1Git User's Manual (for version 1.5.3 or newer) 2______________________________________________ 3 4 5Git is a fast distributed revision control system. 6 7This manual is designed to be readable by someone with basic UNIX 8command-line skills, but no previous knowledge of git. 9 10<<repositories-and-branches>> and <<exploring-git-history>> explain how 11to fetch and study a project using git--read these chapters to learn how 12to build and test a particular version of a software project, search for 13regressions, and so on. 14 15People needing to do actual development will also want to read 16<<Developing-With-git>> and <<sharing-development>>. 17 18Further chapters cover more specialized topics. 19 20Comprehensive reference documentation is available through the man 21pages, or linkgit:git-help[1] command. For example, for the command 22"git clone <repo>", you can either use: 23 24------------------------------------------------ 25$ man git-clone 26------------------------------------------------ 27 28or: 29 30------------------------------------------------ 31$ git help clone 32------------------------------------------------ 33 34With the latter, you can use the manual viewer of your choice; see 35linkgit:git-help[1] for more information. 36 37See also <<git-quick-start>> for a brief overview of git commands, 38without any explanation. 39 40Finally, see <<todo>> for ways that you can help make this manual more 41complete. 42 43 44[[repositories-and-branches]] 45Repositories and Branches 46========================= 47 48[[how-to-get-a-git-repository]] 49How to get a git repository 50--------------------------- 51 52It will be useful to have a git repository to experiment with as you 53read this manual. 54 55The best way to get one is by using the linkgit:git-clone[1] command to 56download a copy of an existing repository. If you don't already have a 57project in mind, here are some interesting examples: 58 59------------------------------------------------ 60 # git itself (approx. 10MB download): 61$ git clone git://git.kernel.org/pub/scm/git/git.git 62 # the Linux kernel (approx. 150MB download): 63$ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git 64------------------------------------------------ 65 66The initial clone may be time-consuming for a large project, but you 67will only need to clone once. 68 69The clone command creates a new directory named after the project ("git" 70or "linux-2.6" in the examples above). After you cd into this 71directory, you will see that it contains a copy of the project files, 72called the <<def_working_tree,working tree>>, together with a special 73top-level directory named ".git", which contains all the information 74about the history of the project. 75 76[[how-to-check-out]] 77How to check out a different version of a project 78------------------------------------------------- 79 80Git is best thought of as a tool for storing the history of a collection 81of files. It stores the history as a compressed collection of 82interrelated snapshots of the project's contents. In git each such 83version is called a <<def_commit,commit>>. 84 85Those snapshots aren't necessarily all arranged in a single line from 86oldest to newest; instead, work may simultaneously proceed along 87parallel lines of development, called <<def_branch,branches>>, which may 88merge and diverge. 89 90A single git repository can track development on multiple branches. It 91does this by keeping a list of <<def_head,heads>> which reference the 92latest commit on each branch; the linkgit:git-branch[1] command shows 93you the list of branch heads: 94 95------------------------------------------------ 96$ git branch 97* master 98------------------------------------------------ 99 100A freshly cloned repository contains a single branch head, by default 101named "master", with the working directory initialized to the state of 102the project referred to by that branch head. 103 104Most projects also use <<def_tag,tags>>. Tags, like heads, are 105references into the project's history, and can be listed using the 106linkgit:git-tag[1] command: 107 108------------------------------------------------ 109$ git tag -l 110v2.6.11 111v2.6.11-tree 112v2.6.12 113v2.6.12-rc2 114v2.6.12-rc3 115v2.6.12-rc4 116v2.6.12-rc5 117v2.6.12-rc6 118v2.6.13 119... 120------------------------------------------------ 121 122Tags are expected to always point at the same version of a project, 123while heads are expected to advance as development progresses. 124 125Create a new branch head pointing to one of these versions and check it 126out using linkgit:git-checkout[1]: 127 128------------------------------------------------ 129$ git checkout -b new v2.6.13 130------------------------------------------------ 131 132The working directory then reflects the contents that the project had 133when it was tagged v2.6.13, and linkgit:git-branch[1] shows two 134branches, with an asterisk marking the currently checked-out branch: 135 136------------------------------------------------ 137$ git branch 138 master 139* new 140------------------------------------------------ 141 142If you decide that you'd rather see version 2.6.17, you can modify 143the current branch to point at v2.6.17 instead, with 144 145------------------------------------------------ 146$ git reset --hard v2.6.17 147------------------------------------------------ 148 149Note that if the current branch head was your only reference to a 150particular point in history, then resetting that branch may leave you 151with no way to find the history it used to point to; so use this command 152carefully. 153 154[[understanding-commits]] 155Understanding History: Commits 156------------------------------ 157 158Every change in the history of a project is represented by a commit. 159The linkgit:git-show[1] command shows the most recent commit on the 160current branch: 161 162------------------------------------------------ 163$ git show 164commit 17cf781661e6d38f737f15f53ab552f1e95960d7 165Author: Linus Torvalds <torvalds@ppc970.osdl.org.(none)> 166Date: Tue Apr 19 14:11:06 2005 -0700 167 168 Remove duplicate getenv(DB_ENVIRONMENT) call 169 170 Noted by Tony Luck. 171 172diff --git a/init-db.c b/init-db.c 173index 65898fa..b002dc6 100644 174--- a/init-db.c 175+++ b/init-db.c 176@@ -7,7 +7,7 @@ 177 178 int main(int argc, char **argv) 179 { 180- char *sha1_dir = getenv(DB_ENVIRONMENT), *path; 181+ char *sha1_dir, *path; 182 int len, i; 183 184 if (mkdir(".git", 0755) < 0) { 185------------------------------------------------ 186 187As you can see, a commit shows who made the latest change, what they 188did, and why. 189 190Every commit has a 40-hexdigit id, sometimes called the "object name" or the 191"SHA-1 id", shown on the first line of the "git show" output. You can usually 192refer to a commit by a shorter name, such as a tag or a branch name, but this 193longer name can also be useful. Most importantly, it is a globally unique 194name for this commit: so if you tell somebody else the object name (for 195example in email), then you are guaranteed that name will refer to the same 196commit in their repository that it does in yours (assuming their repository 197has that commit at all). Since the object name is computed as a hash over the 198contents of the commit, you are guaranteed that the commit can never change 199without its name also changing. 200 201In fact, in <<git-concepts>> we shall see that everything stored in git 202history, including file data and directory contents, is stored in an object 203with a name that is a hash of its contents. 204 205[[understanding-reachability]] 206Understanding history: commits, parents, and reachability 207~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 208 209Every commit (except the very first commit in a project) also has a 210parent commit which shows what happened before this commit. 211Following the chain of parents will eventually take you back to the 212beginning of the project. 213 214However, the commits do not form a simple list; git allows lines of 215development to diverge and then reconverge, and the point where two 216lines of development reconverge is called a "merge". The commit 217representing a merge can therefore have more than one parent, with 218each parent representing the most recent commit on one of the lines 219of development leading to that point. 220 221The best way to see how this works is using the linkgit:gitk[1] 222command; running gitk now on a git repository and looking for merge 223commits will help understand how the git organizes history. 224 225In the following, we say that commit X is "reachable" from commit Y 226if commit X is an ancestor of commit Y. Equivalently, you could say 227that Y is a descendant of X, or that there is a chain of parents 228leading from commit Y to commit X. 229 230[[history-diagrams]] 231Understanding history: History diagrams 232~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 233 234We will sometimes represent git history using diagrams like the one 235below. Commits are shown as "o", and the links between them with 236lines drawn with - / and \. Time goes left to right: 237 238 239................................................ 240 o--o--o <-- Branch A 241 / 242 o--o--o <-- master 243 \ 244 o--o--o <-- Branch B 245................................................ 246 247If we need to talk about a particular commit, the character "o" may 248be replaced with another letter or number. 249 250[[what-is-a-branch]] 251Understanding history: What is a branch? 252~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 253 254When we need to be precise, we will use the word "branch" to mean a line 255of development, and "branch head" (or just "head") to mean a reference 256to the most recent commit on a branch. In the example above, the branch 257head named "A" is a pointer to one particular commit, but we refer to 258the line of three commits leading up to that point as all being part of 259"branch A". 260 261However, when no confusion will result, we often just use the term 262"branch" both for branches and for branch heads. 263 264[[manipulating-branches]] 265Manipulating branches 266--------------------- 267 268Creating, deleting, and modifying branches is quick and easy; here's 269a summary of the commands: 270 271git branch:: 272 list all branches 273git branch <branch>:: 274 create a new branch named <branch>, referencing the same 275 point in history as the current branch 276git branch <branch> <start-point>:: 277 create a new branch named <branch>, referencing 278 <start-point>, which may be specified any way you like, 279 including using a branch name or a tag name 280git branch -d <branch>:: 281 delete the branch <branch>; if the branch you are deleting 282 points to a commit which is not reachable from the current 283 branch, this command will fail with a warning. 284git branch -D <branch>:: 285 even if the branch points to a commit not reachable 286 from the current branch, you may know that that commit 287 is still reachable from some other branch or tag. In that 288 case it is safe to use this command to force git to delete 289 the branch. 290git checkout <branch>:: 291 make the current branch <branch>, updating the working 292 directory to reflect the version referenced by <branch> 293git checkout -b <new> <start-point>:: 294 create a new branch <new> referencing <start-point>, and 295 check it out. 296 297The special symbol "HEAD" can always be used to refer to the current 298branch. In fact, git uses a file named "HEAD" in the .git directory to 299remember which branch is current: 300 301------------------------------------------------ 302$ cat .git/HEAD 303ref: refs/heads/master 304------------------------------------------------ 305 306[[detached-head]] 307Examining an old version without creating a new branch 308------------------------------------------------------ 309 310The `git checkout` command normally expects a branch head, but will also 311accept an arbitrary commit; for example, you can check out the commit 312referenced by a tag: 313 314------------------------------------------------ 315$ git checkout v2.6.17 316Note: moving to "v2.6.17" which isn't a local branch 317If you want to create a new branch from this checkout, you may do so 318(now or later) by using -b with the checkout command again. Example: 319 git checkout -b <new_branch_name> 320HEAD is now at 427abfa... Linux v2.6.17 321------------------------------------------------ 322 323The HEAD then refers to the SHA-1 of the commit instead of to a branch, 324and git branch shows that you are no longer on a branch: 325 326------------------------------------------------ 327$ cat .git/HEAD 328427abfa28afedffadfca9dd8b067eb6d36bac53f 329$ git branch 330* (no branch) 331 master 332------------------------------------------------ 333 334In this case we say that the HEAD is "detached". 335 336This is an easy way to check out a particular version without having to 337make up a name for the new branch. You can still create a new branch 338(or tag) for this version later if you decide to. 339 340[[examining-remote-branches]] 341Examining branches from a remote repository 342------------------------------------------- 343 344The "master" branch that was created at the time you cloned is a copy 345of the HEAD in the repository that you cloned from. That repository 346may also have had other branches, though, and your local repository 347keeps branches which track each of those remote branches, called 348remote-tracking branches, which you 349can view using the "-r" option to linkgit:git-branch[1]: 350 351------------------------------------------------ 352$ git branch -r 353 origin/HEAD 354 origin/html 355 origin/maint 356 origin/man 357 origin/master 358 origin/next 359 origin/pu 360 origin/todo 361------------------------------------------------ 362 363In this example, "origin" is called a remote repository, or "remote" 364for short. The branches of this repository are called "remote 365branches" from our point of view. The remote-tracking branches listed 366above were created based on the remote branches at clone time and will 367be updated by "git fetch" (hence "git pull") and "git push". See 368<<Updating-a-repository-With-git-fetch>> for details. 369 370You might want to build on one of these remote-tracking branches 371on a branch of your own, just as you would for a tag: 372 373------------------------------------------------ 374$ git checkout -b my-todo-copy origin/todo 375------------------------------------------------ 376 377You can also check out "origin/todo" directly to examine it or 378write a one-off patch. See <<detached-head,detached head>>. 379 380Note that the name "origin" is just the name that git uses by default 381to refer to the repository that you cloned from. 382 383[[how-git-stores-references]] 384Naming branches, tags, and other references 385------------------------------------------- 386 387Branches, remote-tracking branches, and tags are all references to 388commits. All references are named with a slash-separated path name 389starting with "refs"; the names we've been using so far are actually 390shorthand: 391 392 - The branch "test" is short for "refs/heads/test". 393 - The tag "v2.6.18" is short for "refs/tags/v2.6.18". 394 - "origin/master" is short for "refs/remotes/origin/master". 395 396The full name is occasionally useful if, for example, there ever 397exists a tag and a branch with the same name. 398 399(Newly created refs are actually stored in the .git/refs directory, 400under the path given by their name. However, for efficiency reasons 401they may also be packed together in a single file; see 402linkgit:git-pack-refs[1]). 403 404As another useful shortcut, the "HEAD" of a repository can be referred 405to just using the name of that repository. So, for example, "origin" 406is usually a shortcut for the HEAD branch in the repository "origin". 407 408For the complete list of paths which git checks for references, and 409the order it uses to decide which to choose when there are multiple 410references with the same shorthand name, see the "SPECIFYING 411REVISIONS" section of linkgit:gitrevisions[7]. 412 413[[Updating-a-repository-With-git-fetch]] 414Updating a repository with git fetch 415------------------------------------ 416 417Eventually the developer cloned from will do additional work in her 418repository, creating new commits and advancing the branches to point 419at the new commits. 420 421The command "git fetch", with no arguments, will update all of the 422remote-tracking branches to the latest version found in her 423repository. It will not touch any of your own branches--not even the 424"master" branch that was created for you on clone. 425 426[[fetching-branches]] 427Fetching branches from other repositories 428----------------------------------------- 429 430You can also track branches from repositories other than the one you 431cloned from, using linkgit:git-remote[1]: 432 433------------------------------------------------- 434$ git remote add linux-nfs git://linux-nfs.org/pub/nfs-2.6.git 435$ git fetch linux-nfs 436* refs/remotes/linux-nfs/master: storing branch 'master' ... 437 commit: bf81b46 438------------------------------------------------- 439 440New remote-tracking branches will be stored under the shorthand name 441that you gave "git remote add", in this case linux-nfs: 442 443------------------------------------------------- 444$ git branch -r 445linux-nfs/master 446origin/master 447------------------------------------------------- 448 449If you run "git fetch <remote>" later, the remote-tracking branches for the 450named <remote> will be updated. 451 452If you examine the file .git/config, you will see that git has added 453a new stanza: 454 455------------------------------------------------- 456$ cat .git/config 457... 458[remote "linux-nfs"] 459 url = git://linux-nfs.org/pub/nfs-2.6.git 460 fetch = +refs/heads/*:refs/remotes/linux-nfs/* 461... 462------------------------------------------------- 463 464This is what causes git to track the remote's branches; you may modify 465or delete these configuration options by editing .git/config with a 466text editor. (See the "CONFIGURATION FILE" section of 467linkgit:git-config[1] for details.) 468 469[[exploring-git-history]] 470Exploring git history 471===================== 472 473Git is best thought of as a tool for storing the history of a 474collection of files. It does this by storing compressed snapshots of 475the contents of a file hierarchy, together with "commits" which show 476the relationships between these snapshots. 477 478Git provides extremely flexible and fast tools for exploring the 479history of a project. 480 481We start with one specialized tool that is useful for finding the 482commit that introduced a bug into a project. 483 484[[using-bisect]] 485How to use bisect to find a regression 486-------------------------------------- 487 488Suppose version 2.6.18 of your project worked, but the version at 489"master" crashes. Sometimes the best way to find the cause of such a 490regression is to perform a brute-force search through the project's 491history to find the particular commit that caused the problem. The 492linkgit:git-bisect[1] command can help you do this: 493 494------------------------------------------------- 495$ git bisect start 496$ git bisect good v2.6.18 497$ git bisect bad master 498Bisecting: 3537 revisions left to test after this 499[65934a9a028b88e83e2b0f8b36618fe503349f8e] BLOCK: Make USB storage depend on SCSI rather than selecting it [try #6] 500------------------------------------------------- 501 502If you run "git branch" at this point, you'll see that git has 503temporarily moved you in "(no branch)". HEAD is now detached from any 504branch and points directly to a commit (with commit id 65934...) that 505is reachable from "master" but not from v2.6.18. Compile and test it, 506and see whether it crashes. Assume it does crash. Then: 507 508------------------------------------------------- 509$ git bisect bad 510Bisecting: 1769 revisions left to test after this 511[7eff82c8b1511017ae605f0c99ac275a7e21b867] i2c-core: Drop useless bitmaskings 512------------------------------------------------- 513 514checks out an older version. Continue like this, telling git at each 515stage whether the version it gives you is good or bad, and notice 516that the number of revisions left to test is cut approximately in 517half each time. 518 519After about 13 tests (in this case), it will output the commit id of 520the guilty commit. You can then examine the commit with 521linkgit:git-show[1], find out who wrote it, and mail them your bug 522report with the commit id. Finally, run 523 524------------------------------------------------- 525$ git bisect reset 526------------------------------------------------- 527 528to return you to the branch you were on before. 529 530Note that the version which `git bisect` checks out for you at each 531point is just a suggestion, and you're free to try a different 532version if you think it would be a good idea. For example, 533occasionally you may land on a commit that broke something unrelated; 534run 535 536------------------------------------------------- 537$ git bisect visualize 538------------------------------------------------- 539 540which will run gitk and label the commit it chose with a marker that 541says "bisect". Choose a safe-looking commit nearby, note its commit 542id, and check it out with: 543 544------------------------------------------------- 545$ git reset --hard fb47ddb2db... 546------------------------------------------------- 547 548then test, run "bisect good" or "bisect bad" as appropriate, and 549continue. 