1029also edit it with your favorite editor.
1030
1031
1032[[creating-a-new-repository]]
1033Creating a new repository
1034-------------------------
1035
1036Creating a new repository from scratch is very easy:
1037
1038-------------------------------------------------
1039$ mkdir project
1040$ cd project
1041$ git init
1042-------------------------------------------------
1043
1044If you have some initial content (say, a tarball):
1045
1046-------------------------------------------------
1047$ tar xzvf project.tar.gz
1048$ cd project
1049$ git init
1050$ git add . # include everything below ./ in the first commit:
1051$ git commit
1052-------------------------------------------------
1053
1054[[how-to-make-a-commit]]
1055How to make a commit
1056--------------------
1057
1058Creating a new commit takes three steps:
1059
1060        1. Making some changes to the working directory using your
1061           favorite editor.
1062        2. Telling Git about your changes.
1063        3. Creating the commit using the content you told Git about
1064           in step 2.
1065
1066In practice, you can interleave and repeat steps 1 and 2 as many
1067times as you want: in order to keep track of what you want committed
1068at step 3, Git maintains a snapshot of the tree's contents in a
1069special staging area called "the index."
1070
1071At the beginning, the content of the index will be identical to
1072that of the HEAD.  The command `git diff --cached`, which shows
1073the difference between the HEAD and the index, should therefore
1074produce no output at that point.
1075
1076Modifying the index is easy:
1077
1078To update the index with the contents of a new or modified file, use
1079
1080-------------------------------------------------
1081$ git add path/to/file
1082-------------------------------------------------
1083
1084To remove a file from the index and from the working tree, use
1085
1086-------------------------------------------------
1087$ git rm path/to/file
1088-------------------------------------------------
1089
1090After each step you can verify that
1091
1092-------------------------------------------------
1093$ git diff --cached
1094-------------------------------------------------
1095
1096always shows the difference between the HEAD and the index file--this
1097is what you'd commit if you created the commit now--and that
1098
1099-------------------------------------------------
1100$ git diff
1101-------------------------------------------------
1102
1103shows the difference between the working tree and the index file.
1104
1105Note that `git add` always adds just the current contents of a file
1106to the index; further changes to the same file will be ignored unless
1107you run `git add` on the file again.
1108
1109When you're ready, just run
1110
1111-------------------------------------------------
1112$ git commit
1113-------------------------------------------------
1114
1115and Git will prompt you for a commit message and then create the new
1116commit.  Check to make sure it looks like what you expected with
1117
1118-------------------------------------------------
1119$ git show
1120-------------------------------------------------
1121
1122As a special shortcut,
1123
1124-------------------------------------------------
1125$ git commit -a
1126-------------------------------------------------
1127
1128will update the index with any files that you've modified or removed
1129and create a commit, all in one step.
1130
1131A number of commands are useful for keeping track of what you're
1132about to commit:
1133
1134-------------------------------------------------
1135$ git diff --cached # difference between HEAD and the index; what
1136                    # would be committed if you ran "commit" now.
1137$ git diff          # difference between the index file and your
1138                    # working directory; changes that would not
1139                    # be included if you ran "commit" now.
1140$ git diff HEAD     # difference between HEAD and working tree; what
1141                    # would be committed if you ran "commit -a" now.
1142$ git status        # a brief per-file summary of the above.
1143-------------------------------------------------
1144
1145You can also use linkgit:git-gui[1] to create commits, view changes in
1146the index and the working tree files, and individually select diff hunks
1147for inclusion in the index (by right-clicking on the diff hunk and
1148choosing "Stage Hunk For Commit").
1149
1150[[creating-good-commit-messages]]
1151Creating good commit messages
1152-----------------------------
1153
1154Though not required, it's a good idea to begin the commit message
1155with a single short (less than 50 character) line summarizing the
1156change, followed by a blank line and then a more thorough
1157description.  The text up to the first blank line in a commit
1158message is treated as the commit title, and that title is used
1159throughout Git.  For example, linkgit:git-format-patch[1] turns a
1160commit into email, and it uses the title on the Subject line and the
1161rest of the commit in the body.
1162
1163
1164[[ignoring-files]]
1165Ignoring files
1166--------------
1167
1168A project will often generate files that you do 'not' want to track with Git.
1169This typically includes files generated by a build process or temporary
1170backup files made by your editor. Of course, 'not' tracking files with Git
1171is just a matter of 'not' calling `git add` on them. But it quickly becomes
1172annoying to have these untracked files lying around; e.g. they make
1173`git add .` practically useless, and they keep showing up in the output of
1174`git status`.
1175
1176You can tell Git to ignore certain files by creating a file called
1177`.gitignore` in the top level of your working directory, with contents
1178such as:
1179
1180-------------------------------------------------
1181# Lines starting with '#' are considered comments.
1182# Ignore any file named foo.txt.
1183foo.txt
1184# Ignore (generated) html files,
1185*.html
1186# except foo.html which is maintained by hand.
1187!foo.html
1188# Ignore objects and archives.
1189*.[oa]
1190-------------------------------------------------
1191
1192See linkgit:gitignore[5] for a detailed explanation of the syntax.  You can
1193also place .gitignore files in other directories in your working tree, and they
1194will apply to those directories and their subdirectories.  The `.gitignore`
1195files can be added to your repository like any other files (just run `git add
1196.gitignore` and `git commit`, as usual), which is convenient when the exclude
1197patterns (such as patterns matching build output files) would also make sense
1198for other users who clone your repository.
1199
1200If you wish the exclude patterns to affect only certain repositories
1201(instead of every repository for a given project), you may instead put
1202them in a file in your repository named `.git/info/exclude`, or in any
1203file specified by the `core.excludesFile` configuration variable.
1204Some Git commands can also take exclude patterns directly on the
1205command line.  See linkgit:gitignore[5] for the details.
1206
1207[[how-to-merge]]
1208How to merge
1209------------
1210
1211You can rejoin two diverging branches of development using
1212linkgit:git-merge[1]:
1213
1214-------------------------------------------------
1215$ git merge branchname
1216-------------------------------------------------
1217
1218merges the development in the branch `branchname` into the current
1219branch.
1220
1221A merge is made by combining the changes made in `branchname` and the
1222changes made up to the latest commit in your current branch since
1223their histories forked. The work tree is overwritten by the result of
1224the merge when this combining is done cleanly, or overwritten by a
1225half-merged results when this combining results in conflicts.
1226Therefore, if you have uncommitted changes touching the same files as
1227the ones impacted by the merge, Git will refuse to proceed. Most of
1228the time, you will want to commit your changes before you can merge,
1229and if you don't, then linkgit:git-stash[1] can take these changes
1230away while you're doing the merge, and reapply them afterwards.
1231
1232If the changes are independent enough, Git will automatically complete
1233the merge and commit the result (or reuse an existing commit in case
1234of <<fast-forwards,fast-forward>>, see below). On the other hand,
1235if there are conflicts--for example, if the same file is
1236modified in two different ways in the remote branch and the local
1237branch--then you are warned; the output may look something like this:
1238
1239-------------------------------------------------
1240$ git merge next
1241 100% (4/4) done
1242Auto-merged file.txt
1243CONFLICT (content): Merge conflict in file.txt
1244Automatic merge failed; fix conflicts and then commit the result.
1245-------------------------------------------------
1246
1247Conflict markers are left in the problematic files, and after
1248you resolve the conflicts manually, you can update the index
1249with the contents and run Git commit, as you normally would when
1250creating a new file.
1251
1252If you examine the resulting commit using gitk, you will see that it
1253has two parents, one pointing to the top of the current branch, and
1254one to the top of the other branch.
1255
1256[[resolving-a-merge]]
1257Resolving a merge
1258-----------------
1259
1260When a merge isn't resolved automatically, Git leaves the index and
1261the working tree in a special state that gives you all the
1262information you need to help resolve the merge.
1263
1264Files with conflicts are marked specially in the index, so until you
1265resolve the problem and update the index, linkgit:git-commit[1] will
1266fail:
1267
1268-------------------------------------------------
1269$ git commit
1270file.txt: needs merge
1271-------------------------------------------------
1272
1273Also, linkgit:git-status[1] will list those files as "unmerged", and the
1274files with conflicts will have conflict markers added, like this:
1275
1276-------------------------------------------------
1277<<<<<<< HEAD:file.txt
1278Hello world
1279=======
1280Goodbye
1281>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1282-------------------------------------------------
1283
1284All you need to do is edit the files to resolve the conflicts, and then
1285
1286-------------------------------------------------
1287$ git add file.txt
1288$ git commit
1289-------------------------------------------------
1290
1291Note that the commit message will already be filled in for you with
1292some information about the merge.  Normally you can just use this
1293default message unchanged, but you may add additional commentary of
1294your own if desired.
1295
1296The above is all you need to know to resolve a simple merge.  But Git
1297also provides more information to help resolve conflicts:
1298
1299[[conflict-resolution]]
1300Getting conflict-resolution help during a merge
1301~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1302
1303All of the changes that Git was able to merge automatically are
1304already added to the index file, so linkgit:git-diff[1] shows only
1305the conflicts.  It uses an unusual syntax:
1306
1307-------------------------------------------------
1308$ git diff
1309diff --cc file.txt
1310index 802992c,2b60207..0000000
1311--- a/file.txt
1312+++ b/file.txt
1313@@@ -1,1 -1,1 +1,5 @@@
1314++<<<<<<< HEAD:file.txt
1315 +Hello world
1316++=======
1317+ Goodbye
1318++>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
1319-------------------------------------------------
1320
1321Recall that the commit which will be committed after we resolve this
1322conflict will have two parents instead of the usual one: one parent
1323will be HEAD, the tip of the current branch; the other will be the
1324tip of the other branch, which is stored temporarily in MERGE_HEAD.
1325
1326During the merge, the index holds three versions of each file.  Each of
1327these three "file stages" represents a different version of the file:
1328
1329-------------------------------------------------
1330$ git show :1:file.txt  # the file in a common ancestor of both branches
1331$ git show :2:file.txt  # the version from HEAD.
1332$ git show :3:file.txt  # the version from MERGE_HEAD.
1333-------------------------------------------------
1334
1335When you ask linkgit:git-diff[1] to show the conflicts, it runs a
1336three-way diff between the conflicted merge results in the work tree with
1337stages 2 and 3 to show only hunks whose contents come from both sides,
1338mixed (in other words, when a hunk's merge results come only from stage 2,
1339that part is not conflicting and is not shown.  Same for stage 3).
1340
1341The diff above shows the differences between the working-tree version of
1342file.txt and the stage 2 and stage 3 versions.  So instead of preceding
1343each line by a single `+` or `-`, it now uses two columns: the first
1344column is used for differences between the first parent and the working
1345directory copy, and the second for differences between the second parent
1346and the working directory copy.  (See the "COMBINED DIFF FORMAT" section
1347of linkgit:git-diff-files[1] for a details of the format.)
1348
1349After resolving the conflict in the obvious way (but before updating the
1350index), the diff will look like:
1351
1352-------------------------------------------------
1353$ git diff
1354diff --cc file.txt
1355index 802992c,2b60207..0000000
1356--- a/file.txt
1357+++ b/file.txt
1358@@@ -1,1 -1,1 +1,1 @@@
1359- Hello world
1360 -Goodbye
1361++Goodbye world
1362-------------------------------------------------
1363
1364This shows that our resolved version deleted "Hello world" from the
1365first parent, deleted "Goodbye" from the second parent, and added
1366"Goodbye world", which was previously absent from both.
1367
1368Some special diff options allow diffing the working directory against
1369any of these stages:
1370
1371-------------------------------------------------
1372$ git diff -1 file.txt          # diff against stage 1
1373$ git diff --base file.txt      # same as the above
1374$ git diff -2 file.txt          # diff against stage 2
1375$ git diff --ours file.txt      # same as the above
1376$ git diff -3 file.txt          # diff against stage 3
1377$ git diff --theirs file.txt    # same as the above.
1378-------------------------------------------------
1379
1380The linkgit:git-log[1] and linkgit:gitk[1] commands also provide special help
1381for merges:
1382
1383-------------------------------------------------
1384$ git log --merge
1385$ gitk --merge
1386-------------------------------------------------
1387
1388These will display all commits which exist only on HEAD or on
1389MERGE_HEAD, and which touch an unmerged file.
1390
1391You may also use linkgit:git-mergetool[1], which lets you merge the
1392unmerged files using external tools such as Emacs or kdiff3.
1393
1394Each time you resolve the conflicts in a file and update the index:
1395
1396-------------------------------------------------
1397$ git add file.txt
1398-------------------------------------------------
1399
1400the different stages of that file will be "collapsed", after which
1401`git diff` will (by default) no longer show diffs for that file.
1402
1403[[undoing-a-merge]]
1404Undoing a merge
1405---------------
1406
1407If you get stuck and decide to just give up and throw the whole mess
1408away, you can always return to the pre-merge state with
1409
1410-------------------------------------------------
1411$ git reset --hard HEAD
1412-------------------------------------------------
1413
1414Or, if you've already committed the merge that you want to throw away,
1415
1416-------------------------------------------------
1417$ git reset --hard ORIG_HEAD
1418-------------------------------------------------
1419
1420However, this last command can be dangerous in some cases--never
1421throw away a commit you have already committed if that commit may
1422itself have been merged into another branch, as doing so may confuse
1423further merges.
1424
1425[[fast-forwards]]
1426Fast-forward merges
1427-------------------
1428
1429There is one special case not mentioned above, which is treated
1430differently.  Normally, a merge results in a merge commit, with two
1431parents, one pointing at each of the two lines of development that
1432were merged.
1433
1434However, if the current branch is an ancestor of the other--so every commit
1435present in the current branch is already contained in the other branch--then Git
1436just performs a "fast-forward"; the head of the current branch is moved forward
1437to point at the head of the merged-in branch, without any new commits being
1438created.
1439
1440[[fixing-mistakes]]
1441Fixing mistakes
1442---------------
1443
1444If you've messed up the working tree, but haven't yet committed your
1445mistake, you can return the entire working tree to the last committed
1446state with
1447
1448-------------------------------------------------
1449$ git reset --hard HEAD
1450-------------------------------------------------
1451
1452If you make a commit that you later wish you hadn't, there are two
1453fundamentally different ways to fix the problem:
1454
1455        1. You can create a new commit that undoes whatever was done
1456        by the old commit.  This is the correct thing if your
1457        mistake has already been made public.
1458
1459        2. You can go back and modify the old commit.  You should
1460        never do this if you have already made the history public;
1461        Git does not normally expect the "history" of a project to
1462        change, and cannot correctly perform repeated merges from
1463        a branch that has had its history changed.
1464
1465[[reverting-a-commit]]
1466Fixing a mistake with a new commit
1467~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1468
1469Creating a new commit that reverts an earlier change is very easy;
1470just pass the linkgit:git-revert[1] command a reference to the bad
1471commit; for example, to revert the most recent commit:
1472
1473-------------------------------------------------
1474$ git revert HEAD
1475-------------------------------------------------
1476
1477This will create a new commit which undoes the change in HEAD.  You
1478will be given a chance to edit the commit message for the new commit.
1479
1480You can also revert an earlier change, for example, the next-to-last:
1481
1482-------------------------------------------------
1483$ git revert HEAD^
1484-------------------------------------------------
1485
1486In this case Git will attempt to undo the old change while leaving
1487intact any changes made since then.  If more recent changes overlap
1488with the changes to be reverted, then you will be asked to fix
1489conflicts manually, just as in the case of <<resolving-a-merge,
1490resolving a merge>>.
1491
1492[[fixing-a-mistake-by-rewriting-history]]
1493Fixing a mistake by rewriting history
1494~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1495
1496If the problematic commit is the most recent commit, and you have not
1497yet made that commit public, then you may just
1498<<undoing-a-merge,destroy it using `git reset`>>.
1499
1500Alternatively, you
1501can edit the working directory and update the index to fix your
1502mistake, just as if you were going to <<how-to-make-a-commit,create a
1503new commit>>, then run
1504
1505-------------------------------------------------
1506$ git commit --amend
1507-------------------------------------------------
1508
1509which will replace the old commit by a new commit incorporating your
1510changes, giving you a chance to edit the old commit message first.