550 551Instead of "git bisect visualize" and then "git reset --hard 552fb47ddb2db...", you might just want to tell git that you want to skip 553the current commit: 554 555------------------------------------------------- 556$ git bisect skip 557------------------------------------------------- 558 559In this case, though, git may not eventually be able to tell the first 560bad one between some first skipped commits and a later bad commit. 561 562There are also ways to automate the bisecting process if you have a 563test script that can tell a good from a bad commit. See 564linkgit:git-bisect[1] for more information about this and other "git 565bisect" features. 566 567[[naming-commits]] 568Naming commits 569-------------- 570 571We have seen several ways of naming commits already: 572 573 - 40-hexdigit object name 574 - branch name: refers to the commit at the head of the given 575 branch 576 - tag name: refers to the commit pointed to by the given tag 577 (we've seen branches and tags are special cases of 578 <<how-git-stores-references,references>>). 579 - HEAD: refers to the head of the current branch 580 581There are many more; see the "SPECIFYING REVISIONS" section of the 582linkgit:gitrevisions[7] man page for the complete list of ways to 583name revisions. Some examples: 584 585------------------------------------------------- 586$ git show fb47ddb2 # the first few characters of the object name 587 # are usually enough to specify it uniquely 588$ git show HEAD^ # the parent of the HEAD commit 589$ git show HEAD^^ # the grandparent 590$ git show HEAD~4 # the great-great-grandparent 591------------------------------------------------- 592 593Recall that merge commits may have more than one parent; by default, 594^ and ~ follow the first parent listed in the commit, but you can 595also choose: 596 597------------------------------------------------- 598$ git show HEAD^1 # show the first parent of HEAD 599$ git show HEAD^2 # show the second parent of HEAD 600------------------------------------------------- 601 602In addition to HEAD, there are several other special names for 603commits: 604 605Merges (to be discussed later), as well as operations such as 606`git reset`, which change the currently checked-out commit, generally 607set ORIG_HEAD to the value HEAD had before the current operation. 608 609The `git fetch` operation always stores the head of the last fetched 610branch in FETCH_HEAD. For example, if you run `git fetch` without 611specifying a local branch as the target of the operation 612 613------------------------------------------------- 614$ git fetch git://example.com/proj.git theirbranch 615------------------------------------------------- 616 617the fetched commits will still be available from FETCH_HEAD. 618 619When we discuss merges we'll also see the special name MERGE_HEAD, 620which refers to the other branch that we're merging in to the current 621branch. 622 623The linkgit:git-rev-parse[1] command is a low-level command that is 624occasionally useful for translating some name for a commit to the object 625name for that commit: 626 627------------------------------------------------- 628$ git rev-parse origin 629e05db0fd4f31dde7005f075a84f96b360d05984b 630------------------------------------------------- 631 632[[creating-tags]] 633Creating tags 634------------- 635 636We can also create a tag to refer to a particular commit; after 637running 638 639------------------------------------------------- 640$ git tag stable-1 1b2e1d63ff 641------------------------------------------------- 642 643You can use stable-1 to refer to the commit 1b2e1d63ff. 644 645This creates a "lightweight" tag. If you would also like to include a 646comment with the tag, and possibly sign it cryptographically, then you 647should create a tag object instead; see the linkgit:git-tag[1] man page 648for details. 649 650[[browsing-revisions]] 651Browsing revisions 652------------------ 653 654The linkgit:git-log[1] command can show lists of commits. On its 655own, it shows all commits reachable from the parent commit; but you 656can also make more specific requests: 657 658------------------------------------------------- 659$ git log v2.5.. # commits since (not reachable from) v2.5 660$ git log test..master # commits reachable from master but not test 661$ git log master..test # ...reachable from test but not master 662$ git log master...test # ...reachable from either test or master, 663 # but not both 664$ git log --since="2 weeks ago" # commits from the last 2 weeks 665$ git log Makefile # commits which modify Makefile 666$ git log fs/ # ... which modify any file under fs/ 667$ git log -S'foo()' # commits which add or remove any file data 668 # matching the string 'foo()' 669------------------------------------------------- 670 671And of course you can combine all of these; the following finds 672commits since v2.5 which touch the Makefile or any file under fs: 673 674------------------------------------------------- 675$ git log v2.5.. Makefile fs/ 676------------------------------------------------- 677 678You can also ask git log to show patches: 679 680------------------------------------------------- 681$ git log -p 682------------------------------------------------- 683 684See the "--pretty" option in the linkgit:git-log[1] man page for more 685display options. 686 687Note that git log starts with the most recent commit and works 688backwards through the parents; however, since git history can contain 689multiple independent lines of development, the particular order that 690commits are listed in may be somewhat arbitrary. 691 692[[generating-diffs]] 693Generating diffs 694---------------- 695 696You can generate diffs between any two versions using 697linkgit:git-diff[1]: 698 699------------------------------------------------- 700$ git diff master..test 701------------------------------------------------- 702 703That will produce the diff between the tips of the two branches. If 704you'd prefer to find the diff from their common ancestor to test, you 705can use three dots instead of two: 706 707------------------------------------------------- 708$ git diff master...test 709------------------------------------------------- 710 711Sometimes what you want instead is a set of patches; for this you can 712use linkgit:git-format-patch[1]: 713 714------------------------------------------------- 715$ git format-patch master..test 716------------------------------------------------- 717 718will generate a file with a patch for each commit reachable from test 719but not from master. 720 721[[viewing-old-file-versions]] 722Viewing old file versions 723------------------------- 724 725You can always view an old version of a file by just checking out the 726correct revision first. But sometimes it is more convenient to be 727able to view an old version of a single file without checking 728anything out; this command does that: 729 730------------------------------------------------- 731$ git show v2.5:fs/locks.c 732------------------------------------------------- 733 734Before the colon may be anything that names a commit, and after it 735may be any path to a file tracked by git. 736 737[[history-examples]] 738Examples 739-------- 740 741[[counting-commits-on-a-branch]] 742Counting the number of commits on a branch 743~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 744 745Suppose you want to know how many commits you've made on "mybranch" 746since it diverged from "origin": 747 748------------------------------------------------- 749$ git log --pretty=oneline origin..mybranch | wc -l 750------------------------------------------------- 751 752Alternatively, you may often see this sort of thing done with the 753lower-level command linkgit:git-rev-list[1], which just lists the SHA-1's 754of all the given commits: 755 756------------------------------------------------- 757$ git rev-list origin..mybranch | wc -l 758------------------------------------------------- 759 760[[checking-for-equal-branches]] 761Check whether two branches point at the same history 762~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 763 764Suppose you want to check whether two branches point at the same point 765in history. 766 767------------------------------------------------- 768$ git diff origin..master 769------------------------------------------------- 770 771will tell you whether the contents of the project are the same at the 772two branches; in theory, however, it's possible that the same project 773contents could have been arrived at by two different historical 774routes. You could compare the object names: 775 776------------------------------------------------- 777$ git rev-list origin 778e05db0fd4f31dde7005f075a84f96b360d05984b 779$ git rev-list master 780e05db0fd4f31dde7005f075a84f96b360d05984b 781------------------------------------------------- 782 783Or you could recall that the ... operator selects all commits 784contained reachable from either one reference or the other but not 785both; so 786 787------------------------------------------------- 788$ git log origin...master 789------------------------------------------------- 790 791will return no commits when the two branches are equal. 792 793[[finding-tagged-descendants]] 794Find first tagged version including a given fix 795~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 796 797Suppose you know that the commit e05db0fd fixed a certain problem. 798You'd like to find the earliest tagged release that contains that 799fix. 800 801Of course, there may be more than one answer--if the history branched 802after commit e05db0fd, then there could be multiple "earliest" tagged 803releases. 804 805You could just visually inspect the commits since e05db0fd: 806 807------------------------------------------------- 808$ gitk e05db0fd.. 809------------------------------------------------- 810 811Or you can use linkgit:git-name-rev[1], which will give the commit a 812name based on any tag it finds pointing to one of the commit's 813descendants: 814 815------------------------------------------------- 816$ git name-rev --tags e05db0fd 817e05db0fd tags/v1.5.0-rc1^0~23 818------------------------------------------------- 819 820The linkgit:git-describe[1] command does the opposite, naming the 821revision using a tag on which the given commit is based: 822 823------------------------------------------------- 824$ git describe e05db0fd 825v1.5.0-rc0-260-ge05db0f 826------------------------------------------------- 827 828but that may sometimes help you guess which tags might come after the 829given commit. 830 831If you just want to verify whether a given tagged version contains a 832given commit, you could use linkgit:git-merge-base[1]: 833 834------------------------------------------------- 835$ git merge-base e05db0fd v1.5.0-rc1 836e05db0fd4f31dde7005f075a84f96b360d05984b 837------------------------------------------------- 838 839The merge-base command finds a common ancestor of the given commits, 840and always returns one or the other in the case where one is a 841descendant of the other; so the above output shows that e05db0fd 842actually is an ancestor of v1.5.0-rc1. 843 844Alternatively, note that 845 846------------------------------------------------- 847$ git log v1.5.0-rc1..e05db0fd 848------------------------------------------------- 849 850will produce empty output if and only if v1.5.0-rc1 includes e05db0fd, 851because it outputs only commits that are not reachable from v1.5.0-rc1. 852 853As yet another alternative, the linkgit:git-show-branch[1] command lists 854the commits reachable from its arguments with a display on the left-hand 855side that indicates which arguments that commit is reachable from. So, 856you can run something like 857 858------------------------------------------------- 859$ git show-branch e05db0fd v1.5.0-rc0 v1.5.0-rc1 v1.5.0-rc2 860! [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if 861available 862 ! [v1.5.0-rc0] GIT v1.5.0 preview 863 ! [v1.5.0-rc1] GIT v1.5.0-rc1 864 ! [v1.5.0-rc2] GIT v1.5.0-rc2 865... 866------------------------------------------------- 867 868then search for a line that looks like 869 870------------------------------------------------- 871+ ++ [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if 872available 873------------------------------------------------- 874 875Which shows that e05db0fd is reachable from itself, from v1.5.0-rc1, and 876from v1.5.0-rc2, but not from v1.5.0-rc0. 877 878[[showing-commits-unique-to-a-branch]] 879Showing commits unique to a given branch 880~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 881 882Suppose you would like to see all the commits reachable from the branch 883head named "master" but not from any other head in your repository. 884 885We can list all the heads in this repository with 886linkgit:git-show-ref[1]: 887 888------------------------------------------------- 889$ git show-ref --heads 890bf62196b5e363d73353a9dcf094c59595f3153b7 refs/heads/core-tutorial 891db768d5504c1bb46f63ee9d6e1772bd047e05bf9 refs/heads/maint 892a07157ac624b2524a059a3414e99f6f44bebc1e7 refs/heads/master 89324dbc180ea14dc1aebe09f14c8ecf32010690627 refs/heads/tutorial-2 8941e87486ae06626c2f31eaa63d26fc0fd646c8af2 refs/heads/tutorial-fixes 895------------------------------------------------- 896 897We can get just the branch-head names, and remove "master", with 898the help of the standard utilities cut and grep: 899 900------------------------------------------------- 901$ git show-ref --heads | cut -d' ' -f2 | grep -v '^refs/heads/master' 902refs/heads/core-tutorial 903refs/heads/maint 904refs/heads/tutorial-2 905refs/heads/tutorial-fixes 906------------------------------------------------- 907 908And then we can ask to see all the commits reachable from master 909but not from these other heads: 910 911------------------------------------------------- 912$ gitk master --not $( git show-ref --heads | cut -d' ' -f2 | 913 grep -v '^refs/heads/master' ) 914------------------------------------------------- 915 916Obviously, endless variations are possible; for example, to see all 917commits reachable from some head but not from any tag in the repository: 918 919------------------------------------------------- 920$ gitk $( git show-ref --heads ) --not $( git show-ref --tags ) 921------------------------------------------------- 922 923(See linkgit:gitrevisions[7] for explanations of commit-selecting 924syntax such as `--not`.) 925 926[[making-a-release]] 927Creating a changelog and tarball for a software release 928~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 929 930The linkgit:git-archive[1] command can create a tar or zip archive from 931any version of a project; for example: 932 933------------------------------------------------- 934$ git archive -o latest.tar.gz --prefix=project/ HEAD 935------------------------------------------------- 936 937will use HEAD to produce a gzipped tar archive in which each filename 938is preceded by `project/`. The output file format is inferred from 939the output file extension if possible, see linkgit:git-archive[1] for 940details. 941 942Versions of Git older than 1.7.7 don't know about the 'tar.gz' format, 943you'll need to use gzip explicitly: 944 945------------------------------------------------- 946$ git archive --format=tar --prefix=project/ HEAD | gzip >latest.tar.gz 947------------------------------------------------- 948 949If you're releasing a new version of a software project, you may want 950to simultaneously make a changelog to include in the release 951announcement. 952 953Linus Torvalds, for example, makes new kernel releases by tagging them, 954then running: 955 956------------------------------------------------- 957$ release-script 2.6.12 2.6.13-rc6 2.6.13-rc7 958------------------------------------------------- 959 960where release-script is a shell script that looks like: 961 962------------------------------------------------- 963#!/bin/sh 964stable="$1" 965last="$2" 966new="$3" 967echo "# git tag v$new" 968echo "git archive --prefix=linux-$new/ v$new | gzip -9 > ../linux-$new.tar.gz" 969echo "git diff v$stable v$new | gzip -9 > ../patch-$new.gz" 970echo "git log --no-merges v$new ^v$last > ../ChangeLog-$new" 971echo "git shortlog --no-merges v$new ^v$last > ../ShortLog" 972echo "git diff --stat --summary -M v$last v$new > ../diffstat-$new" 973------------------------------------------------- 974 975and then he just cut-and-pastes the output commands after verifying that 976they look OK. 977 978[[Finding-commits-With-given-Content]] 979Finding commits referencing a file with given content 980~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 981 982Somebody hands you a copy of a file, and asks which commits modified a 983file such that it contained the given content either before or after the 984commit. You can find out with this: 985 986------------------------------------------------- 987$ git log --raw --abbrev=40 --pretty=oneline | 988 grep -B 1 `git hash-object filename` 989------------------------------------------------- 990 991Figuring out why this works is left as an exercise to the (advanced) 992student. The linkgit:git-log[1], linkgit:git-diff-tree[1], and 993linkgit:git-hash-object[1] man pages may prove helpful. 994 995[[Developing-With-git]] 996Developing with git 997=================== 998 999[[telling-git-your-name]]1000Telling git your name1001---------------------10021003Before creating any commits, you should introduce yourself to Git.1004The easiest way to do so is to use linkgit:git-config[1]:10051006------------------------------------------------1007$ git config --global user.name 'Your Name Comes Here'1008$ git config --global user.email 'you@yourdomain.example.com'1009------------------------------------------------10101011Which will add the following to a file named `.gitconfig` in your1012home directory:10131014------------------------------------------------1015[user]1016 name = Your Name Comes Here1017 email = you@yourdomain.example.com1018------------------------------------------------10191020See the "CONFIGURATION FILE" section of linkgit:git-config[1] for1021details on the configuration file. The file is plain text, so you can1022also edit it with your favorite editor.102310241025[[creating-a-new-repository]]1026Creating a new repository1027-------------------------10281029Creating a new repository from scratch is very easy:10301031-------------------------------------------------1032$ mkdir project1033$ cd project1034$ git init1035-------------------------------------------------10361037If you have some initial content (say, a tarball):10381039-------------------------------------------------1040$ tar xzvf project.tar.gz1041$ cd project1042$ git init1043$ git add . # include everything below ./ in the first commit:1044$ git commit1045-------------------------------------------------10461047[[how-to-make-a-commit]]1048How to make a commit1049--------------------10501051Creating a new commit takes three steps:10521053 1. Making some changes to the working directory using your1054 favorite editor.1055 2. Telling git about your changes.1056 3. Creating the commit using the content you told git about1057 in step 2.10581059In practice, you can interleave and repeat steps 1 and 2 as many1060times as you want: in order to keep track of what you want committed1061at step 3, git maintains a snapshot of the tree's contents in a1062special staging area called "the index."10631064At the beginning, the content of the index will be identical to1065that of the HEAD. The command "git diff --cached", which shows1066the difference between the HEAD and the index, should therefore1067produce no output at that point.10681069Modifying the index is easy:10701071To update the index with the new contents of a modified file, use10721073-------------------------------------------------1074$ git add path/to/file1075-------------------------------------------------10761077To add the contents of a new file to the index, use10781079-------------------------------------------------1080$ git add path/to/file1081-------------------------------------------------10821083To remove a file from the index and from the working tree,10841085-------------------------------------------------1086$ git rm path/to/file1087-------------------------------------------------10881089After each step you can verify that10901091-------------------------------------------------1092$ git diff --cached1093-------------------------------------------------10941095always shows the difference between the HEAD and the index file--this1096is what you'd commit if you created the commit now--and that10971098-------------------------------------------------1099$ git diff1100-------------------------------------------------11011102shows the difference between the working tree and the index file.11031104Note that "git add" always adds just the current contents of a file1105to the index; further changes to the same file will be ignored unless1106you run `git add` on the file again.11071108When you're ready, just run11091110-------------------------------------------------1111$ git commit1112-------------------------------------------------11131114and git will prompt you for a commit message and then create the new1115commit. Check to make sure it looks like what you expected with11161117-------------------------------------------------1118$ git show1119-------------------------------------------------11201121As a special shortcut,11221123-------------------------------------------------1124$ git commit -a1125-------------------------------------------------11261127will update the index with any files that you've modified or removed1128and create a commit, all in one step.11291130A number of commands are useful for keeping track of what you're1131about to commit:11321133-------------------------------------------------1134$ git diff --cached # difference between HEAD and the index; what1135 # would be committed if you ran "commit" now.1136$ git diff # difference between the index file and your1137 # working directory; changes that would not1138 # be included if you ran "commit" now.1139$ git diff HEAD # difference between HEAD and working tree; what1140 # would be committed if you ran "commit -a" now.1141$ git status # a brief per-file summary of the above.1142-------------------------------------------------11431144You can also use linkgit:git-gui[1] to create commits, view changes in1145the index and the working tree files, and individually select diff hunks1146for inclusion in the index (by right-clicking on the diff hunk and1147choosing "Stage Hunk For Commit").11481149[[creating-good-commit-messages]]1150Creating good commit messages1151-----------------------------11521153Though not required, it's a good idea to begin the commit message1154with a single short (less than 50 character) line summarizing the1155change, followed by a blank line and then a more thorough1156description. The text up to the first blank line in a commit1157message is treated as the commit title, and that title is used1158throughout git. For example, linkgit:git-format-patch[1] turns a1159commit into email, and it uses the title on the Subject line and the1160rest of the commit in the body.116111621163[[ignoring-files]]1164Ignoring files1165--------------11661167A project will often generate files that you do 'not' want to track with git.1168This typically includes files generated by a build process or temporary1169backup files made by your editor. Of course, 'not' tracking files with git1170is just a matter of 'not' calling `git add` on them. But it quickly becomes1171annoying to have these untracked files lying around; e.g. they make1172`git add .` practically useless, and they keep showing up in the output of1173`git status`.11741175You can tell git to ignore certain files by creating a file called .gitignore1176in the top level of your working directory, with contents such as:11771178-------------------------------------------------1179# Lines starting with '#' are considered comments.1180# Ignore any file named foo.txt.1181foo.txt1182# Ignore (generated) html files,1183*.html1184# except foo.html which is maintained by hand.1185!foo.html1186# Ignore objects and archives.1187*.[oa]1188-------------------------------------------------11891190See linkgit:gitignore[5] for a detailed explanation of the syntax. You can1191also place .gitignore files in other directories in your working tree, and they1192will apply to those directories and their subdirectories. The `.gitignore`1193files can be added to your repository like any other files (just run `git add1194.gitignore` and `git commit`, as usual), which is convenient when the exclude1195patterns (such as patterns matching build output files) would also make sense1196for other users who clone your repository.11971198If you wish the exclude patterns to affect only certain repositories1199(instead of every repository for a given project), you may instead put1200them in a file in your repository named .git/info/exclude, or in any file1201specified by the `core.excludesfile` configuration variable. Some git1202commands can also take exclude patterns directly on the command line.1203See linkgit:gitignore[5] for the details.12041205[[how-to-merge]]1206How to merge1207------------12081209You can rejoin two diverging branches of development using1210linkgit:git-merge[1]:12111212-------------------------------------------------1213$ git merge branchname1214-------------------------------------------------12151216merges the development in the branch "branchname" into the current1217branch.12181219A merge is made by combining the changes made in "branchname" and the1220changes made up to the latest commit in your current branch since1221their histories forked. The work tree is overwritten by the result of1222the merge when this combining is done cleanly, or overwritten by a1223half-merged results when this combining results in conflicts.1224Therefore, if you have uncommitted changes touching the same files as1225the ones impacted by the merge, Git will refuse to proceed. Most of1226the time, you will want to commit your changes before you can merge,1227and if you don't, then linkgit:git-stash[1] can take these changes1228away while you're doing the merge, and reapply them afterwards.12291230If the changes are independent enough, Git will automatically complete1231the merge and commit the result (or reuse an existing commit in case1232of <<fast-forwards,fast-forward>>, see below). On the other hand,1233if there are conflicts--for example, if the same file is1234modified in two different ways in the remote branch and the local1235branch--then you are warned; the output may look something like this:12361237-------------------------------------------------1238$ git merge next1239 100% (4/4) done1240Auto-merged file.txt1241CONFLICT (content): Merge conflict in file.txt1242Automatic merge failed; fix conflicts and then commit the result.1243-------------------------------------------------12441245Conflict markers are left in the problematic files, and after1246you resolve the conflicts manually, you can update the index1247with the contents and run git commit, as you normally would when1248creating a new file.12491250If you examine the resulting commit using gitk, you will see that it1251has two parents, one pointing to the top of the current branch, and1252one to the top of the other branch.12531254[[resolving-a-merge]]1255Resolving a merge1256-----------------12571258When a merge isn't resolved automatically, git leaves the index and1259the working tree in a special state that gives you all the1260information you need to help resolve the merge.12611262Files with conflicts are marked specially in the index, so until you1263resolve the problem and update the index, linkgit:git-commit[1] will1264fail:12651266-------------------------------------------------1267$ git commit1268file.txt: needs merge1269-------------------------------------------------12701271Also, linkgit:git-status[1] will list those files as "unmerged", and the1272files with conflicts will have conflict markers added, like this:12731274-------------------------------------------------1275<<<<<<< HEAD:file.txt1276Hello world1277=======1278Goodbye1279>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt1280-------------------------------------------------12811282All you need to do is edit the files to resolve the conflicts, and then12831284-------------------------------------------------1285$ git add file.txt1286$ git commit1287-------------------------------------------------12881289Note that the commit message will already be filled in for you with1290some information about the merge. Normally you can just use this1291default message unchanged, but you may add additional commentary of1292your own if desired.12931294The above is all you need to know to resolve a simple merge. But git1295also provides more information to help resolve conflicts:12961297[[conflict-resolution]]1298Getting conflict-resolution help during a merge1299~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~13001301All of the changes that git was able to merge automatically are1302already added to the index file, so linkgit:git-diff[1] shows only1303the conflicts. It uses an unusual syntax:13041305-------------------------------------------------1306$ git diff1307diff --cc file.txt1308index 802992c,2b60207..00000001309--- a/file.txt1310+++ b/file.txt1311@@@ -1,1 -1,1 +1,5 @@@1312++<<<<<<< HEAD:file.txt1313 +Hello world1314++=======1315+ Goodbye1316++>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt1317-------------------------------------------------13181319Recall that the commit which will be committed after we resolve this1320conflict will have two parents instead of the usual one: one parent1321will be HEAD, the tip of the current branch; the other will be the1322tip of the other branch, which is stored temporarily in MERGE_HEAD.13231324During the merge, the index holds three versions of each file. Each of1325these three "file stages" represents a different version of the file:13261327-------------------------------------------------1328$ git show :1:file.txt # the file in a common ancestor of both branches1329$ git show :2:file.txt # the version from HEAD.1330$ git show :3:file.txt # the version from MERGE_HEAD.1331-------------------------------------------------13321333When you ask linkgit:git-diff[1] to show the conflicts, it runs a1334three-way diff between the conflicted merge results in the work tree with1335stages 2 and 3 to show only hunks whose contents come from both sides,1336mixed (in other words, when a hunk's merge results come only from stage 2,1337that part is not conflicting and is not shown. Same for stage 3).13381339The diff above shows the differences between the working-tree version of1340file.txt and the stage 2 and stage 3 versions. So instead of preceding1341each line by a single "+" or "-", it now uses two columns: the first1342column is used for differences between the first parent and the working1343directory copy, and the second for differences between the second parent1344and the working directory copy. (See the "COMBINED DIFF FORMAT" section1345of linkgit:git-diff-files[1] for a details of the format.)13461347After resolving the conflict in the obvious way (but before updating the1348index), the diff will look like:13491350-------------------------------------------------1351$ git diff1352diff --cc file.txt1353index 802992c,2b60207..00000001354--- a/file.txt1355+++ b/file.txt1356@@@ -1,1 -1,1 +1,1 @@@1357- Hello world1358 -Goodbye1359++Goodbye world1360-------------------------------------------------13611362This shows that our resolved version deleted "Hello world" from the1363first parent, deleted "Goodbye" from the second parent, and added1364"Goodbye world", which was previously absent from both.13651366Some special diff options allow diffing the working directory against1367any of these stages:13681369-------------------------------------------------1370$ git diff -1 file.txt # diff against stage 11371$ git diff --base file.txt # same as the above1372$ git diff -2 file.txt # diff against stage 21373$ git diff --ours file.txt # same as the above1374$ git diff -3 file.txt # diff against stage 31375$ git diff --theirs file.txt # same as the above.1376-------------------------------------------------13771378The linkgit:git-log[1] and linkgit:gitk[1] commands also provide special help1379for merges:13801381-------------------------------------------------1382$ git log --merge1383$ gitk --merge1384-------------------------------------------------13851386These will display all commits which exist only on HEAD or on1387MERGE_HEAD, and which touch an unmerged file.13881389You may also use linkgit:git-mergetool[1], which lets you merge the1390unmerged files using external tools such as Emacs or kdiff3.13911392Each time you resolve the conflicts in a file and update the index:13931394-------------------------------------------------1395$ git add file.txt1396-------------------------------------------------13971398the different stages of that file will be "collapsed", after which1399`git diff` will (by default) no longer show diffs for that file.14001401[[undoing-a-merge]]1402Undoing a merge1403---------------14041405If you get stuck and decide to just give up and throw the whole mess1406away, you can always return to the pre-merge state with14071408-------------------------------------------------1409$ git reset --hard HEAD1410-------------------------------------------------14111412Or, if you've already committed the merge that you want to throw away,14131414-------------------------------------------------1415$ git reset --hard ORIG_HEAD1416-------------------------------------------------14171418However, this last command can be dangerous in some cases--never1419throw away a commit you have already committed if that commit may1420itself have been merged into another branch, as doing so may confuse1421further merges.14221423[[fast-forwards]]1424Fast-forward merges1425-------------------14261427There is one special case not mentioned above, which is treated1428differently. Normally, a merge results in a merge commit, with two1429parents, one pointing at each of the two lines of development that1430were merged.14311432However, if the current branch is a descendant of the other--so every1433commit present in the one is already contained in the other--then git1434just performs a "fast-forward"; the head of the current branch is moved1435forward to point at the head of the merged-in branch, without any new1436commits being created.14371438[[fixing-mistakes]]1439Fixing mistakes1440---------------14411442If you've messed up the working tree, but haven't yet committed your1443mistake, you can return the entire working tree to the last committed1444state with14451446-------------------------------------------------1447$ git reset --hard HEAD1448-------------------------------------------------14491450If you make a commit that you later wish you hadn't, there are two1451fundamentally different ways to fix the problem:14521453 1. You can create a new commit that undoes whatever was done1454 by the old commit. This is the correct thing if your1455 mistake has already been made public.14561457 2. You can go back and modify the old commit. You should1458 never do this if you have already made the history public;1459 git does not normally expect the "history" of a project to1460 change, and cannot correctly perform repeated merges from1461 a branch that has had its history changed.14621463[[reverting-a-commit]]1464Fixing a mistake with a new commit1465~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~14661467Creating a new commit that reverts an earlier change is very easy;1468just pass the linkgit:git-revert[1] command a reference to the bad1469commit; for example, to revert the most recent commit:14701471-------------------------------------------------1472$ git revert HEAD1473-------------------------------------------------14741475This will create a new commit which undoes the change in HEAD. You1476will be given a chance to edit the commit message for the new commit.14771478You can also revert an earlier change, for example, the next-to-last:14791480-------------------------------------------------1481$ git revert HEAD^1482-------------------------------------------------14831484In this case git will attempt to undo the old change while leaving1485intact any changes made since then. If more recent changes overlap1486with the changes to be reverted, then you will be asked to fix1487conflicts manually, just as in the case of <<resolving-a-merge,1488resolving a merge>>.14891490[[fixing-a-mistake-by-rewriting-history]]1491Fixing a mistake by rewriting history1492~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~14931494If the problematic commit is the most recent commit, and you have not1495yet made that commit public, then you may just1496<<undoing-a-merge,destroy it using `git reset`>>.14971498Alternatively, you1499can edit the working directory and update the index to fix your1500mistake, just as if you were going to <<how-to-make-a-commit,create a1501new commit>>, then run15021503-------------------------------------------------1504$ git commit --amend1505-------------------------------------------------15061507which will replace the old commit by a new commit incorporating your1508changes, giving you a chance to edit the old commit message first.15091510Again, you should never do this to a commit that may already have1511been merged into another branch; use linkgit:git-revert[1] instead in1512that case.15131514It is also possible to replace commits further back in the history, but1515this is an advanced topic to be left for1516<<cleaning-up-history,another chapter>>.15171518[[checkout-of-path]]1519Checking out an old version of a file1520~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~15211522In the process of undoing a previous bad change, you may find it1523useful to check out an older version of a particular file using1524linkgit:git-checkout[1]. We've used `git checkout` before to switch1525branches, but it has quite different behavior if it is given a path1526name: the command15271528-------------------------------------------------1529$ git checkout HEAD^ path/to/file1530-------------------------------------------------15311532replaces path/to/file by the contents it had in the commit HEAD^, and1533also updates the index to match. It does not change branches.15341535If you just want to look at an old version of the file, without1536modifying the working directory, you can do that with1537linkgit:git-show[1]:15381539-------------------------------------------------1540$ git show HEAD^:path/to/file1541-------------------------------------------------15421543which will display the given version of the file.15441545[[interrupted-work]]1546Temporarily setting aside work in progress1547~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~15481549While you are in the middle of working on something complicated, you1550find an unrelated but obvious and trivial bug. You would like to fix it1551before continuing. You can use linkgit:git-stash[1] to save the current1552state of your work, and after fixing the bug (or, optionally after doing1553so on a different branch and then coming back), unstash the1554work-in-progress changes.15551556------------------------------------------------1557$ git stash save "work in progress for foo feature"1558------------------------------------------------15591560This command will save your changes away to the `stash`, and1561reset your working tree and the index to match the tip of your1562current branch. Then you can make your fix as usual.15631564------------------------------------------------1565... edit and test ...1566$ git commit -a -m "blorpl: typofix"1567------------------------------------------------15681569After that, you can go back to what you were working on with1570`git stash pop`:15711572------------------------------------------------1573$ git stash pop1574------------------------------------------------157515761577[[ensuring-good-performance]]1578Ensuring good performance1579-------------------------15801581On large repositories, Git depends on compression to keep the history1582information from taking up too much space on disk or in memory. Some1583git commands may automatically run linkgit:git-gc[1], so you don't1584have to worry about running it manually. However, compressing a large1585repository may take a while, so you may want to call `gc` explicitly1586to avoid automatic compression kicking in when it is not convenient.158715881589[[ensuring-reliability]]1590Ensuring reliability1591--------------------15921593[[checking-for-corruption]]1594Checking the repository for corruption1595~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~15961597The linkgit:git-fsck[1] command runs a number of self-consistency checks1598on the repository, and reports on any problems. This may take some1599time.16001601-------------------------------------------------1602$ git fsck1603dangling commit 7281251ddd2a61e38657c827739c57015671a6b31604dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a631605dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b51606dangling blob 218761f9d90712d37a9c5e36f406f92202db07eb1607dangling commit bf093535a34a4d35731aa2bd90fe6b176302f14f1608dangling commit 8e4bec7f2ddaa268bef999853c25755452100f8e1609dangling tree d50bb86186bf27b681d25af89d3b5b68382e40851610dangling tree b24c2473f1fd3d91352a624795be026d64c8841f1611...1612-------------------------------------------------16131614You will see informational messages on dangling objects. They are objects1615that still exist in the repository but are no longer referenced by any of1616your branches, and can (and will) be removed after a while with "gc".1617You can run `git fsck --no-dangling` to suppress these messages, and still1618view real errors.16191620[[recovering-lost-changes]]1621Recovering lost changes1622~~~~~~~~~~~~~~~~~~~~~~~16231624[[reflogs]]1625Reflogs1626^^^^^^^16271628Say you modify a branch with +linkgit:git-reset[1] \--hard+, and then1629realize that the branch was the only reference you had to that point in1630history.16311632Fortunately, git also keeps a log, called a "reflog", of all the1633previous values of each branch. So