1511
1512Again, you should never do this to a commit that may already have
1513been merged into another branch; use linkgit:git-revert[1] instead in
1514that case.
1515
1516It is also possible to replace commits further back in the history, but
1517this is an advanced topic to be left for
1518<<cleaning-up-history,another chapter>>.
1519
1520[[checkout-of-path]]
1521Checking out an old version of a file
1522~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1523
1524In the process of undoing a previous bad change, you may find it
1525useful to check out an older version of a particular file using
1526linkgit:git-checkout[1].  We've used `git checkout` before to switch
1527branches, but it has quite different behavior if it is given a path
1528name: the command
1529
1530-------------------------------------------------
1531$ git checkout HEAD^ path/to/file
1532-------------------------------------------------
1533
1534replaces path/to/file by the contents it had in the commit HEAD^, and
1535also updates the index to match.  It does not change branches.
1536
1537If you just want to look at an old version of the file, without
1538modifying the working directory, you can do that with
1539linkgit:git-show[1]:
1540
1541-------------------------------------------------
1542$ git show HEAD^:path/to/file
1543-------------------------------------------------
1544
1545which will display the given version of the file.
1546
1547[[interrupted-work]]
1548Temporarily setting aside work in progress
1549~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1550
1551While you are in the middle of working on something complicated, you
1552find an unrelated but obvious and trivial bug.  You would like to fix it
1553before continuing.  You can use linkgit:git-stash[1] to save the current
1554state of your work, and after fixing the bug (or, optionally after doing
1555so on a different branch and then coming back), unstash the
1556work-in-progress changes.
1557
1558------------------------------------------------
1559$ git stash save "work in progress for foo feature"
1560------------------------------------------------
1561
1562This command will save your changes away to the `stash`, and
1563reset your working tree and the index to match the tip of your
1564current branch.  Then you can make your fix as usual.
1565
1566------------------------------------------------
1567... edit and test ...
1568$ git commit -a -m "blorpl: typofix"
1569------------------------------------------------
1570
1571After that, you can go back to what you were working on with
1572`git stash pop`:
1573
1574------------------------------------------------
1575$ git stash pop
1576------------------------------------------------
1577
1578
1579[[ensuring-good-performance]]
1580Ensuring good performance
1581-------------------------
1582
1583On large repositories, Git depends on compression to keep the history
1584information from taking up too much space on disk or in memory.  Some
1585Git commands may automatically run linkgit:git-gc[1], so you don't
1586have to worry about running it manually.  However, compressing a large
1587repository may take a while, so you may want to call `gc` explicitly
1588to avoid automatic compression kicking in when it is not convenient.
1589
1590
1591[[ensuring-reliability]]
1592Ensuring reliability
1593--------------------
1594
1595[[checking-for-corruption]]
1596Checking the repository for corruption
1597~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1598
1599The linkgit:git-fsck[1] command runs a number of self-consistency checks
1600on the repository, and reports on any problems.  This may take some
1601time.
1602
1603-------------------------------------------------
1604$ git fsck
1605dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1606dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1607dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1608dangling blob 218761f9d90712d37a9c5e36f406f92202db07eb
1609dangling commit bf093535a34a4d35731aa2bd90fe6b176302f14f
1610dangling commit 8e4bec7f2ddaa268bef999853c25755452100f8e
1611dangling tree d50bb86186bf27b681d25af89d3b5b68382e4085
1612dangling tree b24c2473f1fd3d91352a624795be026d64c8841f
1613...
1614-------------------------------------------------
1615
1616You will see informational messages on dangling objects. They are objects
1617that still exist in the repository but are no longer referenced by any of
1618your branches, and can (and will) be removed after a while with `gc`.
1619You can run `git fsck --no-dangling` to suppress these messages, and still
1620view real errors.
1621
1622[[recovering-lost-changes]]
1623Recovering lost changes
1624~~~~~~~~~~~~~~~~~~~~~~~
1625
1626[[reflogs]]
1627Reflogs
1628^^^^^^^
1629
1630Say you modify a branch with <<fixing-mistakes,`git reset --hard`>>,
1631and then realize that the branch was the only reference you had to
1632that point in history.
1633
1634Fortunately, Git also keeps a log, called a "reflog", of all the
1635previous values of each branch.  So in this case you can still find the
1636old history using, for example,
1637
1638-------------------------------------------------
1639$ git log master@{1}
1640-------------------------------------------------
1641
1642This lists the commits reachable from the previous version of the
1643`master` branch head.  This syntax can be used with any Git command
1644that accepts a commit, not just with `git log`.  Some other examples:
1645
1646-------------------------------------------------
1647$ git show master@{2}           # See where the branch pointed 2,
1648$ git show master@{3}           # 3, ... changes ago.
1649$ gitk master@{yesterday}       # See where it pointed yesterday,
1650$ gitk master@{"1 week ago"}    # ... or last week
1651$ git log --walk-reflogs master # show reflog entries for master
1652-------------------------------------------------
1653
1654A separate reflog is kept for the HEAD, so
1655
1656-------------------------------------------------
1657$ git show HEAD@{"1 week ago"}
1658-------------------------------------------------
1659
1660will show what HEAD pointed to one week ago, not what the current branch
1661pointed to one week ago.  This allows you to see the history of what
1662you've checked out.
1663
1664The reflogs are kept by default for 30 days, after which they may be
1665pruned.  See linkgit:git-reflog[1] and linkgit:git-gc[1] to learn
1666how to control this pruning, and see the "SPECIFYING REVISIONS"
1667section of linkgit:gitrevisions[7] for details.
1668
1669Note that the reflog history is very different from normal Git history.
1670While normal history is shared by every repository that works on the
1671same project, the reflog history is not shared: it tells you only about
1672how the branches in your local repository have changed over time.
1673
1674[[dangling-object-recovery]]
1675Examining dangling objects
1676^^^^^^^^^^^^^^^^^^^^^^^^^^
1677
1678In some situations the reflog may not be able to save you.  For example,
1679suppose you delete a branch, then realize you need the history it
1680contained.  The reflog is also deleted; however, if you have not yet
1681pruned the repository, then you may still be able to find the lost
1682commits in the dangling objects that `git fsck` reports.  See
1683<<dangling-objects>> for the details.
1684
1685-------------------------------------------------
1686$ git fsck
1687dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
1688dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
1689dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
1690...
1691-------------------------------------------------
1692
1693You can examine
1694one of those dangling commits with, for example,
1695
1696------------------------------------------------
1697$ gitk 7281251ddd --not --all
1698------------------------------------------------
1699
1700which does what it sounds like: it says that you want to see the commit
1701history that is described by the dangling commit(s), but not the
1702history that is described by all your existing branches and tags.  Thus
1703you get exactly the history reachable from that commit that is lost.
1704(And notice that it might not be just one commit: we only report the
1705"tip of the line" as being dangling, but there might be a whole deep
1706and complex commit history that was dropped.)
1707
1708If you decide you want the history back, you can always create a new
1709reference pointing to it, for example, a new branch:
1710
1711------------------------------------------------
1712$ git branch recovered-branch 7281251ddd
1713------------------------------------------------
1714
1715Other types of dangling objects (blobs and trees) are also possible, and
1716dangling objects can arise in other situations.
1717
1718
1719[[sharing-development]]
1720Sharing development with others
1721===============================
1722
1723[[getting-updates-With-git-pull]]
1724Getting updates with git pull
1725-----------------------------
1726
1727After you clone a repository and commit a few changes of your own, you
1728may wish to check the original repository for updates and merge them
1729into your own work.
1730
1731We have already seen <<Updating-a-repository-With-git-fetch,how to
1732keep remote-tracking branches up to date>> with linkgit:git-fetch[1],
1733and how to merge two branches.  So you can merge in changes from the
1734original repository's master branch with:
1735
1736-------------------------------------------------
1737$ git fetch
1738$ git merge origin/master
1739-------------------------------------------------
1740
1741However, the linkgit:git-pull[1] command provides a way to do this in
1742one step:
1743
1744-------------------------------------------------
1745$ git pull origin master
1746-------------------------------------------------
1747
1748In fact, if you have `master` checked out, then this branch has been
1749configured by `git clone` to get changes from the HEAD branch of the
1750origin repository.  So often you can
1751accomplish the above with just a simple
1752
1753-------------------------------------------------
1754$ git pull
1755-------------------------------------------------
1756
1757This command will fetch changes from the remote branches to your
1758remote-tracking branches `origin/*`, and merge the default branch into
1759the current branch.
1760
1761More generally, a branch that is created from a remote-tracking branch
1762will pull
1763by default from that branch.  See the descriptions of the
1764`branch.<name>.remote` and `branch.<name>.merge` options in
1765linkgit:git-config[1], and the discussion of the `--track` option in
1766linkgit:git-checkout[1], to learn how to control these defaults.
1767
1768In addition to saving you keystrokes, `git pull` also helps you by
1769producing a default commit message documenting the branch and
1770repository that you pulled from.
1771
1772(But note that no such commit will be created in the case of a
1773<<fast-forwards,fast-forward>>; instead, your branch will just be
1774updated to point to the latest commit from the upstream branch.)
1775
1776The `git pull` command can also be given `.` as the "remote" repository,
1777in which case it just merges in a branch from the current repository; so
1778the commands
1779
1780-------------------------------------------------
1781$ git pull . branch
1782$ git merge branch
1783-------------------------------------------------
1784
1785are roughly equivalent.
1786
1787[[submitting-patches]]
1788Submitting patches to a project
1789-------------------------------
1790
1791If you just have a few changes, the simplest way to submit them may
1792just be to send them as patches in email:
1793
1794First, use linkgit:git-format-patch[1]; for example:
1795
1796-------------------------------------------------
1797$ git format-patch origin
1798-------------------------------------------------
1799
1800will produce a numbered series of files in the current directory, one
1801for each patch in the current branch but not in `origin/HEAD`.
1802
1803`git format-patch` can include an initial "cover letter". You can insert
1804commentary on individual patches after the three dash line which
1805`format-patch` places after the commit message but before the patch
1806itself.  If you use `git notes` to track your cover letter material,
1807`git format-patch --notes` will include the commit's notes in a similar
1808manner.
1809
1810You can then import these into your mail client and send them by
1811hand.  However, if you have a lot to send at once, you may prefer to
1812use the linkgit:git-send-email[1] script to automate the process.
1813Consult the mailing list for your project first to determine
1814their requirements for submitting patches.
1815
1816[[importing-patches]]
1817Importing patches to a project
1818------------------------------
1819
1820Git also provides a tool called linkgit:git-am[1] (am stands for
1821"apply mailbox"), for importing such an emailed series of patches.
1822Just save all of the patch-containing messages, in order, into a
1823single mailbox file, say `patches.mbox`, then run
1824
1825-------------------------------------------------
1826$ git am -3 patches.mbox
1827-------------------------------------------------
1828
1829Git will apply each patch in order; if any conflicts are found, it
1830will stop, and you can fix the conflicts as described in
1831"<<resolving-a-merge,Resolving a merge>>".  (The `-3` option tells
1832Git to perform a merge; if you would prefer it just to abort and
1833leave your tree and index untouched, you may omit that option.)
1834
1835Once the index is updated with the results of the conflict
1836resolution, instead of creating a new commit, just run
1837
1838-------------------------------------------------
1839$ git am --continue
1840-------------------------------------------------
1841
1842and Git will create the commit for you and continue applying the
1843remaining patches from the mailbox.
1844
1845The final result will be a series of commits, one for each patch in
1846the original mailbox, with authorship and commit log message each
1847taken from the message containing each patch.
1848
1849[[public-repositories]]
1850Public Git repositories
1851-----------------------
1852
1853Another way to submit changes to a project is to tell the maintainer
1854of that project to pull the changes from your repository using
1855linkgit:git-pull[1].  In the section "<<getting-updates-With-git-pull,
1856Getting updates with `git pull`>>" we described this as a way to get
1857updates from the "main" repository, but it works just as well in the
1858other direction.
1859
1860If you and the maintainer both have accounts on the same machine, then
1861you can just pull changes from each other's repositories directly;
1862commands that accept repository URLs as arguments will also accept a
1863local directory name:
1864
1865-------------------------------------------------
1866$ git clone /path/to/repository
1867$ git pull /path/to/other/repository
1868-------------------------------------------------
1869
1870or an ssh URL:
1871
1872-------------------------------------------------
1873$ git clone ssh://yourhost/~you/repository
1874-------------------------------------------------
1875
1876For projects with few developers, or for synchronizing a few private
1877repositories, this may be all you need.
1878
1879However, the more common way to do this is to maintain a separate public
1880repository (usually on a different host) for others to pull changes
1881from.  This is usually more convenient, and allows you to cleanly
1882separate private work in progress from publicly visible work.
1883
1884You will continue to do your day-to-day work in your personal
1885repository, but periodically "push" changes from your personal
1886repository into your public repository, allowing other developers to
1887pull from that repository.  So the flow of changes, in a situation
1888where there is one other developer with a public repository, looks
1889like this:
1890
1891                        you push
1892  your personal repo ------------------> your public repo
1893        ^                                     |
1894        |                                     |
1895        | you pull                            | they pull
1896        |                                     |
1897        |                                     |
1898        |               they push             V
1899  their public repo <------------------- their repo
1900
1901We explain how to do this in the following sections.
1902
1903[[setting-up-a-public-repository]]
1904Setting up a public repository
1905~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1906
1907Assume your personal repository is in the directory `~/proj`.  We
1908first create a new clone of the repository and tell `git daemon` that it
1909is meant to be public:
1910
1911-------------------------------------------------
1912$ git clone --bare ~/proj proj.git
1913$ touch proj.git/git-daemon-export-ok
1914-------------------------------------------------
1915
1916The resulting directory proj.git contains a "bare" git repository--it is
1917just the contents of the `.git` directory, without any files checked out
1918around it.
1919
1920Next, copy `proj.git` to the server where you plan to host the
1921public repository.  You can use scp, rsync, or whatever is most
1922convenient.
1923
1924[[exporting-via-git]]
1925Exporting a Git repository via the Git protocol
1926~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1927
1928This is the preferred method.
1929
1930If someone else administers the server, they should tell you what
1931directory to put the repository in, and what `git://` URL it will
1932appear at.  You can then skip to the section
1933"<<pushing-changes-to-a-public-repository,Pushing changes to a public
1934repository>>", below.
1935
1936Otherwise, all you need to do is start linkgit:git-daemon[1]; it will
1937listen on port 9418.  By default, it will allow access to any directory
1938that looks like a Git directory and contains the magic file
1939git-daemon-export-ok.  Passing some directory paths as `git daemon`
1940arguments will further restrict the exports to those paths.
1941
1942You can also run `git daemon` as an inetd service; see the
1943linkgit:git-daemon[1] man page for details.  (See especially the
1944examples section.)
1945
1946[[exporting-via-http]]
1947Exporting a git repository via HTTP
1948~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1949
1950The Git protocol gives better performance and reliability, but on a
1951host with a web server set up, HTTP exports may be simpler to set up.
1952
1953All you need to do is place the newly created bare Git repository in
1954a directory that is exported by the web server, and make some
1955adjustments to give web clients some extra information they need:
1956
1957-------------------------------------------------
1958$ mv proj.git /home/you/public_html/proj.git
1959$ cd proj.git
1960$ git --bare update-server-info
1961$ mv hooks/post-update.sample hooks/post-update
1962-------------------------------------------------
1963
1964(For an explanation of the last two lines, see
1965linkgit:git-update-server-info[1] and linkgit:githooks[5].)
1966
1967Advertise the URL of `proj.git`.  Anybody else should then be able to
1968clone or pull from that URL, for example with a command line like:
1969
1970-------------------------------------------------
1971$ git clone http://yourserver.com/~you/proj.git
1972-------------------------------------------------
1973
1974(See also
1975link:howto/setup-git-server-over-http.html[setup-git-server-over-http]
1976for a slightly more sophisticated setup using WebDAV which also
1977allows pushing over HTTP.)
1978
1979[[pushing-changes-to-a-public-repository]]
1980Pushing changes to a public repository
1981~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1982
1983Note that the two techniques outlined above (exporting via
1984<<exporting-via-http,http>> or <<exporting-via-git,git>>) allow other
1985maintainers to fetch your latest changes, but they do not allow write
1986access, which you will need to update the public repository with the
1987latest changes created in your private repository.
1988
1989The simplest way to do this is using linkgit:git-push[1] and ssh; to
1990update the remote branch named `master` with the latest state of your
1991branch named `master`, run
1992
1993-------------------------------------------------
1994$ git push ssh://yourserver.com/~you/proj.git master:master
1995-------------------------------------------------
1996
1997or just
1998
1999-------------------------------------------------
2000$ git push ssh://yourserver.com/~you/proj.git master
2001-------------------------------------------------
2002
2003As with `git fetch`, `git push` will complain if this does not result in a
2004<<fast-forwards,fast-forward>>; see the following section for details on
2005handling this case.
2006
2007Note that the target of a `push` is normally a
2008<<def_bare_repository,bare>> repository.  You can also push to a
2009repository that has a checked-out working tree, but a push to update the
2010currently checked-out branch is denied by default to prevent confusion.
2011See the description of the receive.denyCurrentBranch option
2012in linkgit:git-config[1] for details.
2013
2014As with `git fetch`, you may also set up configuration options to
2015save typing; so, for example:
2016
2017-------------------------------------------------
2018$ git remote add public-repo ssh://yourserver.com/~you/proj.git
2019-------------------------------------------------
2020
2021adds the following to `.git/config`:
2022
2023-------------------------------------------------
2024[remote "public-repo"]
2025        url = yourserver.com:proj.git
2026        fetch = +refs/heads/*:refs/remotes/example/*
2027-------------------------------------------------
2028
2029which lets you do the same push with just
2030
2031-------------------------------------------------
2032$ git push public-repo master
2033-------------------------------------------------
2034
2035See the explanations of the `remote.<name>.url`,
2036`branch.<name>.remote`, and `remote.<name>.push` options in
2037linkgit:git-config[1] for details.
2038
2039[[forcing-push]]
2040What to do when a push fails
2041~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2042
2043If a push would not result in a <<fast-forwards,fast-forward>> of the
2044remote branch, then it will fail with an error like:
2045
2046-------------------------------------------------
2047error: remote 'refs/heads/master' is not an ancestor of
2048 local  'refs/heads/master'.
2049 Maybe you are not up-to-date and need to pull first?
2050error: failed to push to 'ssh://yourserver.com/~you/proj.git'
2051-------------------------------------------------
2052
2053This can happen, for example, if you:
2054
2055        - use `git reset --hard` to remove already-published commits, or
2056        - use `git commit --amend` to replace already-published commits
2057          (as in <<fixing-a-mistake-by-rewriting-history>>), or
2058        - use `git rebase` to rebase any already-published commits (as
2059          in <<using-git-rebase>>).
2060
2061You may force `git push` to perform the update anyway by preceding the
2062branch name with a plus sign:
2063
2064-------------------------------------------------
2065$ git push ssh://yourserver.com/~you/proj.git +master
2066-------------------------------------------------
2067
2068Note the addition of the `+` sign.  Alternatively, you can use the
2069`-f` flag to force the remote update, as in:
2070
2071-------------------------------------------------
2072$ git push -f ssh://yourserver.com/~you/proj.git master
2073-------------------------------------------------
2074
2075Normally whenever a branch head in a public repository is modified, it
2076is modified to point to a descendant of the commit that it pointed to
2077before.  By forcing a push in this situation, you break that convention.
2078(See <<problems-With-rewriting-history>>.)
2079
2080Nevertheless, this is a common practice for people that need a simple
2081way to publish a work-in-progress patch series, and it is an acceptable
2082compromise as long as you warn other developers that this is how you
2083intend to manage the branch.
2084
2085It's also possible for a push to fail in this way when other people have
2086the right to push to the same repository.  In that case, the correct
2087solution is to retry the push after first updating your work: either by a
2088pull, or by a fetch followed by a rebase; see the
2089<<setting-up-a-shared-repository,next section>> and
2090linkgit:gitcvs-migration[7] for more.
2091
2092[[setting-up-a-shared-repository]]
2093Setting up a shared repository
2094~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2095
2096Another way to collaborate is by using a model similar to that
2097commonly used in CVS, where several developers with special rights
2098all push to and pull from a single shared repository.  See
2099linkgit:gitcvs-migration[7] for instructions on how to
2100set this up.
2101
2102However, while there is nothing wrong with Git's support for shared
2103repositories, this mode of operation is not generally recommended,
2104simply because the mode of collaboration that Git supports--by
2105exchanging patches and pulling from public repositories--has so many
2106advantages over the central shared repository:
2107
2108        - Git's ability to quickly import and merge patches allows a
2109          single maintainer to process incoming changes even at very
2110          high rates.  And when that becomes too much, `git pull` provides
2111          an easy way for that maintainer to delegate this job to other
2112          maintainers while still allowing optional review of incoming
2113          changes.
2114        - Since every developer's repository has the same complete copy
2115          of the project history, no repository is special, and it is
2116          trivial for another developer to take over maintenance of a
2117          project, either by mutual agreement, or because a maintainer
2118          becomes unresponsive or difficult to work with.
2119        - The lack of a central group of "committers" means there is
2120          less need for formal decisions about who is "in" and who is
2121          "out".
2122
2123[[setting-up-gitweb]]
2124Allowing web browsing of a repository
2125~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2126
2127The gitweb cgi script provides users an easy way to browse your
2128project's revisions, file contents and logs without having to install
2129Git. Features like RSS/Atom feeds and blame/annotation details may
2130optionally be enabled.
2131
2132The linkgit:git-instaweb[1] command provides a simple way to start
2133browsing the repository using gitweb. The default server when using
2134instaweb is lighttpd.
2135
2136See the file gitweb/INSTALL in the Git source tree and
2137linkgit:gitweb[1] for instructions on details setting up a permanent
2138installation with a CGI or Perl capable server.
2139
2140[[how-to-get-a-git-repository-with-minimal-history]]
2141How to get a Git repository with minimal history
2142------------------------------------------------
2143
2144A <<def_shallow_clone,shallow clone>>, with its truncated
2145history, is useful when one is interested only in recent history
2146of a project and getting full history from the upstream is
2147expensive.
2148
2149A <<def_shallow_clone,shallow clone>> is created by specifying
2150the linkgit:git-clone[1] `--depth` switch. The depth can later be
2151changed with the linkgit:git-fetch[1] `--depth` switch, or full
2152history restored with `--unshallow`.
2153
2154Merging inside a <<def_shallow_clone,shallow clone>> will work as long
2155as a merge base is in the recent history.
2156Otherwise, it will be like merging unrelated histories and may
2157have to result in huge conflicts.  This limitation may make such
2158a repository unsuitable to be used in merge based workflows.
2159
2160[[sharing-development-examples]]
2161Examples
2162--------
2163
2164[[maintaining-topic-branches]]
2165Maintaining topic branches for a Linux subsystem maintainer
2166~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2167
2168This describes how Tony Luck uses Git in his role as maintainer of the
2169IA64 architecture for the Linux kernel.
2170
2171He uses two public branches:
2172
2173 - A "test" tree into which patches are initially placed so that they
2174   can get some exposure when integrated with other ongoing development.
2175   This tree is available to Andrew for pulling into -mm whenever he
2176   wants.
2177
2178 - A "release" tree into which tested patches are moved for final sanity
2179   checking, and as a vehicle to send them upstream to Linus (by sending
2180   him a "please pull" request.)
2181
2182He also uses a set of temporary branches ("topic branches"), each
2183containing a logical grouping of patches.
2184
2185To set this up, first create your work tree by cloning Linus's public
2186tree:
2187
2188-------------------------------------------------
2189$ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git work
2190$ cd work
2191-------------------------------------------------
2192
2193Linus's tree will be stored in the remote-tracking branch named origin/master,
2194and can be updated using linkgit:git-fetch[1]; you can track other
2195public trees using linkgit:git-remote[1] to set up a "remote" and
2196linkgit:git-fetch[1] to keep them up-to-date; see
2197<<repositories-and-branches>>.
2198
2199Now create the branches in which you are going to work; these start out
2200at the current tip of origin/master branch, and should be set up (using
2201the `--track` option to linkgit:git-branch[1]) to merge changes in from
2202Linus by default.
2203
2204-------------------------------------------------
2205$ git branch --track test origin/master
2206$ git branch --track release origin/master
2207-------------------------------------------------
2208
2209These can be easily kept up to date using linkgit:git-pull[1].
2210
2211-------------------------------------------------
2212$ git checkout test && git pull
2213$ git checkout release && git pull
2214-------------------------------------------------
2215
2216Important note!  If you have any local changes in these branches, then
2217this merge will create a commit object in the history (with no local
2218changes Git will simply do a "fast-forward" merge).  Many people dislike
2219the "noise" that this creates in the Linux history, so you should avoid
2220doing this capriciously in the `release` branch, as these noisy commits
2221will become part of the permanent history when you ask Linus to pull
2222from the release branch.
2223
2224A few configuration variables (see linkgit:git-config[1]) can
2225make it easy to push both branches to your public tree.  (See
2226<<setting-up-a-public-repository>>.)
2227
2228-------------------------------------------------
2229$ cat >> .git/config <<EOF
2230[remote "mytree"]
2231        url =  master.kernel.org:/pub/scm/linux/kernel/git/aegl/linux.git
2232        push = release
2233        push = test
2234EOF
2235-------------------------------------------------
2236
2237Then you can push both the test and release trees using
2238linkgit:git-push[1]:
2239
2240-------------------------------------------------
2241$ git push mytree
2242-------------------------------------------------
2243
2244or push just one of the test and release branches using:
2245
2246-------------------------------------------------
2247$ git push mytree test
2248-------------------------------------------------
2249
2250or
2251
2252-------------------------------------------------
2253$ git push mytree release
2254-------------------------------------------------
2255
2256Now to apply some patches from the community.  Think of a short
2257snappy name for a branch to hold this patch (or related group of
2258patches), and create a new branch from a recent stable tag of
2259Linus's branch. Picking a stable base for your branch will:
22601) help you: by avoiding inclusion of unrelated and perhaps lightly
2261tested changes
22622) help future bug hunters that use `git bisect` to find problems
2263
2264-------------------------------------------------
2265$ git checkout -b speed-up-spinlocks v2.6.35
2266-------------------------------------------------
2267
2268Now you apply the patch(es), run some tests, and commit the change(s).  If
2269the patch is a multi-part series, then you should apply each as a separate
2270commit to this branch.
2271
2272-------------------------------------------------
2273$ ... patch ... test  ... commit [ ... patch ... test ... commit ]*
2274-------------------------------------------------
2275
2276When you are happy with the state of this change, you can merge it into the
2277"test" branch in preparation to make it public:
2278
2279-------------------------------------------------
2280$ git checkout test && git merge speed-up-spinlocks
2281-------------------------------------------------
2282
2283It is unlikely that you would have any conflicts here ... but you might if you
2284spent a while on this step and had also pulled new versions from upstream.
2285
2286Sometime later when enough time has passed and testing done, you can pull the
2287same branch into the `release` tree ready to go upstream.  This is where you
2288see the value of keeping each patch (or patch series) in its own branch.  It
2289means that the patches can be moved into the `release` tree in any order.
2290
2291-------------------------------------------------
2292$ git checkout release && git merge speed-up-spinlocks
2293-------------------------------------------------
2294
2295After a while, you will have a number of branches, and despite the
2296well chosen names you picked for each of them, you may forget what
2297they are for, or what status they are in.  To get a reminder of what
2298changes are in a specific branch, use:
2299
2300-------------------------------------------------
2301$ git log linux..branchname | git shortlog
2302-------------------------------------------------
2303
2304To see whether it has already been merged into the test or release branches,
2305use:
2306
2307-------------------------------------------------
2308$ git log test..branchname
2309-------------------------------------------------
2310
2311or
2312
2313-------------------------------------------------
2314$ git log release..branchname
2315-------------------------------------------------
2316
2317(If this branch has not yet been merged, you will see some log entries.
2318If it has been merged, then there will be no output.)
2319
2320Once a patch completes the great cycle (moving from test to release,
2321then pulled by Linus, and finally coming back into your local
2322`origin/master` branch), the branch for this change is no longer needed.
2323You detect this when the output from:
2324
2325-------------------------------------------------
2326$ git log origin..branchname
2327-------------------------------------------------
2328
2329is empty.  At this point the branch can be deleted:
2330
2331-------------------------------------------------
2332$ git branch -d branchname
2333-------------------------------------------------
2334
2335Some changes are so trivial that it is not necessary to create a separate
2336branch and then merge into each of the test and release branches.  For
2337these changes, just apply directly to the `release` branch, and then
2338merge that into the `test` branch.
2339
2340After pushing your work to `mytree`, you can use
2341linkgit:git-request-pull[1] to prepare a "please pull" request message
2342to send to Linus:
2343
2344-------------------------------------------------
2345$ git push mytree
2346$ git request-pull origin mytree release
2347-------------------------------------------------
2348
2349Here are some of the scripts that simplify all this even further.
2350
2351-------------------------------------------------
2352==== update script ====
2353# Update a branch in my Git tree.  If the branch to be updated
2354# is origin, then pull from kernel.org.  Otherwise merge
2355# origin/master branch into test|release branch
2356
2357case "$1" in
2358test|release)
2359        git checkout $1 && git pull . origin
2360        ;;
2361origin)
2362        before=$(git rev-parse refs/remotes/origin/master)
2363        git fetch origin
2364        after=$(git rev-parse refs/remotes/origin/master)
2365        if [ $before != $after ]
2366        then
2367                git log $before..$after | git shortlog
2368        fi
2369        ;;
2370*)
2371        echo "usage: $0 origin|test|release" 1>&2
2372        exit 1
2373        ;;
2374esac
2375-------------------------------------------------
2376
2377-------------------------------------------------
2378==== merge script ====
2379# Merge a branch into either the test or release branch
2380
2381pname=$0
2382
2383usage()
2384{
2385        echo "usage: $pname branch test|release" 1>&2
2386        exit 1
2387}
2388
2389git show-ref -q --verify -- refs/heads/"$1" || {
2390        echo "Can't see branch <$1>" 1>&2
2391        usage
2392}
2393
2394case "$2" in
2395test|release)
2396        if [ $(git log $2..$1 | wc -c) -eq 0 ]
2397        then
2398                echo $1 already merged into $2 1>&2
2399                exit 1
2400        fi
2401        git checkout $2 && git pull . $1
2402        ;;
2403*)
2404        usage
2405        ;;
2406esac
2407-------------------------------------------------
2408
2409-------------------------------------------------
2410==== status script ====
2411# report on status of my ia64 Git tree
2412
2413gb=$(tput setab 2)
2414rb=$(tput setab 1)
2415restore=$(tput setab 9)
2416
2417if [ `git rev-list test..release | wc -c` -gt 0 ]
2418then
2419        echo $rb Warning: commits in release that are not in test $restore
2420        git log test..release
2421fi
2422
2423for branch in `git show-ref --heads | sed 's|^.*/||'`
2424do
2425        if [ $branch = test -o $branch = release ]
2426        then
2427                continue
2428        fi
2429
2430        echo -n $gb ======= $branch ====== $restore " "
2431        status=
2432        for ref in test release origin/master
2433        do
2434                if [ `git rev-list $ref..$branch | wc -c` -gt 0 ]
2435                then
2436                        status=$status${ref:0:1}
2437                fi
2438        done
2439        case $status in
2440        trl)
2441                echo $rb Need to pull into test $restore
2442                ;;
2443        rl)
2444                echo "In test"
2445                ;;
2446        l)
2447                echo "Waiting for linus"
2448                ;;
2449        "")
2450                echo $rb All done $restore
2451                ;;
2452        *)
2453                echo $rb "<$status>" $restore
2454                ;;
2455        esac
2456        git log origin/master..$branch | git shortlog
2457done
2458-------------------------------------------------
2459
2460
2461[[cleaning-up-history]]
2462Rewriting history and maintaining patch series
2463==============================================
2464
2465Normally commits are only added to a project, never taken away or
2466replaced.  Git is designed with this assumption, and violating it will
2467cause Git's merge machinery (for example) to do the wrong thing.