in this case you can still find the1634old history using, for example,16351636-------------------------------------------------1637$ git log master@{1}1638-------------------------------------------------16391640This lists the commits reachable from the previous version of the1641"master" branch head. This syntax can be used with any git command1642that accepts a commit, not just with git log. Some other examples:16431644-------------------------------------------------1645$ git show master@{2} # See where the branch pointed 2,1646$ git show master@{3} # 3, ... changes ago.1647$ gitk master@{yesterday} # See where it pointed yesterday,1648$ gitk master@{"1 week ago"} # ... or last week1649$ git log --walk-reflogs master # show reflog entries for master1650-------------------------------------------------16511652A separate reflog is kept for the HEAD, so16531654-------------------------------------------------1655$ git show HEAD@{"1 week ago"}1656-------------------------------------------------16571658will show what HEAD pointed to one week ago, not what the current branch1659pointed to one week ago. This allows you to see the history of what1660you've checked out.16611662The reflogs are kept by default for 30 days, after which they may be1663pruned. See linkgit:git-reflog[1] and linkgit:git-gc[1] to learn1664how to control this pruning, and see the "SPECIFYING REVISIONS"1665section of linkgit:gitrevisions[7] for details.16661667Note that the reflog history is very different from normal git history.1668While normal history is shared by every repository that works on the1669same project, the reflog history is not shared: it tells you only about1670how the branches in your local repository have changed over time.16711672[[dangling-object-recovery]]1673Examining dangling objects1674^^^^^^^^^^^^^^^^^^^^^^^^^^16751676In some situations the reflog may not be able to save you. For example,1677suppose you delete a branch, then realize you need the history it1678contained. The reflog is also deleted; however, if you have not yet1679pruned the repository, then you may still be able to find the lost1680commits in the dangling objects that `git fsck` reports. See1681<<dangling-objects>> for the details.16821683-------------------------------------------------1684$ git fsck1685dangling commit 7281251ddd2a61e38657c827739c57015671a6b31686dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a631687dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b51688...1689-------------------------------------------------16901691You can examine1692one of those dangling commits with, for example,16931694------------------------------------------------1695$ gitk 7281251ddd --not --all1696------------------------------------------------16971698which does what it sounds like: it says that you want to see the commit1699history that is described by the dangling commit(s), but not the1700history that is described by all your existing branches and tags. Thus1701you get exactly the history reachable from that commit that is lost.1702(And notice that it might not be just one commit: we only report the1703"tip of the line" as being dangling, but there might be a whole deep1704and complex commit history that was dropped.)17051706If you decide you want the history back, you can always create a new1707reference pointing to it, for example, a new branch:17081709------------------------------------------------1710$ git branch recovered-branch 7281251ddd1711------------------------------------------------17121713Other types of dangling objects (blobs and trees) are also possible, and1714dangling objects can arise in other situations.171517161717[[sharing-development]]1718Sharing development with others1719===============================17201721[[getting-updates-With-git-pull]]1722Getting updates with git pull1723-----------------------------17241725After you clone a repository and commit a few changes of your own, you1726may wish to check the original repository for updates and merge them1727into your own work.17281729We have already seen <<Updating-a-repository-With-git-fetch,how to1730keep remote-tracking branches up to date>> with linkgit:git-fetch[1],1731and how to merge two branches. So you can merge in changes from the1732original repository's master branch with:17331734-------------------------------------------------1735$ git fetch1736$ git merge origin/master1737-------------------------------------------------17381739However, the linkgit:git-pull[1] command provides a way to do this in1740one step:17411742-------------------------------------------------1743$ git pull origin master1744-------------------------------------------------17451746In fact, if you have "master" checked out, then this branch has been1747configured by "git clone" to get changes from the HEAD branch of the1748origin repository. So often you can1749accomplish the above with just a simple17501751-------------------------------------------------1752$ git pull1753-------------------------------------------------17541755This command will fetch changes from the remote branches to your1756remote-tracking branches `origin/*`, and merge the default branch into1757the current branch.17581759More generally, a branch that is created from a remote-tracking branch1760will pull1761by default from that branch. See the descriptions of the1762branch.<name>.remote and branch.<name>.merge options in1763linkgit:git-config[1], and the discussion of the `--track` option in1764linkgit:git-checkout[1], to learn how to control these defaults.17651766In addition to saving you keystrokes, "git pull" also helps you by1767producing a default commit message documenting the branch and1768repository that you pulled from.17691770(But note that no such commit will be created in the case of a1771<<fast-forwards,fast-forward>>; instead, your branch will just be1772updated to point to the latest commit from the upstream branch.)17731774The `git pull` command can also be given "." as the "remote" repository,1775in which case it just merges in a branch from the current repository; so1776the commands17771778-------------------------------------------------1779$ git pull . branch1780$ git merge branch1781-------------------------------------------------17821783are roughly equivalent. The former is actually very commonly used.17841785[[submitting-patches]]1786Submitting patches to a project1787-------------------------------17881789If you just have a few changes, the simplest way to submit them may1790just be to send them as patches in email:17911792First, use linkgit:git-format-patch[1]; for example:17931794-------------------------------------------------1795$ git format-patch origin1796-------------------------------------------------17971798will produce a numbered series of files in the current directory, one1799for each patch in the current branch but not in origin/HEAD.18001801`git format-patch` can include an initial "cover letter". You can insert1802commentary on individual patches after the three dash line which1803`format-patch` places after the commit message but before the patch1804itself. If you use `git notes` to track your cover letter material,1805`git format-patch --notes` will include the commit's notes in a similar1806manner.18071808You can then import these into your mail client and send them by1809hand. However, if you have a lot to send at once, you may prefer to1810use the linkgit:git-send-email[1] script to automate the process.1811Consult the mailing list for your project first to determine how they1812prefer such patches be handled.18131814[[importing-patches]]1815Importing patches to a project1816------------------------------18171818Git also provides a tool called linkgit:git-am[1] (am stands for1819"apply mailbox"), for importing such an emailed series of patches.1820Just save all of the patch-containing messages, in order, into a1821single mailbox file, say "patches.mbox", then run18221823-------------------------------------------------1824$ git am -3 patches.mbox1825-------------------------------------------------18261827Git will apply each patch in order; if any conflicts are found, it1828will stop, and you can fix the conflicts as described in1829"<<resolving-a-merge,Resolving a merge>>". (The "-3" option tells1830git to perform a merge; if you would prefer it just to abort and1831leave your tree and index untouched, you may omit that option.)18321833Once the index is updated with the results of the conflict1834resolution, instead of creating a new commit, just run18351836-------------------------------------------------1837$ git am --resolved1838-------------------------------------------------18391840and git will create the commit for you and continue applying the1841remaining patches from the mailbox.18421843The final result will be a series of commits, one for each patch in1844the original mailbox, with authorship and commit log message each1845taken from the message containing each patch.18461847[[public-repositories]]1848Public git repositories1849-----------------------18501851Another way to submit changes to a project is to tell the maintainer1852of that project to pull the changes from your repository using1853linkgit:git-pull[1]. In the section "<<getting-updates-With-git-pull,1854Getting updates with `git pull`>>" we described this as a way to get1855updates from the "main" repository, but it works just as well in the1856other direction.18571858If you and the maintainer both have accounts on the same machine, then1859you can just pull changes from each other's repositories directly;1860commands that accept repository URLs as arguments will also accept a1861local directory name:18621863-------------------------------------------------1864$ git clone /path/to/repository1865$ git pull /path/to/other/repository1866-------------------------------------------------18671868or an ssh URL:18691870-------------------------------------------------1871$ git clone ssh://yourhost/~you/repository1872-------------------------------------------------18731874For projects with few developers, or for synchronizing a few private1875repositories, this may be all you need.18761877However, the more common way to do this is to maintain a separate public1878repository (usually on a different host) for others to pull changes1879from. This is usually more convenient, and allows you to cleanly1880separate private work in progress from publicly visible work.18811882You will continue to do your day-to-day work in your personal1883repository, but periodically "push" changes from your personal1884repository into your public repository, allowing other developers to1885pull from that repository. So the flow of changes, in a situation1886where there is one other developer with a public repository, looks1887like this:18881889 you push1890 your personal repo ------------------> your public repo1891 ^ |1892 | |1893 | you pull | they pull1894 | |1895 | |1896 | they push V1897 their public repo <------------------- their repo18981899We explain how to do this in the following sections.19001901[[setting-up-a-public-repository]]1902Setting up a public repository1903~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~19041905Assume your personal repository is in the directory ~/proj. We1906first create a new clone of the repository and tell `git daemon` that it1907is meant to be public:19081909-------------------------------------------------1910$ git clone --bare ~/proj proj.git1911$ touch proj.git/git-daemon-export-ok1912-------------------------------------------------19131914The resulting directory proj.git contains a "bare" git repository--it is1915just the contents of the ".git" directory, without any files checked out1916around it.19171918Next, copy proj.git to the server where you plan to host the1919public repository. You can use scp, rsync, or whatever is most1920convenient.19211922[[exporting-via-git]]1923Exporting a git repository via the git protocol1924~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~19251926This is the preferred method.19271928If someone else administers the server, they should tell you what1929directory to put the repository in, and what git:// URL it will appear1930at. You can then skip to the section1931"<<pushing-changes-to-a-public-repository,Pushing changes to a public1932repository>>", below.19331934Otherwise, all you need to do is start linkgit:git-daemon[1]; it will1935listen on port 9418. By default, it will allow access to any directory1936that looks like a git directory and contains the magic file1937git-daemon-export-ok. Passing some directory paths as `git daemon`1938arguments will further restrict the exports to those paths.19391940You can also run `git daemon` as an inetd service; see the1941linkgit:git-daemon[1] man page for details. (See especially the1942examples section.)19431944[[exporting-via-http]]1945Exporting a git repository via HTTP1946~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~19471948The git protocol gives better performance and reliability, but on a1949host with a web server set up, HTTP exports may be simpler to set up.19501951All you need to do is place the newly created bare git repository in1952a directory that is exported by the web server, and make some1953adjustments to give web clients some extra information they need:19541955-------------------------------------------------1956$ mv proj.git /home/you/public_html/proj.git1957$ cd proj.git1958$ git --bare update-server-info1959$ mv hooks/post-update.sample hooks/post-update1960-------------------------------------------------19611962(For an explanation of the last two lines, see1963linkgit:git-update-server-info[1] and linkgit:githooks[5].)19641965Advertise the URL of proj.git. Anybody else should then be able to1966clone or pull from that URL, for example with a command line like:19671968-------------------------------------------------1969$ git clone http://yourserver.com/~you/proj.git1970-------------------------------------------------19711972(See also1973link:howto/setup-git-server-over-http.txt[setup-git-server-over-http]1974for a slightly more sophisticated setup using WebDAV which also1975allows pushing over HTTP.)19761977[[pushing-changes-to-a-public-repository]]1978Pushing changes to a public repository1979~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~19801981Note that the two techniques outlined above (exporting via1982<<exporting-via-http,http>> or <<exporting-via-git,git>>) allow other1983maintainers to fetch your latest changes, but they do not allow write1984access, which you will need to update the public repository with the1985latest changes created in your private repository.19861987The simplest way to do this is using linkgit:git-push[1] and ssh; to1988update the remote branch named "master" with the latest state of your1989branch named "master", run19901991-------------------------------------------------1992$ git push ssh://yourserver.com/~you/proj.git master:master1993-------------------------------------------------19941995or just19961997-------------------------------------------------1998$ git push ssh://yourserver.com/~you/proj.git master1999-------------------------------------------------20002001As with `git fetch`, `git push` will complain if this does not result in a2002<<fast-forwards,fast-forward>>; see the following section for details on2003handling this case.20042005Note that the target of a "push" is normally a2006<<def_bare_repository,bare>> repository. You can also push to a2007repository that has a checked-out working tree, but the working tree2008will not be updated by the push. This may lead to unexpected results if2009the branch you push to is the currently checked-out branch!20102011As with `git fetch`, you may also set up configuration options to2012save typing; so, for example:20132014-------------------------------------------------2015$ git remote add public-repo ssh://yourserver.com/~you/proj.git2016-------------------------------------------------20172018adds the following to `.git/config`:20192020-------------------------------------------------2021[remote "public-repo"]2022 url = yourserver.com:proj.git2023 fetch = +refs/heads/*:refs/remotes/example/*2024-------------------------------------------------20252026which lets you do the same push with just20272028-------------------------------------------------2029$ git push public-repo master2030-------------------------------------------------20312032See the explanations of the remote.<name>.url, branch.<name>.remote,2033and remote.<name>.push options in linkgit:git-config[1] for2034details.20352036[[forcing-push]]2037What to do when a push fails2038~~~~~~~~~~~~~~~~~~~~~~~~~~~~20392040If a push would not result in a <<fast-forwards,fast-forward>> of the2041remote branch, then it will fail with an error like:20422043-------------------------------------------------2044error: remote 'refs/heads/master' is not an ancestor of2045 local 'refs/heads/master'.2046 Maybe you are not up-to-date and need to pull first?2047error: failed to push to 'ssh://yourserver.com/~you/proj.git'2048-------------------------------------------------20492050This can happen, for example, if you:20512052 - use `git reset --hard` to remove already-published commits, or2053 - use `git commit --amend` to replace already-published commits2054 (as in <<fixing-a-mistake-by-rewriting-history>>), or2055 - use `git rebase` to rebase any already-published commits (as2056 in <<using-git-rebase>>).20572058You may force `git push` to perform the update anyway by preceding the2059branch name with a plus sign:20602061-------------------------------------------------2062$ git push ssh://yourserver.com/~you/proj.git +master2063-------------------------------------------------20642065Note the addition of the `+` sign. Alternatively, you can use the2066`-f` flag to force the remote update, as in:20672068-------------------------------------------------2069$ git push -f ssh://yourserver.com/~you/proj.git master2070-------------------------------------------------20712072Normally whenever a branch head in a public repository is modified, it2073is modified to point to a descendant of the commit that it pointed to2074before. By forcing a push in this situation, you break that convention.2075(See <<problems-With-rewriting-history>>.)20762077Nevertheless, this is a common practice for people that need a simple2078way to publish a work-in-progress patch series, and it is an acceptable2079compromise as long as you warn other developers that this is how you2080intend to manage the branch.20812082It's also possible for a push to fail in this way when other people have2083the right to push to the same repository. In that case, the correct2084solution is to retry the push after first updating your work: either by a2085pull, or by a fetch followed by a rebase; see the2086<<setting-up-a-shared-repository,next section>> and2087linkgit:gitcvs-migration[7] for more.20882089[[setting-up-a-shared-repository]]2090Setting up a shared repository2091~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~20922093Another way to collaborate is by using a model similar to that2094commonly used in CVS, where several developers with special rights2095all push to and pull from a single shared repository. See2096linkgit:gitcvs-migration[7] for instructions on how to2097set this up.20982099However, while there is nothing wrong with git's support for shared2100repositories, this mode of operation is not generally recommended,2101simply because the mode of collaboration that git supports--by2102exchanging patches and pulling from public repositories--has so many2103advantages over the central shared repository:21042105 - Git's ability to quickly import and merge patches allows a2106 single maintainer to process incoming changes even at very2107 high rates. And when that becomes too much, `git pull` provides2108 an easy way for that maintainer to delegate this job to other2109 maintainers while still allowing optional review of incoming2110 changes.2111 - Since every developer's repository has the same complete copy2112 of the project history, no repository is special, and it is2113 trivial for another developer to take over maintenance of a2114 project, either by mutual agreement, or because a maintainer2115 becomes unresponsive or difficult to work with.2116 - The lack of a central group of "committers" means there is2117 less need for formal decisions about who is "in" and who is2118 "out".21192120[[setting-up-gitweb]]2121Allowing web browsing of a repository2122~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~21232124The gitweb cgi script provides users an easy way to browse your2125project's files and history without having to install git; see the file2126gitweb/INSTALL in the git source tree for instructions on setting it up.21272128[[sharing-development-examples]]2129Examples2130--------21312132[[maintaining-topic-branches]]2133Maintaining topic branches for a Linux subsystem maintainer2134~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~21352136This describes how Tony Luck uses git in his role as maintainer of the2137IA64 architecture for the Linux kernel.21382139He uses two public branches:21402141 - A "test" tree into which patches are initially placed so that they2142 can get some exposure when integrated with other ongoing development.2143 This tree is available to Andrew for pulling into -mm whenever