2468
2469However, there is a situation in which it can be useful to violate this
2470assumption.
2471
2472[[patch-series]]
2473Creating the perfect patch series
2474---------------------------------
2475
2476Suppose you are a contributor to a large project, and you want to add a
2477complicated feature, and to present it to the other developers in a way
2478that makes it easy for them to read your changes, verify that they are
2479correct, and understand why you made each change.
2480
2481If you present all of your changes as a single patch (or commit), they
2482may find that it is too much to digest all at once.
2483
2484If you present them with the entire history of your work, complete with
2485mistakes, corrections, and dead ends, they may be overwhelmed.
2486
2487So the ideal is usually to produce a series of patches such that:
2488
2489        1. Each patch can be applied in order.
2490
2491        2. Each patch includes a single logical change, together with a
2492           message explaining the change.
2493
2494        3. No patch introduces a regression: after applying any initial
2495           part of the series, the resulting project still compiles and
2496           works, and has no bugs that it didn't have before.
2497
2498        4. The complete series produces the same end result as your own
2499           (probably much messier!) development process did.
2500
2501We will introduce some tools that can help you do this, explain how to
2502use them, and then explain some of the problems that can arise because
2503you are rewriting history.
2504
2505[[using-git-rebase]]
2506Keeping a patch series up to date using git rebase
2507--------------------------------------------------
2508
2509Suppose that you create a branch `mywork` on a remote-tracking branch
2510`origin`, and create some commits on top of it:
2511
2512-------------------------------------------------
2513$ git checkout -b mywork origin
2514$ vi file.txt
2515$ git commit
2516$ vi otherfile.txt
2517$ git commit
2518...
2519-------------------------------------------------
2520
2521You have performed no merges into mywork, so it is just a simple linear
2522sequence of patches on top of `origin`:
2523
2524................................................
2525 o--o--O <-- origin
2526        \
2527         a--b--c <-- mywork
2528................................................
2529
2530Some more interesting work has been done in the upstream project, and
2531`origin` has advanced:
2532
2533................................................
2534 o--o--O--o--o--o <-- origin
2535        \
2536         a--b--c <-- mywork
2537................................................
2538
2539At this point, you could use `pull` to merge your changes back in;
2540the result would create a new merge commit, like this:
2541
2542................................................
2543 o--o--O--o--o--o <-- origin
2544        \        \
2545         a--b--c--m <-- mywork
2546................................................
2547
2548However, if you prefer to keep the history in mywork a simple series of
2549commits without any merges, you may instead choose to use
2550linkgit:git-rebase[1]:
2551
2552-------------------------------------------------
2553$ git checkout mywork
2554$ git rebase origin
2555-------------------------------------------------
2556
2557This will remove each of your commits from mywork, temporarily saving
2558them as patches (in a directory named `.git/rebase-apply`), update mywork to
2559point at the latest version of origin, then apply each of the saved
2560patches to the new mywork.  The result will look like:
2561
2562
2563................................................
2564 o--o--O--o--o--o <-- origin
2565                 \
2566                  a'--b'--c' <-- mywork
2567................................................
2568
2569In the process, it may discover conflicts.  In that case it will stop
2570and allow you to fix the conflicts; after fixing conflicts, use `git add`
2571to update the index with those contents, and then, instead of
2572running `git commit`, just run
2573
2574-------------------------------------------------
2575$ git rebase --continue
2576-------------------------------------------------
2577
2578and Git will continue applying the rest of the patches.
2579
2580At any point you may use the `--abort` option to abort this process and
2581return mywork to the state it had before you started the rebase:
2582
2583-------------------------------------------------
2584$ git rebase --abort
2585-------------------------------------------------
2586
2587If you need to reorder or edit a number of commits in a branch, it may
2588be easier to use `git rebase -i`, which allows you to reorder and
2589squash commits, as well as marking them for individual editing during
2590the rebase.  See <<interactive-rebase>> for details, and
2591<<reordering-patch-series>> for alternatives.
2592
2593[[rewriting-one-commit]]
2594Rewriting a single commit
2595-------------------------
2596
2597We saw in <<fixing-a-mistake-by-rewriting-history>> that you can replace the
2598most recent commit using
2599
2600-------------------------------------------------
2601$ git commit --amend
2602-------------------------------------------------
2603
2604which will replace the old commit by a new commit incorporating your
2605changes, giving you a chance to edit the old commit message first.
2606This is useful for fixing typos in your last commit, or for adjusting
2607the patch contents of a poorly staged commit.
2608
2609If you need to amend commits from deeper in your history, you can
2610use <<interactive-rebase,interactive rebase's `edit` instruction>>.
2611
2612[[reordering-patch-series]]
2613Reordering or selecting from a patch series
2614-------------------------------------------
2615
2616Sometimes you want to edit a commit deeper in your history.  One
2617approach is to use `git format-patch` to create a series of patches
2618and then reset the state to before the patches:
2619
2620-------------------------------------------------
2621$ git format-patch origin
2622$ git reset --hard origin
2623-------------------------------------------------
2624
2625Then modify, reorder, or eliminate patches as needed before applying
2626them again with linkgit:git-am[1]:
2627
2628-------------------------------------------------
2629$ git am *.patch
2630-------------------------------------------------
2631
2632[[interactive-rebase]]
2633Using interactive rebases
2634-------------------------
2635
2636You can also edit a patch series with an interactive rebase.  This is
2637the same as <<reordering-patch-series,reordering a patch series using
2638`format-patch`>>, so use whichever interface you like best.
2639
2640Rebase your current HEAD on the last commit you want to retain as-is.
2641For example, if you want to reorder the last 5 commits, use:
2642
2643-------------------------------------------------
2644$ git rebase -i HEAD~5
2645-------------------------------------------------
2646
2647This will open your editor with a list of steps to be taken to perform
2648your rebase.
2649
2650-------------------------------------------------
2651pick deadbee The oneline of this commit
2652pick fa1afe1 The oneline of the next commit
2653...
2654
2655# Rebase c0ffeee..deadbee onto c0ffeee
2656#
2657# Commands:
2658#  p, pick = use commit
2659#  r, reword = use commit, but edit the commit message
2660#  e, edit = use commit, but stop for amending
2661#  s, squash = use commit, but meld into previous commit
2662#  f, fixup = like "squash", but discard this commit's log message
2663#  x, exec = run command (the rest of the line) using shell
2664#
2665# These lines can be re-ordered; they are executed from top to bottom.
2666#
2667# If you remove a line here THAT COMMIT WILL BE LOST.
2668#
2669# However, if you remove everything, the rebase will be aborted.
2670#
2671# Note that empty commits are commented out
2672-------------------------------------------------
2673
2674As explained in the comments, you can reorder commits, squash them
2675together, edit commit messages, etc. by editing the list.  Once you
2676are satisfied, save the list and close your editor, and the rebase
2677will begin.
2678
2679The rebase will stop where `pick` has been replaced with `edit` or
2680when a step in the list fails to mechanically resolve conflicts and
2681needs your help.  When you are done editing and/or resolving conflicts
2682you can continue with `git rebase --continue`.  If you decide that
2683things are getting too hairy, you can always bail out with `git rebase
2684--abort`.  Even after the rebase is complete, you can still recover
2685the original branch by using the <<reflogs,reflog>>.
2686
2687For a more detailed discussion of the procedure and additional tips,
2688see the "INTERACTIVE MODE" section of linkgit:git-rebase[1].
2689
2690[[patch-series-tools]]
2691Other tools
2692-----------
2693
2694There are numerous other tools, such as StGit, which exist for the
2695purpose of maintaining a patch series.  These are outside of the scope of
2696this manual.
2697
2698[[problems-With-rewriting-history]]
2699Problems with rewriting history
2700-------------------------------
2701
2702The primary problem with rewriting the history of a branch has to do
2703with merging.  Suppose somebody fetches your branch and merges it into
2704their branch, with a result something like this:
2705
2706................................................
2707 o--o--O--o--o--o <-- origin
2708        \        \
2709         t--t--t--m <-- their branch:
2710................................................
2711
2712Then suppose you modify the last three commits:
2713
2714................................................
2715         o--o--o <-- new head of origin
2716        /
2717 o--o--O--o--o--o <-- old head of origin
2718................................................
2719
2720If we examined all this history together in one repository, it will
2721look like:
2722
2723................................................
2724         o--o--o <-- new head of origin
2725        /
2726 o--o--O--o--o--o <-- old head of origin
2727        \        \
2728         t--t--t--m <-- their branch:
2729................................................
2730
2731Git has no way of knowing that the new head is an updated version of
2732the old head; it treats this situation exactly the same as it would if
2733two developers had independently done the work on the old and new heads
2734in parallel.  At this point, if someone attempts to merge the new head
2735in to their branch, Git will attempt to merge together the two (old and
2736new) lines of development, instead of trying to replace the old by the
2737new.  The results are likely to be unexpected.
2738
2739You may still choose to publish branches whose history is rewritten,
2740and it may be useful for others to be able to fetch those branches in
2741order to examine or test them, but they should not attempt to pull such
2742branches into their own work.
2743
2744For true distributed development that supports proper merging,
2745published branches should never be rewritten.
2746
2747[[bisect-merges]]
2748Why bisecting merge commits can be harder than bisecting linear history
2749-----------------------------------------------------------------------
2750
2751The linkgit:git-bisect[1] command correctly handles history that
2752includes merge commits.  However, when the commit that it finds is a
2753merge commit, the user may need to work harder than usual to figure out
2754why that commit introduced a problem.
2755
2756Imagine this history:
2757
2758................................................
2759      ---Z---o---X---...---o---A---C---D
2760          \                       /
2761           o---o---Y---...---o---B
2762................................................
2763
2764Suppose that on the upper line of development, the meaning of one
2765of the functions that exists at Z is changed at commit X.  The
2766commits from Z leading to A change both the function's
2767implementation and all calling sites that exist at Z, as well
2768as new calling sites they add, to be consistent.  There is no
2769bug at A.
2770
2771Suppose that in the meantime on the lower line of development somebody
2772adds a new calling site for that function at commit Y.  The
2773commits from Z leading to B all assume the old semantics of that
2774function and the callers and the callee are consistent with each
2775other.  There is no bug at B, either.
2776
2777Suppose further that the two development lines merge cleanly at C,
2778so no conflict resolution is required.
2779
2780Nevertheless, the code at C is broken, because the callers added
2781on the lower line of development have not been converted to the new
2782semantics introduced on the upper line of development.  So if all
2783you know is that D is bad, that Z is good, and that
2784linkgit:git-bisect[1] identifies C as the culprit, how will you
2785figure out that the problem is due to this change in semantics?
2786
2787When the result of a `git bisect` is a non-merge commit, you should
2788normally be able to discover the problem by examining just that commit.
2789Developers can make this easy by breaking their changes into small
2790self-contained commits.  That won't help in the case above, however,
2791because the problem isn't obvious from examination of any single
2792commit; instead, a global view of the development is required.  To
2793make matters worse, the change in semantics in the problematic
2794function may be just one small part of the changes in the upper
2795line of development.
2796
2797On the other hand, if instead of merging at C you had rebased the
2798history between Z to B on top of A, you would have gotten this
2799linear history:
2800
2801................................................................
2802    ---Z---o---X--...---o---A---o---o---Y*--...---o---B*--D*
2803................................................................
2804
2805Bisecting between Z and D* would hit a single culprit commit Y*,
2806and understanding why Y* was broken would probably be easier.
2807
2808Partly for this reason, many experienced Git users, even when
2809working on an otherwise merge-heavy project, keep the history
2810linear by rebasing against the latest upstream version before
2811publishing.
2812
2813[[advanced-branch-management]]
2814Advanced branch management
2815==========================
2816
2817[[fetching-individual-branches]]
2818Fetching individual branches
2819----------------------------
2820
2821Instead of using linkgit:git-remote[1], you can also choose just
2822to update one branch at a time, and to store it locally under an
2823arbitrary name:
2824
2825-------------------------------------------------
2826$ git fetch origin todo:my-todo-work
2827-------------------------------------------------
2828
2829The first argument, `origin`, just tells Git to fetch from the
2830repository you originally cloned from.  The second argument tells Git
2831to fetch the branch named `todo` from the remote repository, and to
2832store it locally under the name `refs/heads/my-todo-work`.
2833
2834You can also fetch branches from other repositories; so
2835
2836-------------------------------------------------
2837$ git fetch git://example.com/proj.git master:example-master
2838-------------------------------------------------
2839
2840will create a new branch named `example-master` and store in it the
2841branch named `master` from the repository at the given URL.  If you
2842already have a branch named example-master, it will attempt to
2843<<fast-forwards,fast-forward>> to the commit given by example.com's
2844master branch.  In more detail:
2845
2846[[fetch-fast-forwards]]
2847git fetch and fast-forwards
2848---------------------------
2849
2850In the previous example, when updating an existing branch, `git fetch`
2851checks to make sure that the most recent commit on the remote
2852branch is a descendant of the most recent commit on your copy of the
2853branch before updating your copy of the branch to point at the new
2854commit.  Git calls this process a <<fast-forwards,fast-forward>>.
2855
2856A fast-forward looks something like this:
2857
2858................................................
2859 o--o--o--o <-- old head of the branch
2860           \
2861            o--o--o <-- new head of the branch
2862................................................
2863
2864
2865In some cases it is possible that the new head will *not* actually be
2866a descendant of the old head.  For example, the developer may have
2867realized she made a serious mistake, and decided to backtrack,
2868resulting in a situation like:
2869
2870................................................
2871 o--o--o--o--a--b <-- old head of the branch
2872           \
2873            o--o--o <-- new head of the branch
2874................................................
2875
2876In this case, `git fetch` will fail, and print out a warning.
2877
2878In that case, you can still force Git to update to the new head, as
2879described in the following section.  However, note that in the
2880situation above this may mean losing the commits labeled `a` and `b`,
2881unless you've already created a reference of your own pointing to
2882them.
2883
2884[[forcing-fetch]]
2885Forcing git fetch to do non-fast-forward updates
2886------------------------------------------------
2887
2888If git fetch fails because the new head of a branch is not a
2889descendant of the old head, you may force the update with:
2890
2891-------------------------------------------------
2892$ git fetch git://example.com/proj.git +master:refs/remotes/example/master
2893-------------------------------------------------
2894
2895Note the addition of the `+` sign.  Alternatively, you can use the `-f`
2896flag to force updates of all the fetched branches, as in:
2897
2898-------------------------------------------------
2899$ git fetch -f origin
2900-------------------------------------------------
2901
2902Be aware that commits that the old version of example/master pointed at
2903may be lost, as we saw in the previous section.
2904
2905[[remote-branch-configuration]]
2906Configuring remote-tracking branches
2907------------------------------------
2908
2909We saw above that `origin` is just a shortcut to refer to the
2910repository that you originally cloned from.  This information is
2911stored in Git configuration variables, which you can see using
2912linkgit:git-config[1]:
2913
2914-------------------------------------------------
2915$ git config -l
2916core.repositoryformatversion=0
2917core.filemode=true
2918core.logallrefupdates=true
2919remote.origin.url=git://git.kernel.org/pub/scm/git/git.git
2920remote.origin.fetch=+refs/heads/*:refs/remotes/origin/*
2921branch.master.remote=origin
2922branch.master.merge=refs/heads/master
2923-------------------------------------------------
2924
2925If there are other repositories that you also use frequently, you can
2926create similar configuration options to save typing; for example,
2927
2928-------------------------------------------------
2929$ git remote add example git://example.com/proj.git
2930-------------------------------------------------
2931
2932adds the following to `.git/config`:
2933
2934-------------------------------------------------
2935[remote "example"]
2936        url = git://example.com/proj.git
2937        fetch = +refs/heads/*:refs/remotes/example/*
2938-------------------------------------------------
2939
2940Also note that the above configuration can be performed by directly
2941editing the file `.git/config` instead of using linkgit:git-remote[1].