he2144 wants.21452146 - A "release" tree into which tested patches are moved for final sanity2147 checking, and as a vehicle to send them upstream to Linus (by sending2148 him a "please pull" request.)21492150He also uses a set of temporary branches ("topic branches"), each2151containing a logical grouping of patches.21522153To set this up, first create your work tree by cloning Linus's public2154tree:21552156-------------------------------------------------2157$ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git work2158$ cd work2159-------------------------------------------------21602161Linus's tree will be stored in the remote-tracking branch named origin/master,2162and can be updated using linkgit:git-fetch[1]; you can track other2163public trees using linkgit:git-remote[1] to set up a "remote" and2164linkgit:git-fetch[1] to keep them up-to-date; see2165<<repositories-and-branches>>.21662167Now create the branches in which you are going to work; these start out2168at the current tip of origin/master branch, and should be set up (using2169the --track option to linkgit:git-branch[1]) to merge changes in from2170Linus by default.21712172-------------------------------------------------2173$ git branch --track test origin/master2174$ git branch --track release origin/master2175-------------------------------------------------21762177These can be easily kept up to date using linkgit:git-pull[1].21782179-------------------------------------------------2180$ git checkout test && git pull2181$ git checkout release && git pull2182-------------------------------------------------21832184Important note! If you have any local changes in these branches, then2185this merge will create a commit object in the history (with no local2186changes git will simply do a "fast-forward" merge). Many people dislike2187the "noise" that this creates in the Linux history, so you should avoid2188doing this capriciously in the "release" branch, as these noisy commits2189will become part of the permanent history when you ask Linus to pull2190from the release branch.21912192A few configuration variables (see linkgit:git-config[1]) can2193make it easy to push both branches to your public tree. (See2194<<setting-up-a-public-repository>>.)21952196-------------------------------------------------2197$ cat >> .git/config <<EOF2198[remote "mytree"]2199 url = master.kernel.org:/pub/scm/linux/kernel/git/aegl/linux-2.6.git2200 push = release2201 push = test2202EOF2203-------------------------------------------------22042205Then you can push both the test and release trees using2206linkgit:git-push[1]:22072208-------------------------------------------------2209$ git push mytree2210-------------------------------------------------22112212or push just one of the test and release branches using:22132214-------------------------------------------------2215$ git push mytree test2216-------------------------------------------------22172218or22192220-------------------------------------------------2221$ git push mytree release2222-------------------------------------------------22232224Now to apply some patches from the community. Think of a short2225snappy name for a branch to hold this patch (or related group of2226patches), and create a new branch from a recent stable tag of2227Linus's branch. Picking a stable base for your branch will:22281) help you: by avoiding inclusion of unrelated and perhaps lightly2229tested changes22302) help future bug hunters that use "git bisect" to find problems22312232-------------------------------------------------2233$ git checkout -b speed-up-spinlocks v2.6.352234-------------------------------------------------22352236Now you apply the patch(es), run some tests, and commit the change(s). If2237the patch is a multi-part series, then you should apply each as a separate2238commit to this branch.22392240-------------------------------------------------2241$ ... patch ... test ... commit [ ... patch ... test ... commit ]*2242-------------------------------------------------22432244When you are happy with the state of this change, you can pull it into the2245"test" branch in preparation to make it public:22462247-------------------------------------------------2248$ git checkout test && git pull . speed-up-spinlocks2249-------------------------------------------------22502251It is unlikely that you would have any conflicts here ... but you might if you2252spent a while on this step and had also pulled new versions from upstream.22532254Some time later when enough time has passed and testing done, you can pull the2255same branch into the "release" tree ready to go upstream. This is where you2256see the value of keeping each patch (or patch series) in its own branch. It2257means that the patches can be moved into the "release" tree in any order.22582259-------------------------------------------------2260$ git checkout release && git pull . speed-up-spinlocks2261-------------------------------------------------22622263After a while, you will have a number of branches, and despite the2264well chosen names you picked for each of them, you may forget what2265they are for, or what status they are in. To get a reminder of what2266changes are in a specific branch, use:22672268-------------------------------------------------2269$ git log linux..branchname | git shortlog2270-------------------------------------------------22712272To see whether it has already been merged into the test or release branches,2273use:22742275-------------------------------------------------2276$ git log test..branchname2277-------------------------------------------------22782279or22802281-------------------------------------------------2282$ git log release..branchname2283-------------------------------------------------22842285(If this branch has not yet been merged, you will see some log entries.2286If it has been merged, then there will be no output.)22872288Once a patch completes the great cycle (moving from test to release,2289then pulled by Linus, and finally coming back into your local2290"origin/master" branch), the branch for this change is no longer needed.2291You detect this when the output from:22922293-------------------------------------------------2294$ git log origin..branchname2295-------------------------------------------------22962297is empty. At this point the branch can be deleted:22982299-------------------------------------------------2300$ git branch -d branchname2301-------------------------------------------------23022303Some changes are so trivial that it is not necessary to create a separate2304branch and then merge into each of the test and release branches. For2305these changes, just apply directly to the "release" branch, and then2306merge that into the "test" branch.23072308To create diffstat and shortlog summaries of changes to include in a "please2309pull" request to Linus you can use:23102311-------------------------------------------------2312$ git diff --stat origin..release2313-------------------------------------------------23142315and23162317-------------------------------------------------2318$ git log -p origin..release | git shortlog2319-------------------------------------------------23202321Here are some of the scripts that simplify all this even further.23222323-------------------------------------------------2324==== update script ====2325# Update a branch in my GIT tree. If the branch to be updated2326# is origin, then pull from kernel.org. Otherwise merge2327# origin/master branch into test|release branch23282329case "$1" in2330test|release)2331 git checkout $1 && git pull . origin2332 ;;2333origin)2334 before=$(git rev-parse refs/remotes/origin/master)2335 git fetch origin2336 after=$(git rev-parse refs/remotes/origin/master)2337 if [ $before != $after ]2338 then2339 git log $before..$after | git shortlog2340 fi2341 ;;2342*)2343 echo "Usage: $0 origin|test|release" 1>&22344 exit 12345 ;;2346esac2347-------------------------------------------------23482349-------------------------------------------------2350==== merge script ====2351# Merge a branch into either the test or release branch23522353pname=$023542355usage()2356{2357 echo "Usage: $pname branch test|release" 1>&22358 exit 12359}23602361git show-ref -q --verify -- refs/heads/"$1" || {2362 echo "Can't see branch <$1>" 1>&22363 usage2364}23652366case "$2" in2367test|release)2368 if [ $(git log $2..$1 | wc -c) -eq 0 ]2369 then2370 echo $1 already merged into $2 1>&22371 exit 12372 fi2373 git checkout $2 && git pull . $12374 ;;2375*)2376 usage2377 ;;2378esac2379-------------------------------------------------23802381-------------------------------------------------2382==== status script ====2383# report on status of my ia64 GIT tree23842385gb=$(tput setab 2)2386rb=$(tput setab 1)2387restore=$(tput setab 9)23882389if [ `git rev-list test..release | wc -c` -gt 0 ]2390then2391 echo $rb Warning: commits in release that are not in test $restore2392 git log test..release2393fi23942395for branch in `git show-ref --heads | sed 's|^.*/||'`2396do2397 if [ $branch = test -o $branch = release ]2398 then2399 continue2400 fi24012402 echo -n $gb ======= $branch ====== $restore " "2403 status=2404 for ref in test release origin/master2405 do2406 if [ `git rev-list $ref..$branch | wc -c` -gt 0 ]2407 then2408 status=$status${ref:0:1}2409 fi2410 done2411 case $status in2412 trl)2413 echo $rb Need to pull into test $restore2414 ;;2415 rl)2416 echo "In test"2417 ;;2418 l)2419 echo "Waiting for linus"2420 ;;2421 "")2422 echo $rb All done $restore2423 ;;2424 *)2425 echo $rb "<$status>" $restore2426 ;;2427 esac2428 git log origin/master..$branch | git shortlog2429done2430-------------------------------------------------243124322433[[cleaning-up-history]]2434Rewriting history and maintaining patch series2435==============================================24362437Normally commits are only added to a project, never taken away or2438replaced. Git is designed with this assumption, and violating it will2439cause git's merge machinery (for example) to do the wrong thing.24402441However, there is a situation in which it can be useful to violate this2442assumption.24432444[[patch-series]]2445Creating the perfect patch series2446---------------------------------24472448Suppose you are a contributor to a large project, and you want to add a2449complicated feature, and to present it to the other developers in a way2450that makes it easy for them to read your changes, verify that they are2451correct, and understand why you made each change.24522453If you present all of your changes as a single patch (or commit), they2454may find that it is too much to digest all at once.24552456If you present them with the entire history of your work, complete with2457mistakes, corrections, and dead ends, they may be overwhelmed.24582459So the ideal is usually to produce a series of patches such that:24602461 1. Each patch can be applied in order.24622463 2. Each patch includes a single logical change, together with a2464 message explaining the change.24652466 3. No patch introduces a regression: after applying any initial2467 part of the series, the resulting project still compiles and2468 works, and has no bugs that it didn't have before.24692470 4. The complete series produces the same end result as your own2471 (probably much messier!) development process did.24722473We will introduce some tools that can help you do this, explain how to2474use them, and then explain some of the problems that can arise because2475you are rewriting history.24762477[[using-git-rebase]]2478Keeping a patch series up to date using git rebase2479--------------------------------------------------24802481Suppose that you create a branch "mywork" on a remote-tracking branch2482"origin", and create some commits on top of it:24832484-------------------------------------------------2485$ git checkout -b mywork origin2486$ vi file.txt2487$ git commit2488$ vi otherfile.txt2489$ git commit2490...2491-------------------------------------------------24922493You have performed no merges into mywork, so it is just a simple linear2494sequence of patches on top of "origin":24952496................................................2497 o--o--O <-- origin2498 \2499 a--b--c <-- mywork2500................................................25012502Some more interesting work has been done in the upstream project, and2503"origin" has advanced:25042505................................................2506 o--o--O--o--o--o <-- origin2507 \2508 a--b--c <-- mywork2509................................................25102511At this point, you could use "pull" to merge your changes back in;2512the result would create a new merge commit, like this:25132514................................................2515 o--o--O--o--o--o <-- origin2516 \ \2517 a--b--c--m <-- mywork2518................................................25192520However, if you prefer to keep the history in mywork a simple series of2521commits without any merges, you may instead choose to use2522linkgit:git-rebase[1]:25232524-------------------------------------------------2525$ git checkout mywork2526$ git rebase origin2527-------------------------------------------------25282529This will remove each of your commits from mywork, temporarily saving2530them as patches (in a directory named ".git/rebase-apply"), update mywork to2531point at the latest version of origin, then apply each of the saved2532patches to the new mywork. The result will look like:253325342535................................................2536 o--o--O--o--o--o <-- origin2537 \2538 a'--b'--c' <-- mywork2539................................................25402541In the process, it may discover conflicts. In that case it will stop2542and allow you to fix the conflicts; after fixing conflicts, use `git add`2543to update the index with those contents, and then, instead of2544running `git commit`, just run25452546-------------------------------------------------2547$ git rebase --continue2548-------------------------------------------------25492550and git will continue applying the rest of the patches.25512552At any point you may use the `--abort` option to abort this process and2553return mywork to the state it had before you started the rebase:25542555-------------------------------------------------2556$ git rebase --abort2557-------------------------------------------------25582559If you need to reorder or edit a number of commits in a branch, it may2560be easier to use `git rebase -i`, which allows you to reorder and2561squash commits, as well as marking them for individual editing during2562the rebase. See <<interactive-rebase>> for details, and2563<<reordering-patch-series>> for alternatives.25642565[[rewriting-one-commit]]2566Rewriting a single commit2567-------------------------25682569We saw in <<fixing-a-mistake-by-rewriting-history>> that you can replace the2570most recent commit using25712572-------------------------------------------------2573$ git commit --amend2574-------------------------------------------------25752576which will replace the old commit by a new commit incorporating your2577changes, giving you a chance to edit the old commit message first.2578This is useful for fixing typos in your last commit, or for adjusting2579the patch contents of a poorly staged commit.25802581If you need to amend commits from deeper in your history, you can2582use <<interactive-rebase,interactive rebase's `edit` instruction>>.25832584[[reordering-patch-series]]2585Reordering or selecting from a patch series2586-------------------------------------------25872588Sometimes you want to edit a commit deeper in your history. One2589approach is to use `git format-patch` to create a series of patches2590and then reset the state to before the patches:25912592-------------------------------------------------2593$ git format-patch origin2594$ git reset --hard origin2595-------------------------------------------------25962597Then modify, reorder, or eliminate patches as needed before applying2598them again with linkgit:git-am[1]:25992600-------------------------------------------------2601$ git am *.patch2602-------------------------------------------------26032604[[interactive-rebase]]2605Using interactive rebases2606-------------------------26072608You can also edit a patch series with an interactive rebase. This is2609the same as <<reordering-patch-series,reordering a patch series using2610`format-patch`>>, so use whichever interface you like best.26112612Rebase your current HEAD on the last commit you want to retain as-is.2613For example, if you want to reorder the last 5 commits, use:26142615-------------------------------------------------2616$ git rebase -i HEAD~52617-------------------------------------------------26182619This will open your editor with a list of steps to be taken to perform2620your rebase.26212622-------------------------------------------------2623pick deadbee The oneline of this commit2624pick fa1afe1 The oneline of the next commit2625...26262627# Rebase c0ffeee..deadbee onto c0ffeee2628#2629# Commands:2630# p, pick = use commit2631# r, reword = use commit, but edit the commit message2632# e, edit = use commit, but stop for amending2633# s, squash = use commit, but meld into previous commit2634# f, fixup = like "squash", but discard this commit's log message2635# x, exec = run command (the rest of the line) using shell2636#2637# These lines can be re-ordered; they are executed from top to bottom.2638#2639# If you remove a line here THAT COMMIT WILL BE LOST.2640#2641# However, if you remove everything, the rebase will be aborted.2642#2643# Note that empty commits are commented out2644-------------------------------------------------26452646As explained in the comments, you can reorder commits, squash them2647together, edit commit messages, etc. by editing the list. Once you2648are satisfied, save the list and close your editor, and the rebase2649will begin.26502651The rebase will stop where `pick` has been replaced with `edit` or2652when a step in the list fails to mechanically resolve conflicts and2653needs your help. When you are done editing and/or resolving conflicts2654you can continue with `git rebase --continue`. If you decide that2655things are getting too hairy, you can always bail out with `git rebase2656--abort`. Even after the rebase is complete, you can still recover2657the original branch by using the <<reflogs,reflog>>.26582659For a more detailed discussion of the procedure and additional tips,2660see the "INTERACTIVE MODE" section of linkgit:git-rebase[1].26612662[[patch-series-tools]]2663Other tools2664-----------26652666There are numerous other tools, such as StGit, which exist for the2667purpose of maintaining a patch series. These are outside of the scope of2668this manual.26692670[[problems-With-rewriting-history]]2671Problems with rewriting history2672-------------------------------26732674The primary problem with rewriting the history of a branch has to do2675with merging. Suppose somebody fetches your branch and merges it into2676their branch, with a result something like this:26772678................................................2679 o--o--O--o--o--o <-- origin2680 \ \2681 t--t--t--m <-- their branch:2682................................................26832684Then suppose you modify the last three commits:26852686................................................2687 o--o--o <-- new head of origin2688 /2689 o--o--O--o--o--o <-- old head of origin2690................................................26912692If we examined all this history together in one repository, it will2693look like:26942695................................................2696 o--o--o <-- new head of origin2697 /2698 o--o--O--o--o--o <-- old head of origin2699 \ \2700 t--t--t--m <-- their branch:2701................................................27022703Git has no way of knowing that the new head is an updated version of2704the old head; it treats this situation exactly the same as it would if2705two developers had independently done the work on the old and new heads2706in parallel. At this point, if someone attempts to merge the new head2707in to their branch, git will attempt to merge together the two (old and2708new) lines of development, instead of trying to replace the old by the2709new. The results are likely to be unexpected.27102711You may still choose to publish branches whose history is rewritten,2712and it may be useful for others to be able to fetch those branches in2713order to examine or test them, but they should not attempt to pull such2714branches into their own work.27152716For true distributed development that supports proper merging,2717published branches should never be rewritten.27182719[[bisect-merges]]2720Why bisecting merge commits can be harder than bisecting linear history2721-----------------------------------------------------------------------27222723The linkgit:git-bisect[1] command correctly handles history that2724includes merge commits. However, when the commit that it finds is a2725merge commit, the user may need to work harder than usual to figure out2726why that commit introduced a problem.27272728Imagine this history:27292730................................................2731 ---Z---o---X---...---o---A---C---D2732 \ /2733 o---o---Y---...