2942
2943After configuring the remote, the following three commands will do the
2944same thing:
2945
2946-------------------------------------------------
2947$ git fetch git://example.com/proj.git +refs/heads/*:refs/remotes/example/*
2948$ git fetch example +refs/heads/*:refs/remotes/example/*
2949$ git fetch example
2950-------------------------------------------------
2951
2952See linkgit:git-config[1] for more details on the configuration
2953options mentioned above and linkgit:git-fetch[1] for more details on
2954the refspec syntax.
2955
2956
2957[[git-concepts]]
2958Git concepts
2959============
2960
2961Git is built on a small number of simple but powerful ideas.  While it
2962is possible to get things done without understanding them, you will find
2963Git much more intuitive if you do.
2964
2965We start with the most important, the  <<def_object_database,object
2966database>> and the <<def_index,index>>.
2967
2968[[the-object-database]]
2969The Object Database
2970-------------------
2971
2972
2973We already saw in <<understanding-commits>> that all commits are stored
2974under a 40-digit "object name".  In fact, all the information needed to
2975represent the history of a project is stored in objects with such names.
2976In each case the name is calculated by taking the SHA-1 hash of the
2977contents of the object.  The SHA-1 hash is a cryptographic hash function.
2978What that means to us is that it is impossible to find two different
2979objects with the same name.  This has a number of advantages; among
2980others:
2981
2982- Git can quickly determine whether two objects are identical or not,
2983  just by comparing names.
2984- Since object names are computed the same way in every repository, the
2985  same content stored in two repositories will always be stored under
2986  the same name.
2987- Git can detect errors when it reads an object, by checking that the
2988  object's name is still the SHA-1 hash of its contents.
2989
2990(See <<object-details>> for the details of the object formatting and
2991SHA-1 calculation.)
2992
2993There are four different types of objects: "blob", "tree", "commit", and
2994"tag".
2995
2996- A <<def_blob_object,"blob" object>> is used to store file data.
2997- A <<def_tree_object,"tree" object>> ties one or more
2998  "blob" objects into a directory structure. In addition, a tree object
2999  can refer to other tree objects, thus creating a directory hierarchy.
3000- A <<def_commit_object,"commit" object>> ties such directory hierarchies
3001  together into a <<def_DAG,directed acyclic graph>> of revisions--each
3002  commit contains the object name of exactly one tree designating the
3003  directory hierarchy at the time of the commit. In addition, a commit
3004  refers to "parent" commit objects that describe the history of how we
3005  arrived at that directory hierarchy.
3006- A <<def_tag_object,"tag" object>> symbolically identifies and can be
3007  used to sign other objects. It contains the object name and type of
3008  another object, a symbolic name (of course!) and, optionally, a
3009  signature.
3010
3011The object types in some more detail:
3012
3013[[commit-object]]
3014Commit Object
3015~~~~~~~~~~~~~
3016
3017The "commit" object links a physical state of a tree with a description
3018of how we got there and why.  Use the `--pretty=raw` option to
3019linkgit:git-show[1] or linkgit:git-log[1] to examine your favorite
3020commit:
3021
3022------------------------------------------------
3023$ git show -s --pretty=raw 2be7fcb476
3024commit 2be7fcb4764f2dbcee52635b91fedb1b3dcf7ab4
3025tree fb3a8bdd0ceddd019615af4d57a53f43d8cee2bf
3026parent 257a84d9d02e90447b149af58b271c19405edb6a
3027author Dave Watson <dwatson@mimvista.com> 1187576872 -0400
3028committer Junio C Hamano <gitster@pobox.com> 1187591163 -0700
3029
3030    Fix misspelling of 'suppress' in docs
3031
3032    Signed-off-by: Junio C Hamano <gitster@pobox.com>
3033------------------------------------------------
3034
3035As you can see, a commit is defined by:
3036
3037- a tree: The SHA-1 name of a tree object (as defined below), representing
3038  the contents of a directory at a certain point in time.
3039- parent(s): The SHA-1 name(s) of some number of commits which represent the
3040  immediately previous step(s) in the history of the project.  The
3041  example above has one parent; merge commits may have more than
3042  one.  A commit with no parents is called a "root" commit, and
3043  represents the initial revision of a project.  Each project must have
3044  at least one root.  A project can also have multiple roots, though
3045  that isn't common (or necessarily a good idea).
3046- an author: The name of the person responsible for this change, together
3047  with its date.
3048- a committer: The name of the person who actually created the commit,
3049  with the date it was done.  This may be different from the author, for
3050  example, if the author was someone who wrote a patch and emailed it
3051  to the person who used it to create the commit.
3052- a comment describing this commit.
3053
3054Note that a commit does not itself contain any information about what
3055actually changed; all changes are calculated by comparing the contents
3056of the tree referred to by this commit with the trees associated with
3057its parents.  In particular, Git does not attempt to record file renames
3058explicitly, though it can identify cases where the existence of the same
3059file data at changing paths suggests a rename.  (See, for example, the
3060`-M` option to linkgit:git-diff[1]).
3061
3062A commit is usually created by linkgit:git-commit[1], which creates a
3063commit whose parent is normally the current HEAD, and whose tree is
3064taken from the content currently stored in the index.
3065
3066[[tree-object]]
3067Tree Object
3068~~~~~~~~~~~
3069
3070The ever-versatile linkgit:git-show[1] command can also be used to
3071examine tree objects, but linkgit:git-ls-tree[1] will give you more
3072details:
3073
3074------------------------------------------------
3075$ git ls-tree fb3a8bdd0ce
3076100644 blob 63c918c667fa005ff12ad89437f2fdc80926e21c    .gitignore
3077100644 blob 5529b198e8d14decbe4ad99db3f7fb632de0439d    .mailmap
3078100644 blob 6ff87c4664981e4397625791c8ea3bbb5f2279a3    COPYING
3079040000 tree 2fb783e477100ce076f6bf57e4a6f026013dc745    Documentation
3080100755 blob 3c0032cec592a765692234f1cba47dfdcc3a9200    GIT-VERSION-GEN
3081100644 blob 289b046a443c0647624607d471289b2c7dcd470b    INSTALL
3082100644 blob 4eb463797adc693dc168b926b6932ff53f17d0b1    Makefile
3083100644 blob 548142c327a6790ff8821d67c2ee1eff7a656b52    README
3084...
3085------------------------------------------------
3086
3087As you can see, a tree object contains a list of entries, each with a
3088mode, object type, SHA-1 name, and name, sorted by name.  It represents
3089the contents of a single directory tree.
3090
3091The object type may be a blob, representing the contents of a file, or
3092another tree, representing the contents of a subdirectory.  Since trees
3093and blobs, like all other objects, are named by the SHA-1 hash of their
3094contents, two trees have the same SHA-1 name if and only if their
3095contents (including, recursively, the contents of all subdirectories)
3096are identical.  This allows Git to quickly determine the differences
3097between two related tree objects, since it can ignore any entries with
3098identical object names.
3099
3100(Note: in the presence of submodules, trees may also have commits as
3101entries.  See <<submodules>> for documentation.)
3102
3103Note that the files all have mode 644 or 755: Git actually only pays
3104attention to the executable bit.
3105
3106[[blob-object]]
3107Blob Object
3108~~~~~~~~~~~
3109
3110You can use linkgit:git-show[1] to examine the contents of a blob; take,
3111for example, the blob in the entry for `COPYING` from the tree above:
3112
3113------------------------------------------------
3114$ git show 6ff87c4664
3115
3116 Note that the only valid version of the GPL as far as this project
3117 is concerned is _this_ particular version of the license (ie v2, not
3118 v2.2 or v3.x or whatever), unless explicitly otherwise stated.
3119...
3120------------------------------------------------
3121
3122A "blob" object is nothing but a binary blob of data.  It doesn't refer
3123to anything else or have attributes of any kind.
3124
3125Since the blob is entirely defined by its data, if two files in a
3126directory tree (or in multiple different versions of the repository)
3127have the same contents, they will share the same blob object. The object
3128is totally independent of its location in the directory tree, and
3129renaming a file does not change the object that file is associated with.
3130
3131Note that any tree or blob object can be examined using
3132linkgit:git-show[1] with the <revision>:<path> syntax.  This can
3133sometimes be useful for browsing the contents of a tree that is not
3134currently checked out.
3135
3136[[trust]]
3137Trust
3138~~~~~
3139
3140If you receive the SHA-1 name of a blob from one source, and its contents
3141from another (possibly untrusted) source, you can still trust that those
3142contents are correct as long as the SHA-1 name agrees.  This is because
3143the SHA-1 is designed so that it is infeasible to find different contents
3144that produce the same hash.
3145
3146Similarly, you need only trust the SHA-1 name of a top-level tree object
3147to trust the contents of the entire directory that it refers to, and if
3148you receive the SHA-1 name of a commit from a trusted source, then you
3149can easily verify the entire history of commits reachable through
3150parents of that commit, and all of those contents of the trees referred
3151to by those commits.
3152
3153So to introduce some real trust in the system, the only thing you need
3154to do is to digitally sign just 'one' special note, which includes the
3155name of a top-level commit.  Your digital signature shows others
3156that you trust that commit, and the immutability of the history of
3157commits tells others that they can trust the whole history.
3158
3159In other words, you can easily validate a whole archive by just
3160sending out a single email that tells the people the name (SHA-1 hash)
3161of the top commit, and digitally sign that email using something
3162like GPG/PGP.
3163
3164To assist in this, Git also provides the tag object...
3165
3166[[tag-object]]
3167Tag Object
3168~~~~~~~~~~
3169
3170A tag object contains an object, object type, tag name, the name of the
3171person ("tagger") who created the tag, and a message, which may contain
3172a signature, as can be seen using linkgit:git-cat-file[1]:
3173
3174------------------------------------------------
3175$ git cat-file tag v1.5.0
3176object 437b1b20df4b356c9342dac8d38849f24ef44f27
3177type commit
3178tag v1.5.0
3179tagger Junio C Hamano <junkio@cox.net> 1171411200 +0000
3180
3181GIT 1.5.0
3182-----BEGIN PGP SIGNATURE-----
3183Version: GnuPG v1.4.6 (GNU/Linux)
3184
3185iD8DBQBF0lGqwMbZpPMRm5oRAuRiAJ9ohBLd7s2kqjkKlq1qqC57SbnmzQCdG4ui
3186nLE/L9aUXdWeTFPron96DLA=
3187=2E+0
3188-----END PGP SIGNATURE-----
3189------------------------------------------------
3190
3191See the linkgit:git-tag[1] command to learn how to create and verify tag
3192objects.  (Note that linkgit:git-tag[1] can also be used to create
3193"lightweight tags", which are not tag objects at all, but just simple
3194references whose names begin with `refs/tags/`).
3195
3196[[pack-files]]
3197How Git stores objects efficiently: pack files
3198~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3199
3200Newly created objects are initially created in a file named after the
3201object's SHA-1 hash (stored in `.git/objects`).
3202
3203Unfortunately this system becomes inefficient once a project has a
3204lot of objects.  Try this on an old project:
3205
3206------------------------------------------------
3207$ git count-objects
32086930 objects, 47620 kilobytes
3209------------------------------------------------
3210
3211The first number is the number of objects which are kept in
3212individual files.  The second is the amount of space taken up by
3213those "loose" objects.
3214
3215You can save space and make Git faster by moving these loose objects in
3216to a "pack file", which stores a group of objects in an efficient
3217compressed format; the details of how pack files are formatted can be
3218found in link:technical/pack-format.html[pack format].
3219
3220To put the loose objects into a pack, just run git repack:
3221
3222------------------------------------------------
3223$ git repack
3224Counting objects: 6020, done.
3225Delta compression using up to 4 threads.
3226Compressing objects: 100% (6020/6020), done.
3227Writing objects: 100% (6020/6020), done.
3228Total 6020 (delta 4070), reused 0 (delta 0)
3229------------------------------------------------
3230
3231This creates a single "pack file" in .git/objects/pack/
3232containing all currently unpacked objects.  You can then run
3233
3234------------------------------------------------
3235$ git prune
3236------------------------------------------------
3237
3238to remove any of the "loose" objects that are now contained in the
3239pack.  This will also remove any unreferenced objects (which may be
3240created when, for example, you use `git reset` to remove a commit).
3241You can verify that the loose objects are gone by looking at the
3242`.git/objects` directory or by running
3243
3244------------------------------------------------
3245$ git count-objects
32460 objects, 0 kilobytes
3247------------------------------------------------
3248
3249Although the object files are gone, any commands that refer to those
3250objects will work exactly as they did before.
3251
3252The linkgit:git-gc[1] command performs packing, pruning, and more for
3253you, so is normally the only high-level command you need.
3254
3255[[dangling-objects]]
3256Dangling objects
3257~~~~~~~~~~~~~~~~
3258
3259The linkgit:git-fsck[1] command will sometimes complain about dangling
3260objects.  They are not a problem.
3261
3262The most common cause of dangling objects is that you've rebased a
3263branch, or you have pulled from somebody else who rebased a branch--see
3264<<cleaning-up-history>>.  In that case, the old head of the original
3265branch still exists, as does everything it pointed to. The branch
3266pointer itself just doesn't, since you replaced it with another one.
3267
3268There are also other situations that cause dangling objects. For
3269example, a "dangling blob" may arise because you did a `git add` of a
3270file, but then, before you actually committed it and made it part of the
3271bigger picture, you changed something else in that file and committed
3272that *updated* thing--the old state that you added originally ends up
3273not being pointed to by any commit or tree, so it's now a dangling blob
3274object.
3275
3276Similarly, when the "recursive" merge strategy runs, and finds that
3277there are criss-cross merges and thus more than one merge base (which is
3278fairly unusual, but it does happen), it will generate one temporary
3279midway tree (or possibly even more, if you had lots of criss-crossing
3280merges and more than two merge bases) as a temporary internal merge
3281base, and again, those are real objects, but the end result will not end
3282up pointing to them, so they end up "dangling" in your repository.
3283
3284Generally, dangling objects aren't anything to worry about. They can
3285even be very useful: if you screw something up, the dangling objects can
3286be how you recover your old tree (say, you did a rebase, and realized
3287that you really didn't want to--you can look at what dangling objects
3288you have, and decide to reset your head to some old dangling state).
3289
3290For commits, you can just use:
3291
3292------------------------------------------------
3293$ gitk <dangling-commit-sha-goes-here> --not --all
3294------------------------------------------------
3295
3296This asks for all the history reachable from the given commit but not
3297from any branch, tag, or other reference.  If you decide it's something
3298you want, you can always create a new reference to it, e.g.,
3299
3300------------------------------------------------
3301$ git branch recovered-branch <dangling-commit-sha-goes-here>
3302------------------------------------------------
3303
3304For blobs and trees, you can't do the same, but you can still examine
3305them.  You can just do
3306
3307------------------------------------------------
3308$ git show <dangling-blob/tree-sha-goes-here>
3309------------------------------------------------
3310
3311to show what the contents of the blob were (or, for a tree, basically
3312what the `ls` for that directory was), and that may give you some idea
3313of what the operation was that left that dangling object.
3314
3315Usually, dangling blobs and trees aren't very interesting. They're
3316almost always the result of either being a half-way mergebase (the blob
3317will often even have the conflict markers from a merge in it, if you
3318have had conflicting merges that you fixed up by hand), or simply
3319because you interrupted a `git fetch` with ^C or something like that,
3320leaving _some_ of the new objects in the object database, but just
3321dangling and useless.