---o---B2734................................................27352736Suppose that on the upper line of development, the meaning of one2737of the functions that exists at Z is changed at commit X. The2738commits from Z leading to A change both the function's2739implementation and all calling sites that exist at Z, as well2740as new calling sites they add, to be consistent. There is no2741bug at A.27422743Suppose that in the meantime on the lower line of development somebody2744adds a new calling site for that function at commit Y. The2745commits from Z leading to B all assume the old semantics of that2746function and the callers and the callee are consistent with each2747other. There is no bug at B, either.27482749Suppose further that the two development lines merge cleanly at C,2750so no conflict resolution is required.27512752Nevertheless, the code at C is broken, because the callers added2753on the lower line of development have not been converted to the new2754semantics introduced on the upper line of development. So if all2755you know is that D is bad, that Z is good, and that2756linkgit:git-bisect[1] identifies C as the culprit, how will you2757figure out that the problem is due to this change in semantics?27582759When the result of a `git bisect` is a non-merge commit, you should2760normally be able to discover the problem by examining just that commit.2761Developers can make this easy by breaking their changes into small2762self-contained commits. That won't help in the case above, however,2763because the problem isn't obvious from examination of any single2764commit; instead, a global view of the development is required. To2765make matters worse, the change in semantics in the problematic2766function may be just one small part of the changes in the upper2767line of development.27682769On the other hand, if instead of merging at C you had rebased the2770history between Z to B on top of A, you would have gotten this2771linear history:27722773................................................................2774 ---Z---o---X--...---o---A---o---o---Y*--...---o---B*--D*2775................................................................27762777Bisecting between Z and D* would hit a single culprit commit Y*,2778and understanding why Y* was broken would probably be easier.27792780Partly for this reason, many experienced git users, even when2781working on an otherwise merge-heavy project, keep the history2782linear by rebasing against the latest upstream version before2783publishing.27842785[[advanced-branch-management]]2786Advanced branch management2787==========================27882789[[fetching-individual-branches]]2790Fetching individual branches2791----------------------------27922793Instead of using linkgit:git-remote[1], you can also choose just2794to update one branch at a time, and to store it locally under an2795arbitrary name:27962797-------------------------------------------------2798$ git fetch origin todo:my-todo-work2799-------------------------------------------------28002801The first argument, "origin", just tells git to fetch from the2802repository you originally cloned from. The second argument tells git2803to fetch the branch named "todo" from the remote repository, and to2804store it locally under the name refs/heads/my-todo-work.28052806You can also fetch branches from other repositories; so28072808-------------------------------------------------2809$ git fetch git://example.com/proj.git master:example-master2810-------------------------------------------------28112812will create a new branch named "example-master" and store in it the2813branch named "master" from the repository at the given URL. If you2814already have a branch named example-master, it will attempt to2815<<fast-forwards,fast-forward>> to the commit given by example.com's2816master branch. In more detail:28172818[[fetch-fast-forwards]]2819git fetch and fast-forwards2820---------------------------28212822In the previous example, when updating an existing branch, "git fetch"2823checks to make sure that the most recent commit on the remote2824branch is a descendant of the most recent commit on your copy of the2825branch before updating your copy of the branch to point at the new2826commit. Git calls this process a <<fast-forwards,fast-forward>>.28272828A fast-forward looks something like this:28292830................................................2831 o--o--o--o <-- old head of the branch2832 \2833 o--o--o <-- new head of the branch2834................................................283528362837In some cases it is possible that the new head will *not* actually be2838a descendant of the old head. For example, the developer may have2839realized she made a serious mistake, and decided to backtrack,2840resulting in a situation like:28412842................................................2843 o--o--o--o--a--b <-- old head of the branch2844 \2845 o--o--o <-- new head of the branch2846................................................28472848In this case, "git fetch" will fail, and print out a warning.28492850In that case, you can still force git to update to the new head, as2851described in the following section. However, note that in the2852situation above this may mean losing the commits labeled "a" and "b",2853unless you've already created a reference of your own pointing to2854them.28552856[[forcing-fetch]]2857Forcing git fetch to do non-fast-forward updates2858------------------------------------------------28592860If git fetch fails because the new head of a branch is not a2861descendant of the old head, you may force the update with:28622863-------------------------------------------------2864$ git fetch git://example.com/proj.git +master:refs/remotes/example/master2865-------------------------------------------------28662867Note the addition of the "+" sign. Alternatively, you can use the "-f"2868flag to force updates of all the fetched branches, as in:28692870-------------------------------------------------2871$ git fetch -f origin2872-------------------------------------------------28732874Be aware that commits that the old version of example/master pointed at2875may be lost, as we saw in the previous section.28762877[[remote-branch-configuration]]2878Configuring remote-tracking branches2879------------------------------------28802881We saw above that "origin" is just a shortcut to refer to the2882repository that you originally cloned from. This information is2883stored in git configuration variables, which you can see using2884linkgit:git-config[1]:28852886-------------------------------------------------2887$ git config -l2888core.repositoryformatversion=02889core.filemode=true2890core.logallrefupdates=true2891remote.origin.url=git://git.kernel.org/pub/scm/git/git.git2892remote.origin.fetch=+refs/heads/*:refs/remotes/origin/*2893branch.master.remote=origin2894branch.master.merge=refs/heads/master2895-------------------------------------------------28962897If there are other repositories that you also use frequently, you can2898create similar configuration options to save typing; for example,28992900-------------------------------------------------2901$ git remote add example git://example.com/proj.git2902-------------------------------------------------29032904adds the following to `.git/config`:29052906-------------------------------------------------2907[remote "example"]2908 url = git://example.com/proj.git2909 fetch = +refs/heads/*:refs/remotes/example/*2910-------------------------------------------------29112912Also note that the above configuration can be performed by directly2913editing the file `.git/config` instead of using linkgit:git-remote[1].29142915After configuring the remote, the following three commands will do the2916same thing:29172918-------------------------------------------------2919$ git fetch git://example.com/proj.git +refs/heads/*:refs/remotes/example/*2920$ git fetch example +refs/heads/*:refs/remotes/example/*2921$ git fetch example2922-------------------------------------------------29232924See linkgit:git-config[1] for more details on the configuration2925options mentioned above and linkgit:git-fetch[1] for more details on2926the refspec syntax.292729282929[[git-concepts]]2930Git concepts2931============29322933Git is built on a small number of simple but powerful ideas. While it2934is possible to get things done without understanding them, you will find2935git much more intuitive if you do.29362937We start with the most important, the <<def_object_database,object2938database>> and the <<def_index,index>>.29392940[[the-object-database]]2941The Object Database2942-------------------294329442945We already saw in <<understanding-commits>> that all commits are stored2946under a 40-digit "object name". In fact, all the information needed to2947represent the history of a project is stored in objects with such names.2948In each case the name is calculated by taking the SHA-1 hash of the2949contents of the object. The SHA-1 hash is a cryptographic hash function.2950What that means to us is that it is impossible to find two different2951objects with the same name. This has a number of advantages; among2952others:29532954- Git can quickly determine whether two objects are identical or not,2955 just by comparing names.2956- Since object names are computed the same way in every repository, the2957 same content stored in two repositories will always be stored under2958 the same name.2959- Git can detect errors when it reads an object, by checking that the2960 object's name is still the SHA-1 hash of its contents.29612962(See <<object-details>> for the details of the object formatting and2963SHA-1 calculation.)29642965There are four different types of objects: "blob", "tree", "commit", and2966"tag".29672968- A <<def_blob_object,"blob" object>> is used to store file data.2969- A <<def_tree_object,"tree" object>> ties one or more2970 "blob" objects into a directory structure. In addition, a tree object2971 can refer to other tree objects, thus creating a directory hierarchy.2972- A <<def_commit_object,"commit" object>> ties such directory hierarchies2973 together into a <<def_DAG,directed acyclic graph>> of revisions--each2974 commit contains the object name of exactly one tree designating the2975 directory hierarchy at the time of the commit. In addition, a commit2976 refers to "parent" commit objects that describe the history of how we2977 arrived at that directory hierarchy.2978- A <<def_tag_object,"tag" object>> symbolically identifies and can be2979 used to sign other objects. It contains the object name and type of2980 another object, a symbolic name (of course!) and, optionally, a2981 signature.29822983The object types in some more detail:29842985[[commit-object]]2986Commit Object2987~~~~~~~~~~~~~29882989The "commit" object links a physical state of a tree with a description2990of how we got there and why. Use the --pretty=raw option to2991linkgit:git-show[1] or linkgit:git-log[1] to examine your favorite2992commit:29932994------------------------------------------------2995$ git show -s --pretty=raw 2be7fcb4762996commit 2be7fcb4764f2dbcee52635b91fedb1b3dcf7ab42997tree fb3a8bdd0ceddd019615af4d57a53f43d8cee2bf2998parent 257a84d9d02e90447b149af58b271c19405edb6a2999author Dave Watson <dwatson@mimvista.com> 1187576872 -04003000committer Junio C Hamano <gitster@pobox.com> 1187591163 -070030013002 Fix misspelling of 'suppress' in docs30033004 Signed-off-by: Junio C Hamano <gitster@pobox.com>3005------------------------------------------------30063007As you can see, a commit is defined by:30083009- a tree: The SHA-1 name of a tree object (as defined below), representing3010 the contents of a directory at a certain point in time.3011- parent(s): The SHA-1 name(s) of some number of commits which represent the3012 immediately previous step(s) in the history of the project. The3013 example above has one parent; merge commits may have more than3014 one. A commit with no parents is called a "root" commit, and3015 represents the initial revision of a project. Each project must have3016 at least one root. A project can also have multiple roots, though3017 that isn't common (or necessarily a good idea).3018- an author: The name of the person responsible for this change, together3019 with its date.3020- a committer: The name of the person who actually created the commit,3021 with the date it was done. This may be different from the author, for3022 example, if the author was someone who wrote a patch and emailed it3023 to the person who used it to create the commit.3024- a comment describing this commit.30253026Note that a commit does not itself contain any information about what3027actually changed; all changes are calculated by comparing the contents3028of the tree referred to by this commit with the trees associated with3029its parents. In particular, git does not attempt to record file renames3030explicitly, though it can identify cases where the existence of the same3031file data at changing paths suggests a rename. (See, for example, the3032-M option to linkgit:git-diff[1]).30333034A commit is usually created by linkgit:git-commit[1], which creates a3035commit whose parent is normally the current HEAD, and whose tree is3036taken from the content currently stored in the index.30373038[[tree-object]]3039Tree Object3040~~~~~~~~~~~30413042The ever-versatile linkgit:git-show[1] command can also be used to3043examine tree objects, but linkgit:git-ls-tree[1] will give you more3044details:30453046------------------------------------------------3047$ git ls-tree fb3a8bdd0ce3048100644 blob 63c918c667fa005ff12ad89437f2fdc80926e21c .gitignore3049100644 blob 5529b198e8d14decbe4ad99db3f7fb632de0439d .mailmap3050100644 blob 6ff87c4664981e4397625791c8ea3bbb5f2279a3 COPYING3051040000 tree 2fb783e477100ce076f6bf57e4a6f026013dc745 Documentation3052100755 blob 3c0032cec592a765692234f1cba47dfdcc3a9200 GIT-VERSION-GEN3053100644 blob 289b046a443c0647624607d471289b2c7dcd470b INSTALL3054100644 blob 4eb463797adc693dc168b926b6932ff53f17d0b1 Makefile3055100644 blob 548142c327a6790ff8821d67c2ee1eff7a656b52 README3056...3057------------------------------------------------30583059As you can see, a tree object contains a list of entries, each with a3060mode, object type, SHA-1 name, and name, sorted by name. It represents3061the contents of a single directory tree.30623063The object type may be a blob, representing the contents of a file, or3064another tree, representing the contents of a subdirectory. Since trees3065and blobs, like all other objects, are named by the SHA-1 hash of their3066contents, two trees have the same SHA-1 name if and only if their3067contents (including, recursively, the contents of all subdirectories)3068are identical. This allows git to quickly determine the differences3069between two related tree objects, since it can ignore any entries with3070identical object names.30713072(Note: in the presence of submodules, trees may also have commits as3073entries. See <<submodules>> for documentation.)30743075Note that the files all have mode 644 or 755: git actually only pays3076attention to the executable bit.30773078[[blob-object]]3079Blob Object3080~~~~~~~~~~~30813082You can use linkgit:git-show[1] to examine the contents of a blob; take,3083for example, the blob in the entry for "COPYING" from the tree above:30843085------------------------------------------------3086$ git show 6ff87c466430873088 Note that the only valid version of the GPL as far as this project3089 is concerned is _this_ particular version of the license (ie v2, not3090 v2.2 or v3.x or whatever), unless explicitly otherwise stated.3091...3092------------------------------------------------30933094A "blob" object is nothing but a binary blob of data. It doesn't refer3095to anything else or have attributes of any kind.30963097Since the blob is entirely defined by its data, if two files in a3098directory tree (or in multiple different versions of the repository)3099have the same contents, they will share the same blob object. The object3100is totally independent of its location in the directory tree, and3101renaming a file does not change the object that file is associated with.31023103Note that any tree or blob object can be examined using3104linkgit:git-show[1] with the <revision>:<path> syntax. This can3105sometimes be useful for browsing the contents of a tree that is not3106currently checked out.31073108[[trust]]3109Trust3110~~~~~31113112If you receive the SHA-1 name of a blob from one source, and its contents3113from another (possibly untrusted) source, you can still trust that those3114contents are correct as long as the SHA-1 name agrees. This is because3115the SHA-1 is designed so that it is infeasible to find different contents3116that produce the same hash.31173118Similarly, you need only trust the SHA-1 name of a top-level tree object3119to trust the contents of the entire directory that it refers to, and if3120you receive the SHA-1 name of a commit from a trusted source, then you3121can easily verify the entire history of commits reachable through3122parents of that commit, and all of those contents of the trees referred3123to by those commits.31243125So to introduce some real trust in the system, the only thing you need3126to do is to digitally sign just 'one' special note, which includes the3127name of a top-level commit. Your digital signature shows others3128that you trust that commit, and the immutability of the history of3129commits tells others that they can trust the whole history.31303131In other words, you can easily validate a whole archive by just3132sending out a single email that tells the people the name (SHA-1 hash)3133of the top commit, and digitally sign that email using something3134like GPG/PGP.31353136To assist in this, git also provides the tag object...31373138[[tag-object]]3139Tag Object3140~~~~~~~~~~31413142A tag object contains an object, object type, tag name, the name of the3143person ("tagger") who created the tag, and a message, which may contain3144a signature, as can be seen using linkgit:git-cat-file[1]:31453146------------------------------------------------3147$ git cat-file tag v1.5.03148object 437b1b20df4b356c9342dac8d38849f24ef44f273149type commit3150tag v1.5.03151tagger Junio C Hamano <junkio@cox.net> 1171411200 +000031523153GIT 1.5.03154-----BEGIN PGP SIGNATURE-----3155Version: GnuPG v1.4.6 (GNU/Linux)31563157iD8DBQBF0lGqwMbZpPMRm5oRAuRiAJ9ohBLd7s2kqjkKlq1qqC57SbnmzQCdG4ui3158nLE/L9aUXdWeTFPron96DLA=3159=2E+03160-----END PGP SIGNATURE-----3161------------------------------------------------31623163See the linkgit:git-tag[1] command to learn how to create and verify tag3164objects. (Note that linkgit:git-tag[1] can also be used to create3165"lightweight tags", which are not tag objects at all, but just simple3166references whose names begin with "refs/tags/").31673168[[pack-files]]3169How git stores objects efficiently: pack files3170~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~31713172Newly created objects are initially created in a file named after the3173object's SHA-1 hash (stored in .git/objects).31743175Unfortunately this system becomes inefficient once a project has a3176lot of objects. Try this on an old project:31773178------------------------------------------------3179$ git count-objects31806930 objects, 47620 kilobytes3181------------------------------------------------31823183The first number is the number of objects which are kept in3184individual files. The second is the amount of space taken up by3185those "loose" objects.31863187You can save space and make git faster by moving these loose objects in3188to a "pack file", which stores a group of objects in an efficient3189compressed format; the details of how pack files are formatted can be3190found in link:technical/pack-format.txt[technical/pack-format.txt].31913192To put the loose objects into a pack, just run git repack:31933194------------------------------------------------3195$ git repack3196Generating pack...3197Done counting 6020 objects.3198Deltifying 6020 objects.3199 100% (6020/6020) done3200Writing 6020 objects.3201 100% (6020/6020) done3202Total 6020, written 6020 (delta 4070), reused 0 (delta 0)3203Pack pack-3e54ad29d5b2e05838c75df582c65257b8d08e1c created.3204------------------------------------------------32053206You can then run32073208------------------------------------------------3209$ git prune3210------------------------------------------------32113212to remove any of the "loose" objects that are now contained in the3213pack. This will also remove any unreferenced objects (which may be3214created when, for example, you use "git reset" to remove a commit).3215You can verify that the loose objects are gone by looking at the3216.git/objects directory or by running32173218------------------------------------------------3219$ git count-objects32200 objects, 0 kilobytes3221------------------------------------------------32223223Although the object files are gone, any commands that refer to those3224objects will work exactly as they did before.32253226The linkgit:git-gc[1] command performs packing, pruning, and more for3227you, so is normally the only high-level command you need.32283229[[dangling-objects]]3230Dangling objects3231~~~~~~~~~~~~~~~~32323233The linkgit:git-fsck[1] command will sometimes complain about dangling3234objects. They are not a problem.32353236The most common cause of dangling objects is that you've rebased a3237branch, or you have pulled from somebody else who rebased a branch--see3238<<cleaning-up-history>>. In that case, the old head of the original3239branch still exists, as does everything it pointed to. The branch3240pointer itself just doesn't, since you replaced it with another one.32413242There are also other situations that cause dangling objects. For3243example, a "dangling blob" may arise because you did a "git add" of a3244file, but then, before you actually committed it and made it part of the3245bigger picture, you changed something else in that file and committed3246that *updated* thing--the old state that you added originally ends up3247not being pointed to by any commit or tree, so it's now a dangling blob3248object.32493250Similarly, when the "recursive" merge strategy runs, and finds that3251there are criss-cross merges and thus more than one merge base (which is3252fairly unusual, but it does happen), it will generate one temporary3253midway tree (or possibly even more, if you had lots of criss-crossing3254merges and more than two merge bases) as a temporary internal merge3255base, and again, those are real objects, but the end result will not end3256up pointing to them, so they end up "dangling" in your repository.32573258Generally, dangling objects aren't anything to worry about. They can3259even be very useful: if you screw something up, the dangling objects can3260be how you recover your old tree (say, you did a rebase, and realized3261that you really didn't want to--you can look at what dangling objects3262you have, and decide to reset your head to some old dangling state).32633264For commits, you can just use:32653266------------------------------------------------3267$ gitk <dangling-commit-sha-goes-here> --not --all3268------------------------------------------------32693270This asks for all the history reachable from the given commit but not3271from any branch, tag, or other reference. If you decide it's something3272you want, you can always create a new reference to it, e.g.,32733274------------------------------------------------3275$ git branch recovered-branch <dangling-commit-sha-goes-here>3276------------------------------------------------32773278For blobs and trees, you can't do the same, but you can still examine3279them. You can just do32803281------------------------------------------------3282$ git show <dangling-blob/tree-sha-goes-here>3283------------------------------------------------32843285to show what the contents of the blob were (or, for a tree, basically3286what the "ls" for that directory was), and that may give you some idea3287of what the operation was that left that dangling object.32883289Usually, dangling blobs and trees aren't very interesting. They're3290almost always the result of either being a half-way mergebase (the blob3291will often even have the conflict markers from a merge in it, if you3292have had conflicting merges that you fixed up by hand), or simply3293because you interrupted a "git fetch" with ^C or something like that,3294leaving _some_ of the new objects in the object database, but just3295dangling and useless.32963297Anyway, once you are sure that you're not interested in any dangling3298state, you can just prune all unreachable objects:32993300------------------------------------------------3301$ git prune3302------------------------------------------------33033304and they'll be gone. But you should only run "git prune" on a quiescent3305repository--it's kind of like doing a filesystem fsck recovery: you3306don't want to do that while the filesystem is mounted.33073308(The same is true of "git fsck" itself, btw, but since3309`git fsck` never actually *changes* the repository, it just reports3310on what it found, `git fsck` itself is never 'dangerous' to run.3311Running it while somebody is actually changing the repository can cause3312confusing and scary messages, but it won't actually do anything bad. In3313contrast, running "git prune" while somebody is actively changing the3314repository is a *BAD* idea).33153316[[recovering-from-repository-corruption]]3317Recovering from repository corruption3318~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~33193320By design, git treats data trusted to it with caution. However, even in3321the absence of bugs in git itself, it is still possible that hardware or3322operating system errors could corrupt data.33233324The first defense against such problems is backups. You can back up a3325git directory using clone, or just using cp, tar, or any other backup3326mechanism.33273328As a last resort, you can search for the corrupted objects and attempt3329to replace them by hand. Back up your repository before attempting this3330in case you corrupt things even more in the process.33313332We'll assume that the problem is a single missing or corrupted blob,3333which is sometimes a solvable problem. (Recovering missing trees and3334especially commits is *much* harder).33353336Before starting, verify that there is corruption, and figure out where3337it is with linkgit:git-fsck[1]; this may be time-consuming.33383339Assume the output looks like this:33403341------------------------------------------------3342$ git fsck --full --no-dangling3343broken link from tree 2d9263c6d23595e7cb2a21e5ebbb53655278dff83344 to blob 4b9458b3786228369c63936db65827de3cc062003345missing blob 4b9458b3786228369c63936db65827de3cc062003346------------------------------------------------33473348Now you know that blob 4b9458b3 is missing, and that the tree 2d9263c63349points to it. If you could find just one copy of that missing blob3350object, possibly in some other repository, you could move it into3351.git/objects/4b/9458b3... and be done. Suppose you can't. You can3352still examine the tree that pointed to it with linkgit:git-ls-tree[1],3353which might output something like:33543355------------------------------------------------3356$ git ls-tree 2d9263c6d23595e7cb2a21e5ebbb53655278dff83357100644 blob 8d14531846b95bfa3564b58ccfb7913a034323b8 .gitignore3358100644 blob ebf9bf84da0aab5ed944264a5db2a65fe3a3e883 .mailmap3359100644 blob ca442d313d86dc67e0a2e5d584b465bd382cbf5c COPYING3360...3361100644 blob 4b9458b3786228369c63936db65827de3cc06200 myfile3362...3363------------------------------------------------33643365So now you know that the missing blob was the data for a file named3366"myfile". And chances are you can also identify the directory--let's3367say it's in "somedirectory". If you're lucky the missing copy might be3368the same as the copy you have checked out in your working tree at3369"somedirectory/myfile"; you can test whether that's right with3370linkgit:git-hash-object[1]:33713372------------------------------------------------3373$ git hash-object -w somedirectory/myfile3374------------------------------------------------33753376which will create and store a blob object with the contents of3377somedirectory/myfile, and output the SHA-1 of that object. if you're3378extremely lucky it might be 4b9458b3786228369c63936db65827de3cc06200, in3379which case you've guessed right, and the corruption is fixed!33803381Otherwise, you need more information. How do you tell which version of3382the file has been lost?33833384The easiest way to do this is with:33853386------------------------------------------------3387$ git log --raw --all --full-history -- somedirectory/myfile3388------------------------------------------------33893390Because you're asking for raw output, you'll now get something like33913392------------------------------------------------3393commit abc3394Author:3395Date:3396...3397:100644 100644 4b9458b... newsha... M somedirectory/myfile339833993400commit xyz3401Author:3402Date:34033404...3405:100644 100644 oldsha... 4b9458b... M somedirectory/myfile3406------------------------------------------------34073408This tells you that the immediately following version of the file was3409"newsha", and that the immediately preceding version was "oldsha".3410You also know the commit messages that went with the change from oldsha3411to 4b9458b and with the change from 4b9458b to newsha.34123413If you've been committing small enough changes, you may now have a good3414shot at reconstructing the contents of the in-between state 4b9458b.34153416If you can do that, you can now recreate the missing object with34173418------------------------------------------------3419$ git hash-object -w <recreated-file>3420------------------------------------------------34213422and your repository is good again!34233424(Btw, you could have ignored the fsck, and started with doing a34253426------------------------------------------------3427$ git log --raw --all3428------------------------------------------------34293430and just looked for the sha of the missing object (4b9458b..) in that3431whole thing. It's up to you--Git does *have* a lot of information, it is3432just missing one particular blob version.34333434[[the-index]]3435The index3436-----------34373438The index is a binary file (generally kept in .git/index) containing a3439sorted list of path names, each with permissions and the SHA-1 of a blob3440object; linkgit:git-ls-files[1] can show you the contents of the index:34413442-------------------------------------------------3443$ git ls-files --stage3444100644 63c918c667fa005ff12ad89437f2fdc80926e21c 0 .gitignore3445100644 5529b198e8d14decbe4ad99db3f7fb632de0439d 0 .mailmap3446100644 6ff87c4664981e4397625791c8ea3bbb5f2279a3 0 COPYING3447100644 a37b2152bd26be2c2289e1f57a292534a51a93c7 0 Documentation/.gitignore3448100644 fbefe9a45b00a54b58d94d06eca48b03d40a50e0 0 Documentation/Makefile3449...3450100644 2511aef8d89ab52be5ec6a5e46236b4b6bcd07ea 0 xdiff/xtypes.h3451100644 2ade97b2574a9f77e7ae4002a4e07a6a38e46d07 0 xdiff/xutils.c3452100644 d5de8292e05e7c36c4b68857c1cf9855e3d2f70a 0 xdiff/xutils.h3453-------------------------------------------------34543455Note that in older documentation you may see the index called the3456"current directory cache" or just the "cache". It has three important3457properties:345834591. The index contains all the information necessary to generate a single3460(uniquely determined) tree object.3461+3462For example, running linkgit:git-commit[1] generates this tree object3463from the index, stores it in the object database, and uses it as the3464tree object associated with the new commit.346534662. The index enables fast comparisons between the tree object it defines3467and the working tree.3468+3469It does this by storing some additional data for each entry (such as3470the last modified time). This data is not displayed above, and is not3471stored in the created tree object, but it can be used to determine3472quickly which files in the working directory differ from what was3473stored in the index, and thus save git from having to read all of the3474data from such files to look for changes.347534763. It can efficiently represent information about merge conflicts3477between different tree objects, allowing each pathname to be3478associated with sufficient information about the trees involved that3479you can create a three-way merge between them.3480+3481We saw in <<conflict-resolution>> that during a merge the index can3482store multiple versions of a single file (called "stages"). The third3483column in the linkgit:git-ls-files[1] output above is the stage3484number, and will take on values other than 0 for files with merge3485conflicts.34863487The index is thus a sort of temporary staging area, which is filled with3488a tree which you are in the process of working on.34893490If you blow the index away entirely, you generally haven't lost any3491information as long as you have the name of the tree that it described.34923493[[submodules]]3494Submodules3495==========34963497Large projects are often composed of smaller, self-contained modules. For3498example, an embedded Linux distribution's source tree would include every3499piece of software in the distribution with some local modifications; a movie3500player might need to build against a specific, known-working version of a3501decompression library; several independent programs might all share the same3502build scripts.35033504With centralized revision control systems this is often accomplished by3505including every module in one single repository. Developers can check out3506all modules or only the modules they need to work with. They can even modify3507files across several modules in a single commit while moving things around3508or updating APIs and translations.35093510Git does not allow partial checkouts, so duplicating this approach in Git3511would force developers to keep a local copy of modules they are not3512interested in touching. Commits in an enormous checkout would be slower3513than you'd expect as Git would have to scan every directory for changes.3514If modules have a lot of local history, clones would take forever.35153516On the plus side, distributed revision control systems can much better3517integrate with external sources. In a centralized model, a single arbitrary3518snapshot of the external project is exported from its own revision control3519and then imported into the local revision control on a vendor branch. All3520the history is hidden. With distributed revision control you can clone the3521entire external history and much more easily follow development and re-merge3522local changes.35233524Git's submodule support allows a repository to contain, as a subdirectory, a3525checkout of an external project. Submodules maintain their own identity;3526the submodule support just stores the submodule repository location and3527commit ID, so other developers who clone the containing project3528("superproject") can easily clone all the submodules at the same revision.3529Partial checkouts of the superproject are possible: you can tell Git to3530clone none, some or all of the submodules.35313532The linkgit:git-submodule[1] command is available since Git 1.5.3. Users3533with Git 1.5.2 can look up the submodule commits in the repository and3534manually check them out; earlier versions won't recognize the submodules at3535all.35363537To see how submodule support works, create (for example) four example3538repositories that can be used later as a submodule:35393540-------------------------------------------------3541$ mkdir ~/git3542$ cd ~/git3543$ for i in a b c d3544do3545 mkdir $i3546 cd $i3547 git init3548 echo "module $i" > $i.txt3549 git add $i.txt3550 git commit -m "Initial commit, submodule $i"3551 cd ..3552done3553-------------------------------------------------35543555Now create the superproject and add all the submodules:35563557-------------------------------------------------3558$ mkdir super3559$ cd super3560$ git init3561$ for i in a b c d3562do3563 git submodule add ~/git/$i $i3564done3565-------------------------------------------------35663567NOTE: Do not use local URLs here if you plan to publish your superproject!35683569See what files `git submodule` created:35703571-------------------------------------------------3572$ ls -a3573. .. .git .gitmodules a b c d3574-------------------------------------------------35753576The `git submodule add <repo> <path>` command does a couple of things:35773578- It clones the submodule from <repo> to the given <path> under the3579 current directory and by default checks out the master branch.3580- It adds the submodule's clone path to the linkgit:gitmodules[5] file and3581 adds this file to the index, ready to be committed.3582- It adds the submodule's current commit ID to the index, ready to be3583 committed.35843585Commit the superproject:35863587-------------------------------------------------3588$ git commit -m "Add submodules a, b, c and d."3589-------------------------------------------------35903591Now clone the superproject:35923593-------------------------------------------------3594$ cd ..3595$ git clone super cloned3596$ cd cloned3597-------------------------------------------------35983599The submodule directories are there, but they're empty:36003601-------------------------------------------------3602$ ls -a a3603. ..3604$ git submodule status3605-d266b9873ad50488163457f025db7cdd9683d88b a3606-e81d457da15309b4fef4249aba9b50187999670d b3607-c1536a972b9affea0f16e0680ba87332dc059146 c3608-d96249ff5d57de5de093e6baff9e0aafa5276a74 d3609-------------------------------------------------36103611NOTE: The commit object names shown above would be different for you, but they3612should match the HEAD commit object names of your repositories. You can check3613it by running `git ls-remote ../a`.36143615Pulling down the submodules is a two-step process. First run `git submodule3616init` to add the submodule repository URLs to `.git/config`:36173618-------------------------------------------------3619$ git submodule init3620-------------------------------------------------36213622Now use `git submodule update` to clone the repositories and check out the3623commits specified in the superproject:36243625-------------------------------------------------3626$ git submodule update3627$ cd a3628$ ls -a3629. .. .git a.txt3630-------------------------------------------------36313632One major difference between `git submodule update` and `git submodule add` is3633that `git submodule update` checks out a specific commit, rather than the tip3634of a branch. It's like checking out a tag: the head is detached, so you're not3635working on a branch.36363637-------------------------------------------------3638$ git branch3639* (no branch)3640 master3641-------------------------------------------------36423643If you want to make a change within a submodule and you have a detached head,3644then you should create or checkout a branch, make your changes, publish the3645change within the submodule, and then update the superproject to reference the3646new commit:36473648-------------------------------------------------3649$ git checkout master3650-------------------------------------------------36513652or36533654-------------------------------------------------3655$ git checkout -b fix-up3656-------------------------------------------------36573658then36593660-------------------------------------------------3661$ echo "adding a line again" >> a.txt3662$ git commit -a -m "Updated the submodule from within the superproject."3663$ git push3664$ cd ..3665$ git diff3666diff --git a/a b/a3667index d266b98..261dfac 1600003668--- a/a3669+++ b/a3670@@ -1 +1 @@3671-Subproject commit d266b9873ad50488163457f025db7cdd9683d88b3672+Subproject commit 261dfac35cb99d380eb966e102c1197139f7fa243673$ git add a3674$ git commit -m "Updated submodule a."3675$ git push3676-------------------------------------------------36773678You have to run `git submodule update` after `git pull` if you want to update3679submodules, too.36803681Pitfalls with submodules3682------------------------36833684Always publish the submodule change before publishing the change to the3685superproject that references it. If you forget to publish the submodule change,3686others won't be able to clone the repository:36873688-------------------------------------------------3689$ cd ~/git/super/a3690$ echo i added another line to this file >> a.txt3691$ git commit -a -m "doing it wrong this time"3692$ cd ..3693$ git add a3694$ git commit -m "Updated submodule a again."3695$ git push3696$ cd ~/git/cloned3697$ git pull3698$ git submodule update3699error: pathspec '261dfac35cb99d380eb966e102c1197139f7fa24' did not match any file(s) known to git.3700Did you forget to 'git add'?3701Unable to checkout '261dfac35cb99d380eb966e102c1197139f7fa24' in submodule path 'a'3702-------------------------------------------------37033704In older git versions it could be easily forgotten to commit new or modified3705files in a submodule, which silently leads to similar problems as not pushing3706the submodule changes. Starting with git 1.7.0 both "git status" and "git diff"3707in the superproject show submodules as modified when they contain new or3708modified files to protect against accidentally committing such a state. "git3709diff" will also add a "-dirty" to the work tree side when generating patch3710output or used with the --submodule option:37113712-------------------------------------------------3713$ git diff3714diff --git a/sub b/sub3715--- a/sub3716+++ b/sub3717@@ -1 +1 @@3718-Subproject commit 3f356705649b5d566d97ff843cf193359229a4533719+Subproject commit 3f356705649b5d566d97ff843cf193359229a453-dirty3720$ git diff --submodule3721Submodule sub 3f35670..3f35670-dirty:3722-------------------------------------------------37233724You also should not rewind branches in a submodule beyond commits that were3725ever recorded in any superproject.37263727It's not safe to run `git submodule update` if you've made and committed3728changes within a submodule without checking out a branch first. They will be3729silently overwritten:37303731-------------------------------------------------3732$ cat a.txt3733module a3734$ echo line added from private2 >> a.txt3735$ git commit -a -m "line added inside private2"3736$ cd ..3737$ git submodule update3738Submodule path 'a': checked out 'd266b9873ad50488163457f025db7cdd9683d88b'3739$ cd a3740$ cat a.txt3741module a3742-------------------------------------------------37433744NOTE: The changes are still visible in the submodule's reflog.37453746This is not the case if you did not commit your changes.37473748[[low-level-operations]]3749Low-level git operations3750========================37513752Many of the