3322
3323Anyway, once you are sure that you're not interested in any dangling
3324state, you can just prune all unreachable objects:
3325
3326------------------------------------------------
3327$ git prune
3328------------------------------------------------
3329
3330and they'll be gone. (You should only run `git prune` on a quiescent
3331repository--it's kind of like doing a filesystem fsck recovery: you
3332don't want to do that while the filesystem is mounted.
3333`git prune` is designed not to cause any harm in such cases of concurrent
3334accesses to a repository but you might receive confusing or scary messages.)
3335
3336[[recovering-from-repository-corruption]]
3337Recovering from repository corruption
3338~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3339
3340By design, Git treats data trusted to it with caution.  However, even in
3341the absence of bugs in Git itself, it is still possible that hardware or
3342operating system errors could corrupt data.
3343
3344The first defense against such problems is backups.  You can back up a
3345Git directory using clone, or just using cp, tar, or any other backup
3346mechanism.
3347
3348As a last resort, you can search for the corrupted objects and attempt
3349to replace them by hand.  Back up your repository before attempting this
3350in case you corrupt things even more in the process.
3351
3352We'll assume that the problem is a single missing or corrupted blob,
3353which is sometimes a solvable problem.  (Recovering missing trees and
3354especially commits is *much* harder).
3355
3356Before starting, verify that there is corruption, and figure out where
3357it is with linkgit:git-fsck[1]; this may be time-consuming.
3358
3359Assume the output looks like this:
3360
3361------------------------------------------------
3362$ git fsck --full --no-dangling
3363broken link from    tree 2d9263c6d23595e7cb2a21e5ebbb53655278dff8
3364              to    blob 4b9458b3786228369c63936db65827de3cc06200
3365missing blob 4b9458b3786228369c63936db65827de3cc06200
3366------------------------------------------------
3367
3368Now you know that blob 4b9458b3 is missing, and that the tree 2d9263c6
3369points to it.  If you could find just one copy of that missing blob
3370object, possibly in some other repository, you could move it into
3371`.git/objects/4b/9458b3...` and be done.  Suppose you can't.  You can
3372still examine the tree that pointed to it with linkgit:git-ls-tree[1],
3373which might output something like:
3374
3375------------------------------------------------
3376$ git ls-tree 2d9263c6d23595e7cb2a21e5ebbb53655278dff8
3377100644 blob 8d14531846b95bfa3564b58ccfb7913a034323b8    .gitignore
3378100644 blob ebf9bf84da0aab5ed944264a5db2a65fe3a3e883    .mailmap
3379100644 blob ca442d313d86dc67e0a2e5d584b465bd382cbf5c    COPYING
3380...
3381100644 blob 4b9458b3786228369c63936db65827de3cc06200    myfile
3382...
3383------------------------------------------------
3384
3385So now you know that the missing blob was the data for a file named
3386`myfile`.  And chances are you can also identify the directory--let's
3387say it's in `somedirectory`.  If you're lucky the missing copy might be
3388the same as the copy you have checked out in your working tree at
3389`somedirectory/myfile`; you can test whether that's right with
3390linkgit:git-hash-object[1]:
3391
3392------------------------------------------------
3393$ git hash-object -w somedirectory/myfile
3394------------------------------------------------
3395
3396which will create and store a blob object with the contents of
3397somedirectory/myfile, and output the SHA-1 of that object.  if you're
3398extremely lucky it might be 4b9458b3786228369c63936db65827de3cc06200, in
3399which case you've guessed right, and the corruption is fixed!
3400
3401Otherwise, you need more information.  How do you tell which version of
3402the file has been lost?
3403
3404The easiest way to do this is with:
3405
3406------------------------------------------------
3407$ git log --raw --all --full-history -- somedirectory/myfile
3408------------------------------------------------
3409
3410Because you're asking for raw output, you'll now get something like
3411
3412------------------------------------------------
3413commit abc
3414Author:
3415Date:
3416...
3417:100644 100644 4b9458b... newsha... M somedirectory/myfile
3418
3419
3420commit xyz
3421Author:
3422Date:
3423
3424...
3425:100644 100644 oldsha... 4b9458b... M somedirectory/myfile
3426------------------------------------------------
3427
3428This tells you that the immediately following version of the file was
3429"newsha", and that the immediately preceding version was "oldsha".
3430You also know the commit messages that went with the change from oldsha
3431to 4b9458b and with the change from 4b9458b to newsha.
3432
3433If you've been committing small enough changes, you may now have a good
3434shot at reconstructing the contents of the in-between state 4b9458b.
3435
3436If you can do that, you can now recreate the missing object with
3437
3438------------------------------------------------
3439$ git hash-object -w <recreated-file>
3440------------------------------------------------
3441
3442and your repository is good again!
3443
3444(Btw, you could have ignored the `fsck`, and started with doing a
3445
3446------------------------------------------------
3447$ git log --raw --all
3448------------------------------------------------
3449
3450and just looked for the sha of the missing object (4b9458b..) in that
3451whole thing. It's up to you--Git does *have* a lot of information, it is
3452just missing one particular blob version.
3453
3454[[the-index]]
3455The index
3456---------
3457
3458The index is a binary file (generally kept in `.git/index`) containing a
3459sorted list of path names, each with permissions and the SHA-1 of a blob
3460object; linkgit:git-ls-files[1] can show you the contents of the index:
3461
3462-------------------------------------------------
3463$ git ls-files --stage
3464100644 63c918c667fa005ff12ad89437f2fdc80926e21c 0       .gitignore
3465100644 5529b198e8d14decbe4ad99db3f7fb632de0439d 0       .mailmap
3466100644 6ff87c4664981e4397625791c8ea3bbb5f2279a3 0       COPYING
3467100644 a37b2152bd26be2c2289e1f57a292534a51a93c7 0       Documentation/.gitignore
3468100644 fbefe9a45b00a54b58d94d06eca48b03d40a50e0 0       Documentation/Makefile
3469...
3470100644 2511aef8d89ab52be5ec6a5e46236b4b6bcd07ea 0       xdiff/xtypes.h
3471100644 2ade97b2574a9f77e7ae4002a4e07a6a38e46d07 0       xdiff/xutils.c
3472100644 d5de8292e05e7c36c4b68857c1cf9855e3d2f70a 0       xdiff/xutils.h
3473-------------------------------------------------
3474
3475Note that in older documentation you may see the index called the
3476"current directory cache" or just the "cache".  It has three important
3477properties:
3478
34791. The index contains all the information necessary to generate a single
3480(uniquely determined) tree object.
3481+
3482For example, running linkgit:git-commit[1] generates this tree object
3483from the index, stores it in the object database, and uses it as the
3484tree object associated with the new commit.
3485
34862. The index enables fast comparisons between the tree object it defines
3487and the working tree.
3488+
3489It does this by storing some additional data for each entry (such as
3490the last modified time).  This data is not displayed above, and is not
3491stored in the created tree object, but it can be used to determine
3492quickly which files in the working directory differ from what was
3493stored in the index, and thus save Git from having to read all of the
3494data from such files to look for changes.
3495
34963. It can efficiently represent information about merge conflicts
3497between different tree objects, allowing each pathname to be
3498associated with sufficient information about the trees involved that
3499you can create a three-way merge between them.
3500+
3501We saw in <<conflict-resolution>> that during a merge the index can
3502store multiple versions of a single file (called "stages").  The third
3503column in the linkgit:git-ls-files[1] output above is the stage
3504number, and will take on values other than 0 for files with merge
3505conflicts.
3506
3507The index is thus a sort of temporary staging area, which is filled with
3508a tree which you are in the process of working on.
3509
3510If you blow the index away entirely, you generally haven't lost any
3511information as long as you have the name of the tree that it described.
3512
3513[[submodules]]
3514Submodules
3515==========
3516
3517Large projects are often composed of smaller, self-contained modules.  For
3518example, an embedded Linux distribution's source tree would include every
3519piece of software in the distribution with some local modifications; a movie
3520player might need to build against a specific, known-working version of a
3521decompression library; several independent programs might all share the same
3522build scripts.
3523
3524With centralized revision control systems this is often accomplished by
3525including every module in one single repository.  Developers can check out
3526all modules or only the modules they need to work with.  They can even modify
3527files across several modules in a single commit while moving things around
3528or updating APIs and translations.
3529
3530Git does not allow partial checkouts, so duplicating this approach in Git
3531would force developers to keep a local copy of modules they are not
3532interested in touching.  Commits in an enormous checkout would be slower
3533than you'd expect as Git would have to scan every directory for changes.
3534If modules have a lot of local history, clones would take forever.
3535
3536On the plus side, distributed revision control systems can much better
3537integrate with external sources.  In a centralized model, a single arbitrary
3538snapshot of the external project is exported from its own revision control
3539and then imported into the local revision control on a vendor branch.  All
3540the history is hidden.  With distributed revision control you can clone the
3541entire external history and much more easily follow development and re-merge
3542local changes.
3543
3544Git's submodule support allows a repository to contain, as a subdirectory, a
3545checkout of an external project.  Submodules maintain their own identity;
3546the submodule support just stores the submodule repository location and
3547commit ID, so other developers who clone the containing project
3548("superproject") can easily clone all the submodules at the same revision.
3549Partial checkouts of the superproject are possible: you can tell Git to
3550clone none, some or all of the submodules.
3551
3552The linkgit:git-submodule[1] command is available since Git 1.5.3.  Users
3553with Git 1.5.2 can look up the submodule commits in the repository and
3554manually check them out; earlier versions won't recognize the submodules at
3555all.
3556
3557To see how submodule support works, create four example
3558repositories that can be used later as a submodule:
3559
3560-------------------------------------------------
3561$ mkdir ~/git
3562$ cd ~/git
3563$ for i in a b c d
3564do
3565        mkdir $i
3566        cd $i
3567        git init
3568        echo "module $i" > $i.txt
3569        git add $i.txt
3570        git commit -m "Initial commit, submodule $i"
3571        cd ..
3572done
3573-------------------------------------------------
3574
3575Now create the superproject and add all the submodules:
3576
3577-------------------------------------------------
3578$ mkdir super
3579$ cd super
3580$ git init
3581$ for i in a b c d
3582do
3583        git submodule add ~/git/$i $i
3584done
3585-------------------------------------------------
3586
3587NOTE: Do not use local URLs here if you plan to publish your superproject!
3588
3589See what files `git submodule` created:
3590
3591-------------------------------------------------
3592$ ls -a
3593.  ..  .git  .gitmodules  a  b  c  d
3594-------------------------------------------------
3595
3596The `git submodule add <repo> <path>` command does a couple of things:
3597
3598- It clones the submodule from `<repo>` to the given `<path>` under the
3599  current directory and by default checks out the master branch.
3600- It adds the submodule's clone path to the linkgit:gitmodules[5] file and
3601  adds this file to the index, ready to be committed.
3602- It adds the submodule's current commit ID to the index, ready to be
3603  committed.
3604
3605Commit the superproject:
3606
3607-------------------------------------------------
3608$ git commit -m "Add submodules a, b, c and d."
3609-------------------------------------------------
3610
3611Now clone the superproject:
3612
3613-------------------------------------------------
3614$ cd ..
3615$ git clone super cloned
3616$ cd cloned
3617-------------------------------------------------
3618
3619The submodule directories are there, but they're empty:
3620
3621-------------------------------------------------
3622$ ls -a a
3623.  ..
3624$ git submodule status
3625-d266b9873ad50488163457f025db7cdd9683d88b a
3626-e81d457da15309b4fef4249aba9b50187999670d b
3627-c1536a972b9affea0f16e0680ba87332dc059146 c
3628-d96249ff5d57de5de093e6baff9e0aafa5276a74 d
3629-------------------------------------------------
3630
3631NOTE: The commit object names shown above would be different for you, but they
3632should match the HEAD commit object names of your repositories.  You can check
3633it by running `git ls-remote ../a`.
3634
3635Pulling down the submodules is a two-step process. First run `git submodule
3636init` to add the submodule repository URLs to `.git/config`:
3637
3638-------------------------------------------------
3639$ git submodule init
3640-------------------------------------------------
3641
3642Now use `git submodule update` to clone the repositories and check out the
3643commits specified in the superproject:
3644
3645-------------------------------------------------
3646$ git submodule update
3647$ cd a
3648$ ls -a
3649.  ..  .git  a.txt
3650-------------------------------------------------
3651
3652One major difference between `git submodule update` and `git submodule add` is
3653that `git submodule update` checks out a specific commit, rather than the tip
3654of a branch. It's like checking out a tag: the head is detached, so you're not
3655working on a branch.
3656
3657-------------------------------------------------
3658$ git branch
3659* (detached from d266b98)
3660  master
3661-------------------------------------------------
3662
3663If you want to make a change within a submodule and you have a detached head,
3664then you should create or checkout a branch, make your changes, publish the
3665change within the submodule, and then update the superproject to reference the
3666new commit:
3667
3668-------------------------------------------------
3669$ git checkout master
3670-------------------------------------------------
3671
3672or
3673
3674-------------------------------------------------
3675$ git checkout -b fix-up
3676-------------------------------------------------
3677
3678then
3679
3680-------------------------------------------------
3681$ echo "adding a line again" >> a.txt
3682$ git commit -a -m "Updated the submodule from within the superproject."
3683$ git push
3684$ cd ..
3685$ git diff
3686diff --git a/a b/a
3687index d266b98..261dfac 160000
3688--- a/a
3689+++ b/a
3690@@ -1 +1 @@
3691-Subproject commit d266b9873ad50488163457f025db7cdd9683d88b
3692+Subproject commit 261dfac35cb99d380eb966e102c1197139f7fa24
3693$ git add a
3694$ git commit -m "Updated submodule a."
3695$ git push
3696-------------------------------------------------
3697
3698You have to run `git submodule update` after `git pull` if you want to update
3699submodules, too.
3700
3701Pitfalls with submodules
3702------------------------
3703
3704Always publish the submodule change before publishing the change to the
3705superproject that references it. If you forget to publish the submodule change,
3706others won't be able to clone the repository:
3707
3708-------------------------------------------------
3709$ cd ~/git/super/a
3710$ echo i added another line to this file >> a.txt
3711$ git commit -a -m "doing it wrong this time"
3712$ cd ..
3713$ git add a
3714$ git commit -m "Updated submodule a again."
3715$ git push
3716$ cd ~/git/cloned
3717$ git pull
3718$ git submodule update
3719error: pathspec '261dfac35cb99d380eb966e102c1197139f7fa24' did not match any file(s) known to git.
3720Did you forget to 'git add'?
3721Unable to checkout '261dfac35cb99d380eb966e102c1197139f7fa24' in submodule path 'a'
3722-------------------------------------------------
3723
3724In older Git versions it could be easily forgotten to commit new or modified
3725files in a submodule, which silently leads to similar problems as not pushing
3726the submodule changes. Starting with Git 1.7.0 both `git status` and `git diff`
3727in the superproject show submodules as modified when they contain new or
3728modified files to protect against accidentally committing such a state. `git
3729diff` will also add a `-dirty` to the work tree side when generating patch
3730output or used with the `--submodule` option:
3731
3732-------------------------------------------------
3733$ git diff
3734diff --git a/sub b/sub
3735--- a/sub
3736+++ b/sub
3737@@ -1 +1 @@
3738-Subproject commit 3f356705649b5d566d97ff843cf193359229a453
3739+Subproject commit 3f356705649b5d566d97ff843cf193359229a453-dirty
3740$ git diff --submodule
3741Submodule sub 3f35670..3f35670-dirty:
3742-------------------------------------------------
3743
3744You also should not rewind branches in a submodule beyond commits that were
3745ever recorded in any superproject.
3746
3747It's not safe to run `git submodule update` if you've made and committed
3748changes within a submodule without checking out a branch first. They will be
3749silently overwritten:
3750
3751-------------------------------------------------
3752$ cat a.txt
3753module a
3754$ echo line added from private2 >> a.txt
3755$ git commit -a -m "line added inside private2"
3756$ cd ..
3757$ git submodule update
3758Submodule path 'a': checked out 'd266b9873ad50488163457f025db7cdd9683d88b'
3759$ cd a
3760$ cat a.txt
3761module a
3762-------------------------------------------------
3763
3764NOTE: The changes are still visible in the submodule's reflog.
3765
3766If you have uncommitted changes in your submodule working tree, `git
3767submodule update` will not overwrite them.  Instead, you get the usual
3768warning about not being able switch from a dirty branch.
3769
3770[[low-level-operations]]
3771Low-level Git operations
3772========================
3773
3774Many of the higher-level commands were originally implemented as shell
3775scripts using a smaller core of low-level Git commands.  These can still
3776be useful when doing unusual things with Git, or just as a way to
3777understand its inner workings.
3778
3779[[object-manipulation]]
3780Object access and manipulation
3781------------------------------
3782
3783The linkgit:git-cat-file[1] command can show the contents of any object,
3784though the higher-level linkgit:git-show[1] is usually more useful.
3785
3786The linkgit:git-commit-tree[1] command allows constructing commits with
3787arbitrary parents and trees.
3788
3789A tree can be created with linkgit:git-write-tree[1] and its data can be
3790accessed by linkgit:git-ls-tree[1].  Two trees can be compared with
3791linkgit:git-diff-tree[1].
3792
3793A tag is created with linkgit:git-mktag[1], and the signature can be
3794verified by linkgit:git-verify-tag[1], though it is normally simpler to
3795use linkgit:git-tag[1] for both.
3796
3797[[the-workflow]]
3798The Workflow
3799------------
3800
3801High-level operations such as linkgit:git-commit[1],
3802linkgit:git-checkout[1] and linkgit:git-reset[1] work by moving data
3803between the working tree, the index, and the object database.  Git
3804provides low-level operations which perform each of these steps
3805individually.
3806
3807Generally, all Git operations work on the index file. Some operations
3808work *purely* on the index file (showing the current state of the
3809index), but most operations move data between the index file and either
3810the database or the working directory. Thus there are four main
3811combinations:
3812
3813[[working-directory-to-index]]
3814working directory -> index
3815~~~~~~~~~~~~~~~~~~~~~~~~~~
3816
3817The linkgit:git-update-index[1] command updates the index with
3818information from the working directory.  You generally update the
3819index information by just specifying the filename you want to update,
3820like so:
3821
3822-------------------------------------------------
3823$ git update-index filename
3824-------------------------------------------------
3825
3826but to avoid common mistakes with filename globbing etc., the command
3827will not normally add totally new entries or remove old entries,
3828i.e. it will normally just update existing cache entries.
3829
3830To tell Git that yes, you really do realize that certain files no
3831longer exist, or that new files should be added, you
3832should use the `--remove` and `--add` flags respectively.
3833
3834NOTE! A `--remove` flag does 'not' mean that subsequent filenames will
3835necessarily be removed: if the files still exist in your directory
3836structure, the index will be updated with their new status, not
3837removed. The only thing `--remove` means is that update-index will be
3838considering a removed file to be a valid thing, and if the file really
3839does not exist any more, it will update the index accordingly.
3840
3841As a special case, you can also do `git update-index --refresh`, which
3842will refresh the "stat" information of each index to match the current
3843stat information. It will 'not' update the object status itself, and
3844it will only update the fields that are used to quickly test whether
3845an object still matches its old backing store object.
3846
3847The previously introduced linkgit:git-add[1] is just a wrapper for
3848linkgit:git-update-index[1].
3849
3850[[index-to-object-database]]
3851index -> object database
3852~~~~~~~~~~~~~~~~~~~~~~~~
3853
3854You write your current index file to a "tree" object with the program
3855
3856-------------------------------------------------
3857$ git write-tree
3858-------------------------------------------------
3859
3860that doesn't come with any options--it will just write out the
3861current index into the set of tree objects that describe that state,
3862and it will return the name of the resulting top-level tree. You can
3863use that tree to re-generate the index at any time by going in the
3864other direction:
3865
3866[[object-database-to-index]]
3867object database -> index
3868~~~~~~~~~~~~~~~~~~~~~~~~
3869
3870You read a "tree" file from the object database, and use that to
3871populate (and overwrite--don't do this if your index contains any
3872unsaved state that you might want to restore later!) your current
3873index.  Normal operation is just
3874
3875-------------------------------------------------
3876$ git read-tree <SHA-1 of tree>
3877-------------------------------------------------
3878
3879and your index file will now be equivalent to the tree that you saved
3880earlier. However, that is only your 'index' file: your working
3881directory contents have not been modified.
3882
3883[[index-to-working-directory]]
3884index -> working directory
3885~~~~~~~~~~~~~~~~~~~~~~~~~~
3886
3887You update your working directory from the index by "checking out"
3888files. This is not a very common operation, since normally you'd just
3889keep your files updated, and rather than write to your working
3890directory, you'd tell the index files about the changes in your
3891working directory (i.e. `git update-index`).
3892
3893However, if you decide to jump to a new version, or check out somebody
3894else's version, or just restore a previous tree, you'd populate your
3895index file with read-tree, and then you need to check out the result
3896with
3897
3898-------------------------------------------------
3899$ git checkout-index filename
3900-------------------------------------------------
3901
3902or, if you want to check out all of the index, use `-a`.
3903
3904NOTE! `git checkout-index` normally refuses to overwrite old files, so
3905if you have an old version of the tree already checked out, you will
3906need to use the `-f` flag ('before' the `-a` flag or the filename) to
3907'force' the checkout.
3908
3909
3910Finally, there are a few odds and ends which are not purely moving
3911from one representation to the other:
3912
3913[[tying-it-all-together]]
3914Tying it all together
3915~~~~~~~~~~~~~~~~~~~~~
3916
3917To commit a tree you have instantiated with `git write-tree`, you'd
3918create a "commit" object that refers to that tree and the history
3919behind it--most notably the "parent" commits that preceded it in
3920history.
3921
3922Normally a "commit" has one parent: the previous state of the tree
3923before a certain change was made. However, sometimes it can have two
3924or more parent commits, in which case we call it a "merge", due to the
3925fact that such a commit brings together ("merges") two or more
3926previous states represented by other commits.
3927
3928In other words, while a "tree" represents a particular directory state
3929of a working directory, a "commit" represents that state in time,
3930and explains how we got there.
3931
3932You create a commit object by giving it the tree that describes the
3933state at the time of the commit, and a list of parents:
3934
3935-------------------------------------------------
3936$ git commit-tree <tree> -p <parent> [(-p <parent2>)...]
3937-------------------------------------------------
3938
3939and then giving the reason for the commit on stdin (either through
3940redirection from a pipe or file, or by just typing it at the tty).
3941
3942`git commit-tree` will return the name of the object that represents
3943that commit, and you should save it away for later use. Normally,
3944you'd commit a new `HEAD` state, and while Git doesn't care where you
3945save the note about that state, in practice we tend to just write the
3946result to the file pointed at by `.git/HEAD`, so that we can always see
3947what the last committed state was.
3948
3949Here is a picture that illustrates how various pieces fit together:
3950
3951------------
3952
3953                     commit-tree
3954                      commit obj
3955                       +----+
3956                       |    |
3957                       |    |
3958                       V    V
3959                    +-----------+
3960                    | Object DB |
3961                    |  Backing  |
3962                    |   Store   |
3963                    +-----------+
3964                       ^
3965           write-tree  |     |
3966             tree obj  |     |
3967                       |     |  read-tree
3968                       |     |  tree obj
3969                             V
3970                    +-----------+
3971                    |   Index   |
3972                    |  "cache"  |
3973                    +-----------+
3974         update-index  ^
3975             blob obj  |     |
3976                       |     |
3977    checkout-index -u  |     |  checkout-index
3978             stat      |     |  blob obj
3979                             V
3980                    +-----------+
3981                    |  Working  |
3982                    | Directory |
3983                    +-----------+
3984
3985------------
3986
3987
3988[[examining-the-data]]
3989Examining the data
3990------------------
3991
3992You can examine the data represented in the object database and the
3993index with various helper tools. For every object, you can use
3994linkgit:git-cat-file[1] to examine details about the
3995object:
3996
3997-------------------------------------------------
3998$ git cat-file -t <objectname>
3999-------------------------------------------------
4000
4001shows the type of the object, and once you have the type (which is
4002usually implicit in where you find the object), you can use
4003
4004-------------------------------------------------
4005$ git cat-file blob|tree|commit|tag <objectname>
4006-------------------------------------------------
4007
4008to show its contents. NOTE! Trees have binary content, and as a result
4009there is a special helper for showing that content, called
4010`git ls-tree`, which turns the binary content into a more easily
4011readable form.
4012
4013It's especially instructive to look at "commit" objects, since those
4014tend to be small and fairly self-explanatory. In particular, if you
4015follow the convention of having the top commit name in `.git/HEAD`,
4016you can do
4017
4018-------------------------------------------------
4019$ git cat-file commit HEAD
4020-------------------------------------------------
4021
4022to see what the top commit was.
4023
4024[[merging-multiple-trees]]
4025Merging multiple trees
4026----------------------
4027
4028Git can help you perform a three-way merge, which can in turn be
4029used for a many-way merge by repeating the merge procedure several
4030times.  The usual situation is that you only do one three-way merge
4031(reconciling two lines of history) and commit the result, but if
4032you like to, you can merge several branches in one go.
4033
4034To perform a three-way merge, you start with the two commits you
4035want to merge, find their closest common parent (a third commit),
4036and compare the trees corresponding to these three commits.
4037
4038To get the "base" for the merge, look up the common parent of two
4039commits:
4040
4041-------------------------------------------------
4042$ git merge-base <commit1> <commit2>
4043-------------------------------------------------
4044
4045This prints the name of a commit they are both based on. You should
4046now look up the tree objects of those commits, which you can easily
4047do with
4048
4049-------------------------------------------------
4050$ git cat-file commit <commitname> | head -1
4051-------------------------------------------------
4052
4053since the tree object information is always the first line in a commit
4054object.
4055
4056Once you know the three trees you are going to merge (the one "original"
4057tree, aka the common tree, and the two "result" trees, aka the branches
4058you want to merge), you do a "merge" read into the index. This will
4059complain if it has to throw away your old index contents, so you should
4060make sure that you've committed those--in fact you would normally
4061always do a merge against your last commit (which should thus match what
4062you have in your current index anyway).
4063
4064To do the merge, do
4065
4066-------------------------------------------------
4067$ git read-tree -m -u <origtree> <yourtree> <targettree>
4068-------------------------------------------------
4069
4070which will do all trivial merge operations for you directly in the
4071index file, and you can just write the result out with
4072`git write-tree`.
4073
4074
4075[[merging-multiple-trees-2]]
4076Merging multiple trees, continued
4077---------------------------------
4078
4079Sadly, many merges aren't trivial. If there are files that have
4080been added, moved or removed, or if both branches have modified the
4081same file, you will be left with an index tree that contains "merge
4082entries" in it. Such an index tree can 'NOT' be written out to a tree
4083object, and you will have to resolve any such merge clashes using
4084other tools before you can write out the result.
4085
4086You can examine such index state with `git ls-files --unmerged`
4087command.  An example:
4088
4089------------------------------------------------
4090$ git read-tree -m $orig HEAD $target
4091$ git ls-files --unmerged
4092100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1       hello.c
4093100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2       hello.c
4094100644 cc44c73eb783565da5831b4d820c962954019b69 3       hello.c
4095------------------------------------------------
4096
4097Each line of the `git ls-files --unmerged` output begins with
4098the blob mode bits, blob SHA-1, 'stage number', and the
4099filename.  The 'stage number' is Git's way to say which tree it
4100came from: stage 1 corresponds to the `$orig` tree, stage 2 to
4101the `HEAD` tree, and stage 3 to the `$target` tree.
4102
4103Earlier we said that trivial merges are done inside
4104`git read-tree -m`.  For example, if the file did not change
4105from `$orig` to `HEAD` or `$target`, or if the file changed
4106from `$orig` to `HEAD` and `$orig` to `$target` the same way,
4107obviously the final outcome is what is in `HEAD`.  What the
4108above example shows is that file `hello.c` was changed from
4109`$orig` to `HEAD` and `$orig` to `$target` in a different way.
4110You could resolve this by running your favorite 3-way merge
4111program, e.g.  `diff3`, `merge`, or Git's own merge-file, on
4112the blob objects from these three stages yourself, like this:
4113
4114------------------------------------------------
4115$ git cat-file blob 263414f... >hello.c~1
4116$ git cat-file blob 06fa6a2... >hello.c~2
4117$ git cat-file blob cc44c73... >hello.c~3
4118$ git merge-file hello.c~2 hello.c~1 hello.c~3
4119------------------------------------------------
4120
4121This would leave the merge result in `hello.c~2` file, along
4122with conflict markers if there are conflicts.  After verifying
4123the merge result makes sense, you can tell Git what the final
4124merge result for this file is by:
4125
4126-------------------------------------------------
4127$ mv -f hello.c~2 hello.c
4128$ git update-index hello.c
4129-------------------------------------------------
4130
4131When a path is in the "unmerged" state, running `git update-index` for
4132that path tells Git to mark the path resolved.
4133
4134The above is the description of a Git merge at the lowest level,
4135to help you understand what conceptually happens under the hood.
4136In practice, nobody, not even Git itself, runs `git cat-file` three times
4137for this.  There is a `git merge-index` program that extracts the
4138stages to temporary files and calls a "merge" script on it:
4139
4140-------------------------------------------------
4141$ git merge-index git-merge-one-file hello.c
4142-------------------------------------------------
4143
4144and that is what higher level `git merge -s resolve` is implemented with.
4145
4146[[hacking-git]]
4147Hacking Git
4148===========
4149
4150This chapter covers internal details of the Git implementation which
4151probably only Git developers need to understand.
4152
4153[[object-details]]
4154Object storage format
4155---------------------
4156
4157All objects have a statically determined "type" which identifies the
4158format of the object (i.e. how it is used, and how it can refer to other
4159objects).  There are currently four different object types: "blob",
4160"tree", "commit", and "tag".
4161
4162Regardless of object type, all objects share the following
4163characteristics: they are all deflated with zlib, and have a header
4164that not only specifies their type, but also provides size information
4165about the data in the object.  It's worth noting that the SHA-1 hash
4166that is used to name the object is the hash of the original data
4167plus this header, so `sha1sum` 'file' does not match the object name
4168for 'file'.
4169
4170As a result, the general consistency of an object can always be tested
4171independently of the contents or the type of the object: all objects can
4172be validated by verifying that (a) their hashes match the content of the
4173file and (b) the object successfully inflates to a stream of bytes that
4174forms a sequence of
4175`<ascii type without space> + <space> + <ascii decimal size> +
4176<byte\0> + <binary object data>`.
4177
4178The structured objects can further have their structure and
4179connectivity to other objects verified. This is generally done with
4180the `git fsck` program, which generates a full dependency graph
4181of all objects, and verifies their internal consistency (in addition
4182to just verifying their superficial consistency through the hash).
4183
4184[[birdview-on-the-source-code]]
4185A birds-eye view of Git's source code
4186-------------------------------------
4187
4188It is not always easy for new developers to find their way through Git's
4189source code.  This section gives you a little guidance to show where to
4190start.
4191
4192A good place to start is with the contents of the initial commit, with:
4193
4194----------------------------------------------------
4195$ git checkout e83c5163
4196----------------------------------------------------
4197
4198The initial revision lays the foundation for almost everything Git has
4199today, but is small enough to read in one sitting.
4200
4201Note that terminology has changed since that revision.  For example, the
4202README in that revision uses the word "changeset" to describe what we
4203now call a <<def_commit_object,commit>>.
4204
4205Also, we do not call it "cache" any more, but rather "index"; however, the
4206file is still called `cache.h`.  Remark: Not much reason to change it now,
4207especially since there is no good single name for it anyway, because it is
4208basically _the_ header file which is included by _all_ of Git's C sources.
4209
4210If you grasp the ideas in that initial commit, you should check out a
4211more recent version and skim `cache.h`, `object.h` and `commit.h`.
4212
4213In the early days, Git (in the tradition of UNIX) was a bunch of programs
4214which were extremely simple, and which you used in scripts, piping the
4215output of one into another. This turned out to be good for initial
4216development, since it was easier to test new things.  However, recently
4217many of these parts have become builtins, and some of the core has been
4218"libified", i.e. put into libgit.a for performance, portability reasons,
4219and to avoid code duplication.
4220
4221By now, you know what the index is (and find the corresponding data
4222structures in `cache.h`), and that there are just a couple of object types
4223(blobs, trees, commits and tags) which inherit their common structure from
4224`struct object`, which is their first member (and thus, you can cast e.g.
4225`(struct object *)commit` to achieve the _same_ as `&commit->object`, i.e.
4226get at the object name and flags).
4227
4228Now is a good point to take a break to let this information sink in.
4229
4230Next step: get familiar with the object naming.  Read <<naming-commits>>.
4231There are quite a few ways to name an object (and not only revisions!).
4232All of these are handled in `sha1_name.c`. Just have a quick look at
4233the function `get_sha1()`. A lot of the special handling is done by
4234functions like `get_sha1_basic()` or the likes.
4235
4236This is just to get you into the groove for the most libified part of Git:
4237the revision walker.
4238
4239Basically, the initial version of `git log` was a shell script:
4240
4241----------------------------------------------------------------
4242$ git-rev-list --pretty $(git-rev-parse --default HEAD "$@") | \
4243        LESS=-S ${PAGER:-less}
4244----------------------------------------------------------------
4245
4246What does this mean?
4247
4248`git rev-list` is the original version of the revision walker, which
4249_always_ printed a list of revisions to stdout.  It is still functional,
4250and needs to, since most new Git commands start out as scripts using
4251`git rev-list`.
4252
4253`git rev-parse` is not as important any more; it was only used to filter out
4254options that were relevant for the different plumbing commands that were
4255called by the script.
4256
4257Most of what `git rev-list` did is contained in `revision.c` and
4258`revision.h`.  It wraps the options in a struct named `rev_info`, which
4259controls how and what revisions are walked, and more.
4260
4261The original job of `git rev-parse` is now taken by the function
4262`setup_revisions()`, which parses the revisions and the common command-line
4263options for the revision walker. This information is stored in the struct
4264`rev_info` for later consumption. You can do your own command-line option
4265parsing after calling `setup_revisions()`. After that, you have to call
4266`prepare_revision_walk()` for initialization, and then you can get the
4267commits one by one with the function `get_revision()`.
4268
4269If you are interested in more details of the revision walking process,
4270just have a look at the first implementation of `cmd_log()`; call
4271`git show v1.3.0~155^2~4` and scroll down to that function (note that you
4272no longer need to call `setup_pager()` directly).
4273
4274Nowadays, `git log` is a builtin, which means that it is _contained_ in the
4275command `git`.  The source side of a builtin is
4276
4277- a function called `cmd_<bla>`, typically defined in `builtin/<bla.c>`
4278  (note that older versions of Git used to have it in `builtin-<bla>.c`
4279  instead), and declared in `builtin.h`.
4280
4281- an entry in the `commands[]` array in `git.c`, and
4282
4283- an entry in `BUILTIN_OBJECTS` in the `Makefile`.
4284
4285Sometimes, more than one builtin is contained in one source file.  For
4286example, `cmd_whatchanged()` and `cmd_log()` both reside in `builtin/log.c`,
4287since they share quite a bit of code.  In that case, the commands which are
4288_not_ named like the `.c` file in which they live have to be listed in
4289`BUILT_INS` in the `Makefile`.
4290
4291`git log` looks more complicated in C than it does in the original script,
4292but that allows for a much greater flexibility and performance.
4293
4294Here again it is a good point to take a pause.
4295
4296Lesson three is: study the code.  Really, it is the best way to learn about
4297the organization of Git (after you know the basic concepts).
4298
4299So, think about something which you are interested in, say, "how can I
4300access a blob just knowing the object name of it?".  The first step is to
4301find a Git command with which you can do it.  In this example, it is either
4302`git show` or `git cat-file`.
4303
4304For the sake of clarity, let's stay with `git cat-file`, because it
4305
4306- is plumbing, and
4307
4308- was around even in the initial commit (it literally went only through
4309  some 20 revisions as `cat-file.c`, was renamed to `builtin/cat-file.c`
4310  when made a builtin, and then saw less than 10 versions).
4311
4312So, look into `builtin/cat-file.c`, search for `cmd_cat_file()` and look what
4313it does.
4314
4315------------------------------------------------------------------
4316        git_config(git_default_config);
4317        if (argc != 3)
4318                usage("git cat-file [-t|-s|-e|-p|<type>] <sha1>");
4319        if (get_sha1(argv[2], sha1))
4320                die("Not a valid object name %s", argv[2]);
4321------------------------------------------------------------------
4322
4323Let's skip over the obvious details; the only really interesting part
4324here is the call to `get_sha1()`.  It tries to interpret `argv[2]` as an
4325object name, and if it refers to an object which is present in the current
4326repository, it writes the resulting SHA-1 into the variable `sha1`.
4327
4328Two things are interesting here:
4329
4330- `get_sha1()` returns 0 on _success_.  This might surprise some new
4331  Git hackers, but there is a long tradition in UNIX to return different
4332  negative numbers in case of different errors--and 0 on success.
4333
4334- the variable `sha1` in the function signature of `get_sha1()` is `unsigned
4335  char *`, but is actually expected to be a pointer to `unsigned
4336  char[20]`.  This variable will contain the 160-bit SHA-1 of the given
4337  commit.  Note that whenever a SHA-1 is passed as `unsigned char *`, it
4338  is the binary representation, as opposed to the ASCII representation in
4339  hex characters, which is passed as `char *`.
4340
4341You will see both of these things throughout the code.
4342
4343Now, for the meat:
4344
4345-----------------------------------------------------------------------------
4346        case 0:
4347                buf = read_object_with_reference(sha1, argv[1], &size, NULL);
4348-----------------------------------------------------------------------------
4349
4350This is how you read a blob (actually, not only a blob, but any type of
4351object).  To know how the function `read_object_with_reference()` actually
4352works, find the source code for it (something like `git grep
4353read_object_with | grep ":[a-z]"` in the Git repository), and read
4354the source.
4355
4356To find out how the result can be used, just read on in `cmd_cat_file()`:
4357
4358-----------------------------------
4359        write_or_die(1, buf, size);
4360-----------------------------------
4361
4362Sometimes, you do not know where to look for a feature.  In many such cases,
4363it helps to search through the output of `git log`, and then `git show` the
4364corresponding commit.
4365
4366Example: If you know that there was some test case for `git bundle`, but
4367do not remember where it was (yes, you _could_ `git grep bundle t/`, but that
4368does not illustrate the point!):
4369
4370------------------------
4371$ git log --no-merges t/
4372------------------------
4373
4374In the pager (`less`), just search for "bundle", go a few lines back,
4375and see that it is in commit 18449ab0...  Now just copy this object name,
4376and paste it into the command line
4377
4378-------------------
4379$ git show 18449ab0
4380-------------------
4381
4382Voila.
4383
4384Another example: Find out what to do in order to make some script a
4385builtin:
4386
4387-------------------------------------------------
4388$ git log --no-merges --diff-filter=A builtin/*.c
4389-------------------------------------------------
4390
4391You see, Git is actually the best tool to find out about the source of Git
4392itself!
4393
4394[[glossary]]
4395Git Glossary
4396============
4397
4398include::glossary-content.txt[]
4399
4400[[git-quick-start]]
4401Appendix A: Git Quick Reference
4402===============================
4403
4404This is a quick summary of the major commands; the previous chapters
4405explain how these work in more detail.
4406
4407[[quick-creating-a-new-repository]]
4408Creating a new repository
4409-------------------------
4410
4411From a tarball:
4412
4413-----------------------------------------------
4414$ tar xzf project.tar.gz
4415$ cd project
4416$ git init
4417Initialized empty Git repository in .git/
4418$ git add .
4419$ git commit
4420-----------------------------------------------
4421
4422From a remote repository:
4423
4424-----------------------------------------------
4425$ git clone git://example.com/pub/project.git
4426$ cd project
4427-----------------------------------------------
4428
4429[[managing-branches]]
4430Managing branches
4431-----------------
4432
4433-----------------------------------------------
4434$ git branch         # list all local branches in this repo
4435$ git checkout test  # switch working directory to branch "test"
4436$ git branch new     # create branch "new" starting at current HEAD
4437$ git branch -d new  # delete branch "new"
4438-----------------------------------------------
4439
4440Instead of basing a new branch on current HEAD (the default), use:
4441
4442-----------------------------------------------
4443$ git branch new test    # branch named "test"
4444$ git branch new v2.6.15 # tag named v2.6.15
4445$ git branch new HEAD^   # commit before the most recent
4446$ git branch new HEAD^^  # commit before that
4447$ git branch new test~10 # ten commits before tip of branch "test"
4448-----------------------------------------------
4449
4450Create and switch to a new branch at the same time:
4451
4452-----------------------------------------------
4453$ git checkout -b new v2.6.15
4454-----------------------------------------------
4455
4456Update and examine branches from the repository you cloned from:
4457
4458-----------------------------------------------
4459$ git fetch             # update
4460$ git branch -r         # list
4461  origin/master
4462  origin/next
4463  ...
4464$ git checkout -b masterwork origin/master
4465-----------------------------------------------
4466
4467Fetch a branch from a different repository, and give it a new
4468name in your repository:
4469
4470-----------------------------------------------
4471$ git fetch git://example.com/project.git theirbranch:mybranch
4472$ git fetch git://example.com/project.git v2.6.15:mybranch
4473-----------------------------------------------
4474
4475Keep a list of repositories you work with regularly:
4476
4477-----------------------------------------------
4478$ git remote add example git://example.com/project.git
4479$ git remote                    # list remote repositories
4480example
4481origin
4482$ git remote show example       # get details
4483* remote example
4484  URL: git://example.com/project.git
4485  Tracked remote branches
4486    master
4487    next
4488    ...
4489$ git fetch example             # update branches from example
4490$ git branch -r                 # list all remote branches
4491-----------------------------------------------
4492
4493
4494[[exploring-history]]
4495Exploring history
4496-----------------
4497
4498-----------------------------------------------
4499$ gitk                      # visualize and browse history
4500$ git log                   # list all commits
4501$ git log src/              # ...modifying src/
4502$ git log v2.6.15..v2.6.16  # ...in v2.6.16, not in v2.6.15
4503$ git log master..test      # ...in branch test, not in branch master
4504$ git log test..master      # ...in branch master, but not in test
4505$ git log test...master     # ...in one branch, not in both
4506$ git log -S'foo()'         # ...where difference contain "foo()"
4507$ git log --since="2 weeks ago"
4508$ git log -p                # show patches as well
4509$ git show                  # most recent commit
4510$ git diff v2.6.15..v2.6.16 # diff between two tagged versions
4511$ git diff v2.6.15..HEAD    # diff with current head
4512$ git grep "foo()"          # search working directory for "foo()"
4513$ git grep v2.6.15 "foo()"  # search old tree for "foo()"
4514$ git show v2.6.15:a.txt    # look at old version of a.txt
4515-----------------------------------------------
4516
4517Search for regressions:
4518
4519-----------------------------------------------
4520$ git bisect start
4521$ git bisect bad                # current version is bad
4522$ git bisect good v2.6.13-rc2   # last known good revision
4523Bisecting: 675 revisions left to test after this
4524                                # test here, then:
4525$ git bisect good               # if this revision is good, or
4526$ git bisect bad                # if this revision is bad.
4527                                # repeat until done.
4528-----------------------------------------------
4529
4530[[making-changes]]
4531Making changes
4532--------------
4533
4534Make sure Git knows who to blame:
4535
4536------------------------------------------------
4537$ cat >>~/.gitconfig <<\EOF
4538[user]
4539        name = Your Name Comes Here
4540        email = you@yourdomain.example.com
4541EOF
4542------------------------------------------------
4543
4544Select file contents to include in the next commit, then make the
4545commit:
4546
4547-----------------------------------------------
4548$ git add a.txt    # updated file
4549$ git add b.txt    # new file
4550$ git rm c.txt     # old file
4551$ git commit
4552-----------------------------------------------
4553
4554Or, prepare and create the commit in one step:
4555
4556-----------------------------------------------
4557$ git commit d.txt # use latest content only of d.txt
4558$ git commit -a    # use latest content of all tracked files
4559-----------------------------------------------
4560
4561[[merging]]
4562Merging
4563-------
4564
4565-----------------------------------------------
4566$ git merge test   # merge branch "test" into the current branch
4567$ git pull git://example.com/project.git master
4568                   # fetch and merge in remote branch
4569$ git pull . test  # equivalent to git merge test
4570-----------------------------------------------
4571
4572[[sharing-your-changes]]
4573Sharing your changes
4574--------------------
4575
4576Importing or exporting patches:
4577
4578-----------------------------------------------
4579$ git format-patch origin..HEAD # format a patch for each commit
4580                                # in HEAD but not in origin
4581$ git am mbox # import patches from the mailbox "mbox"
4582-----------------------------------------------
4583
4584Fetch a branch in a different Git repository, then merge into the
4585current branch:
4586
4587-----------------------------------------------
4588$ git pull git://example.com/project.git theirbranch
4589-----------------------------------------------
4590
4591Store the fetched branch into a local branch before merging into the
4592current branch:
4593
4594-----------------------------------------------
4595$ git pull git://example.com/project.git theirbranch:mybranch
4596-----------------------------------------------
4597
4598After creating commits on a local branch, update the remote
4599branch with your commits:
4600
4601-----------------------------------------------
4602$ git push ssh://example.com/project.git mybranch:theirbranch
4603-----------------------------------------------
4604
4605When remote and local branch are both named "test":
4606
4607-----------------------------------------------
4608$ git push ssh://example.com/project.git test
4609-----------------------------------------------
4610
4611Shortcut version for a frequently used remote repository:
4612
4613-----------------------------------------------
4614$ git remote add example ssh://example.com/project.git
4615$ git push example test
4616-----------------------------------------------
4617
4618[[repository-maintenance]]
4619Repository maintenance
4620----------------------
4621
4622Check for corruption:
4623
4624-----------------------------------------------
4625$ git fsck
4626-----------------------------------------------
4627
4628Recompress, remove unused cruft:
4629
4630-----------------------------------------------
4631$ git gc
4632-----------------------------------------------
4633
4634
4635[[todo]]
4636Appendix B: Notes and todo list for this manual
4637===============================================
4638
4639This is a work in progress.
4640
4641The basic requirements:
4642
4643- It must be readable in order, from beginning to end, by someone
4644  intelligent with a basic grasp of the UNIX command line, but without
4645  any special knowledge of Git.  If necessary, any other prerequisites
4646  should be specifically mentioned as they arise.
4647- Whenever possible, section headings should clearly describe the task
4648  they explain how to do, in language that requires no more knowledge
4649  than necessary: for example, "importing patches into a project" rather
4650  than "the `git am` command"
4651
4652Think about how to create a clear chapter dependency graph that will
4653allow people to get to important topics without necessarily reading
4654everything in between.
4655
4656Scan `Documentation/` for other stuff left out; in particular:
4657
4658- howto's
4659- some of `technical/`?
4660- hooks
4661- list of commands in linkgit:git[1]
4662
4663Scan email archives for other stuff left out
4664
4665Scan man pages to see if any assume more background than this manual
4666provides.
4667
4668Add more good examples.  Entire sections of just cookbook examples
4669might be a good idea; maybe make an "advanced examples" section a
4670standard end-of-chapter section?
4671
4672Include cross-references to the glossary, where appropriate.
4673
4674Add a section on working with other version control systems, including
4675CVS, Subversion, and just imports of series of release tarballs.
4676
4677Write a chapter on using plumbing and writing scripts.
4678
4679Alternates, clone -reference, etc.
4680
4681More on recovery from repository corruption.  See:
4682        http://marc.info/?l=git&m=117263864820799&w=2
4683        http://marc.info/?l=git&m=117147855503798&w=2