higher-level commands were originally implemented as shell3753scripts using a smaller core of low-level git commands. These can still3754be useful when doing unusual things with git, or just as a way to3755understand its inner workings.37563757[[object-manipulation]]3758Object access and manipulation3759------------------------------37603761The linkgit:git-cat-file[1] command can show the contents of any object,3762though the higher-level linkgit:git-show[1] is usually more useful.37633764The linkgit:git-commit-tree[1] command allows constructing commits with3765arbitrary parents and trees.37663767A tree can be created with linkgit:git-write-tree[1] and its data can be3768accessed by linkgit:git-ls-tree[1]. Two trees can be compared with3769linkgit:git-diff-tree[1].37703771A tag is created with linkgit:git-mktag[1], and the signature can be3772verified by linkgit:git-verify-tag[1], though it is normally simpler to3773use linkgit:git-tag[1] for both.37743775[[the-workflow]]3776The Workflow3777------------37783779High-level operations such as linkgit:git-commit[1],3780linkgit:git-checkout[1] and linkgit:git-reset[1] work by moving data3781between the working tree, the index, and the object database. Git3782provides low-level operations which perform each of these steps3783individually.37843785Generally, all "git" operations work on the index file. Some operations3786work *purely* on the index file (showing the current state of the3787index), but most operations move data between the index file and either3788the database or the working directory. Thus there are four main3789combinations:37903791[[working-directory-to-index]]3792working directory -> index3793~~~~~~~~~~~~~~~~~~~~~~~~~~37943795The linkgit:git-update-index[1] command updates the index with3796information from the working directory. You generally update the3797index information by just specifying the filename you want to update,3798like so:37993800-------------------------------------------------3801$ git update-index filename3802-------------------------------------------------38033804but to avoid common mistakes with filename globbing etc, the command3805will not normally add totally new entries or remove old entries,3806i.e. it will normally just update existing cache entries.38073808To tell git that yes, you really do realize that certain files no3809longer exist, or that new files should be added, you3810should use the `--remove` and `--add` flags respectively.38113812NOTE! A `--remove` flag does 'not' mean that subsequent filenames will3813necessarily be removed: if the files still exist in your directory3814structure, the index will be updated with their new status, not3815removed. The only thing `--remove` means is that update-index will be3816considering a removed file to be a valid thing, and if the file really3817does not exist any more, it will update the index accordingly.38183819As a special case, you can also do `git update-index --refresh`, which3820will refresh the "stat" information of each index to match the current3821stat information. It will 'not' update the object status itself, and3822it will only update the fields that are used to quickly test whether3823an object still matches its old backing store object.38243825The previously introduced linkgit:git-add[1] is just a wrapper for3826linkgit:git-update-index[1].38273828[[index-to-object-database]]3829index -> object database3830~~~~~~~~~~~~~~~~~~~~~~~~38313832You write your current index file to a "tree" object with the program38333834-------------------------------------------------3835$ git write-tree3836-------------------------------------------------38373838that doesn't come with any options--it will just write out the3839current index into the set of tree objects that describe that state,3840and it will return the name of the resulting top-level tree. You can3841use that tree to re-generate the index at any time by going in the3842other direction:38433844[[object-database-to-index]]3845object database -> index3846~~~~~~~~~~~~~~~~~~~~~~~~38473848You read a "tree" file from the object database, and use that to3849populate (and overwrite--don't do this if your index contains any3850unsaved state that you might want to restore later!) your current3851index. Normal operation is just38523853-------------------------------------------------3854$ git read-tree <SHA-1 of tree>3855-------------------------------------------------38563857and your index file will now be equivalent to the tree that you saved3858earlier. However, that is only your 'index' file: your working3859directory contents have not been modified.38603861[[index-to-working-directory]]3862index -> working directory3863~~~~~~~~~~~~~~~~~~~~~~~~~~38643865You update your working directory from the index by "checking out"3866files. This is not a very common operation, since normally you'd just3867keep your files updated, and rather than write to your working3868directory, you'd tell the index files about the changes in your3869working directory (i.e. `git update-index`).38703871However, if you decide to jump to a new version, or check out somebody3872else's version, or just restore a previous tree, you'd populate your3873index file with read-tree, and then you need to check out the result3874with38753876-------------------------------------------------3877$ git checkout-index filename3878-------------------------------------------------38793880or, if you want to check out all of the index, use `-a`.38813882NOTE! `git checkout-index` normally refuses to overwrite old files, so3883if you have an old version of the tree already checked out, you will3884need to use the "-f" flag ('before' the "-a" flag or the filename) to3885'force' the checkout.388638873888Finally, there are a few odds and ends which are not purely moving3889from one representation to the other:38903891[[tying-it-all-together]]3892Tying it all together3893~~~~~~~~~~~~~~~~~~~~~38943895To commit a tree you have instantiated with "git write-tree", you'd3896create a "commit" object that refers to that tree and the history3897behind it--most notably the "parent" commits that preceded it in3898history.38993900Normally a "commit" has one parent: the previous state of the tree3901before a certain change was made. However, sometimes it can have two3902or more parent commits, in which case we call it a "merge", due to the3903fact that such a commit brings together ("merges") two or more3904previous states represented by other commits.39053906In other words, while a "tree" represents a particular directory state3907of a working directory, a "commit" represents that state in "time",3908and explains how we got there.39093910You create a commit object by giving it the tree that describes the3911state at the time of the commit, and a list of parents:39123913-------------------------------------------------3914$ git commit-tree <tree> -p <parent> [(-p <parent2>)...]3915-------------------------------------------------39163917and then giving the reason for the commit on stdin (either through3918redirection from a pipe or file, or by just typing it at the tty).39193920`git commit-tree` will return the name of the object that represents3921that commit, and you should save it away for later use. Normally,3922you'd commit a new `HEAD` state, and while git doesn't care where you3923save the note about that state, in practice we tend to just write the3924result to the file pointed at by `.git/HEAD`, so that we can always see3925what the last committed state was.39263927Here is an ASCII art by Jon Loeliger that illustrates how3928various pieces fit together.39293930------------39313932 commit-tree3933 commit obj3934 +----+3935 | |3936 | |3937 V V3938 +-----------+3939 | Object DB |3940 | Backing |3941 | Store |3942 +-----------+3943 ^3944 write-tree | |3945 tree obj | |3946 | | read-tree3947 | | tree obj3948 V3949 +-----------+3950 | Index |3951 | "cache" |3952 +-----------+3953 update-index ^3954 blob obj | |3955 | |3956 checkout-index -u | | checkout-index3957 stat | | blob obj3958 V3959 +-----------+3960 | Working |3961 | Directory |3962 +-----------+39633964------------396539663967[[examining-the-data]]3968Examining the data3969------------------39703971You can examine the data represented in the object database and the3972index with various helper tools. For every object, you can use3973linkgit:git-cat-file[1] to examine details about the3974object:39753976-------------------------------------------------3977$ git cat-file -t <objectname>3978-------------------------------------------------39793980shows the type of the object, and once you have the type (which is3981usually implicit in where you find the object), you can use39823983-------------------------------------------------3984$ git cat-file blob|tree|commit|tag <objectname>3985-------------------------------------------------39863987to show its contents. NOTE! Trees have binary content, and as a result3988there is a special helper for showing that content, called3989`git ls-tree`, which turns the binary content into a more easily3990readable form.39913992It's especially instructive to look at "commit" objects, since those3993tend to be small and fairly self-explanatory. In particular, if you3994follow the convention of having the top commit name in `.git/HEAD`,3995you can do39963997-------------------------------------------------3998$ git cat-file commit HEAD3999-------------------------------------------------40004001to see what the top commit was.40024003[[merging-multiple-trees]]4004Merging multiple trees4005----------------------40064007Git helps you do a three-way merge, which you can expand to n-way by4008repeating the merge procedure arbitrary times until you finally4009"commit" the state. The normal situation is that you'd only do one4010three-way merge (two parents), and commit it, but if you like to, you4011can do multiple parents in one go.40124013To do a three-way merge, you need the two sets of "commit" objects4014that you want to merge, use those to find the closest common parent (a4015third "commit" object), and then use those commit objects to find the4016state of the directory ("tree" object) at these points.40174018To get the "base" for the merge, you first look up the common parent4019of two commits with40204021-------------------------------------------------4022$ git merge-base <commit1> <commit2>4023-------------------------------------------------40244025which will return you the commit they are both based on. You should4026now look up the "tree" objects of those commits, which you can easily4027do with (for example)40284029-------------------------------------------------4030$ git cat-file commit <commitname> | head -14031-------------------------------------------------40324033since the tree object information is always the first line in a commit4034object.40354036Once you know the three trees you are going to merge (the one "original"4037tree, aka the common tree, and the two "result" trees, aka the branches4038you want to merge), you do a "merge" read into the index. This will4039complain if it has to throw away your old index contents, so you should4040make sure that you've committed those--in fact you would normally4041always do a merge against your last commit (which should thus match what4042you have in your current index anyway).40434044To do the merge, do40454046-------------------------------------------------4047$ git read-tree -m -u <origtree> <yourtree> <targettree>4048-------------------------------------------------40494050which will do all trivial merge operations for you directly in the4051index file, and you can just write the result out with4052`git write-tree`.405340544055[[merging-multiple-trees-2]]4056Merging multiple trees, continued4057---------------------------------40584059Sadly, many merges aren't trivial. If there are files that have4060been added, moved or removed, or if both branches have modified the4061same file, you will be left with an index tree that contains "merge4062entries" in it. Such an index tree can 'NOT' be written out to a tree4063object, and you will have to resolve any such merge clashes using4064other tools before you can write out the result.40654066You can examine such index state with `git ls-files --unmerged`4067command. An example:40684069------------------------------------------------4070$ git read-tree -m $orig HEAD $target4071$ git ls-files --unmerged4072100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1 hello.c4073100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2 hello.c4074100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello.c4075------------------------------------------------40764077Each line of the `git ls-files --unmerged` output begins with4078the blob mode bits, blob SHA-1, 'stage number', and the4079filename. The 'stage number' is git's way to say which tree it4080came from: stage 1 corresponds to the `$orig` tree, stage 2 to4081the `HEAD` tree, and stage 3 to the `$target` tree.40824083Earlier we said that trivial merges are done inside4084`git read-tree -m`. For example, if the file did not change4085from `$orig` to `HEAD` nor `$target`, or if the file changed4086from `$orig` to `HEAD` and `$orig` to `$target` the same way,4087obviously the final outcome is what is in `HEAD`. What the4088above example shows is that file `hello.c` was changed from4089`$orig` to `HEAD` and `$orig` to `$target` in a different way.4090You could resolve this by running your favorite 3-way merge4091program, e.g. `diff3`, `merge`, or git's own merge-file, on4092the blob objects from these three stages yourself, like this:40934094------------------------------------------------4095$ git cat-file blob 263414f... >hello.c~14096$ git cat-file blob 06fa6a2... >hello.c~24097$ git cat-file blob cc44c73... >hello.c~34098$ git merge-file hello.c~2 hello.c~1 hello.c~34099------------------------------------------------41004101This would leave the merge result in `hello.c~2` file, along4102with conflict markers if there are conflicts. After verifying4103the merge result makes sense, you can tell git what the final4104merge result for this file is by:41054106-------------------------------------------------4107$ mv -f hello.c~2 hello.c4108$ git update-index hello.c4109-------------------------------------------------41104111When a path is in the "unmerged" state, running `git update-index` for4112that path tells git to mark the path resolved.41134114The above is the description of a git merge at the lowest level,4115to help you understand what conceptually happens under the hood.4116In practice, nobody, not even git itself, runs `git cat-file` three times4117for this. There is a `git merge-index` program that extracts the4118stages to temporary files and calls a "merge" script on it:41194120-------------------------------------------------4121$ git merge-index git-merge-one-file hello.c4122-------------------------------------------------41234124and that is what higher level `git merge -s resolve` is implemented with.41254126[[hacking-git]]4127Hacking git4128===========41294130This chapter covers internal details of the git implementation which4131probably only git developers need to understand.41324133[[object-details]]4134Object storage format4135---------------------41364137All objects have a statically determined "type" which identifies the4138format of the object (i.e. how it is used, and how it can refer to other4139objects). There are currently four different object types: "blob",4140"tree", "commit", and "tag".41414142Regardless of object type, all objects share the following4143characteristics: they are all deflated with zlib, and have a header4144that not only specifies their type, but also provides size information4145about the data in the object. It's worth noting that the SHA-1 hash4146that is used to name the object is the hash of the original data4147plus this header, so `sha1sum` 'file' does not match the object name4148for 'file'.4149(Historical note: in the dawn of the age of git the hash4150was the SHA-1 of the 'compressed' object.)41514152As a result, the general consistency of an object can always be tested4153independently of the contents or the type of the object: all objects can4154be validated by verifying that (a) their hashes match the content of the4155file and (b) the object successfully inflates to a stream of bytes that4156forms a sequence of <ascii type without space> {plus} <space> {plus} <ascii decimal4157size> {plus} <byte\0> {plus} <binary object data>.41584159The structured objects can further have their structure and4160connectivity to other objects verified. This is generally done with4161the `git fsck` program, which generates a full dependency graph4162of all objects, and verifies their internal consistency (in addition4163to just verifying their superficial consistency through the hash).41644165[[birdview-on-the-source-code]]4166A birds-eye view of Git's source code4167-------------------------------------41684169It is not always easy for new developers to find their way through Git's4170source code. This section gives you a little guidance to show where to4171start.41724173A good place to start is with the contents of the initial commit, with:41744175----------------------------------------------------4176$ git checkout e83c51634177----------------------------------------------------41784179The initial revision lays the foundation for almost everything git has4180today, but is small enough to read in one sitting.41814182Note that terminology has changed since that revision. For example, the4183README in that revision uses the word "changeset" to describe what we4184now call a <<def_commit_object,commit>>.41854186Also, we do not call it "cache" any more, but rather "index"; however, the4187file is still called `cache.h`. Remark: Not much reason to change it now,4188especially since there is no good single name for it anyway, because it is4189basically _the_ header file which is included by _all_ of Git's C sources.41904191If you grasp the ideas in that initial commit, you should check out a4192more recent version and skim `cache.h`, `object.h` and `commit.h`.41934194In the early days, Git (in the tradition of UNIX) was a bunch of programs4195which were extremely simple, and which you used in scripts, piping the4196output of one into another. This turned out to be good for initial4197development, since it was easier to test new things. However, recently4198many of these parts have become builtins, and some of the core has been4199"libified", i.e. put into libgit.a for performance, portability reasons,4200and to avoid code duplication.42014202By now, you know what the index is (and find the corresponding data4203structures in `cache.h`), and that there are just a couple of object types4204(blobs, trees, commits and tags) which inherit their common structure from4205`struct object`, which is their first member (and thus, you can cast e.g.4206`(struct object *)commit` to achieve the _same_ as `&commit->object`, i.e.4207get at the object name and flags).42084209Now is a good point to take a break to let this information sink in.42104211Next step: get familiar with the object naming. Read <<naming-commits>>.4212There are quite a few ways to name an object (and not only revisions!).4213All of these are handled in `sha1_name.c`. Just have a quick look at4214the function `get_sha1()`. A lot of the special handling is done by4215functions like `get_sha1_basic()` or the likes.42164217This is just to get you into the groove for the most libified part of Git:4218the revision walker.42194220Basically, the initial version of `git log` was a shell script:42214222----------------------------------------------------------------4223$ git-rev-list --pretty $(git-rev-parse --default HEAD "$@") | \4224 LESS=-S ${PAGER:-less}4225----------------------------------------------------------------42264227What does this mean?42284229`git rev-list` is the original version of the revision walker, which4230_always_ printed a list of revisions to stdout. It is still functional,4231and needs to, since most new Git commands start out as scripts using4232`git rev-list`.42334234`git rev-parse` is not as important any more; it was only used to filter out4235options that were relevant for the different plumbing commands that were4236called by the script.42374238Most of what `git rev-list` did is contained in `revision.c` and4239`revision.h`. It wraps the options in a struct named `rev_info`, which4240controls how and what revisions are walked, and more.42414242The original job of `git rev-parse` is now taken by the function4243`setup_revisions()`, which parses the revisions and the common command line4244options for the revision walker. This information is stored in the struct4245`rev_info` for later consumption. You can do your own command line option4246parsing after calling `setup_revisions()`. After that, you have to call4247`prepare_revision_walk()` for initialization, and then you can get the4248commits one by one with the function `get_revision()`.42494250If you are interested in more details of the revision walking process,4251just have a look at the first implementation of `cmd_log()`; call4252`git show v1.3.0~155^2~4` and scroll down to that function (note that you4253no longer need to call `setup_pager()` directly).42544255Nowadays, `git log` is a builtin, which means that it is _contained_ in the4256command `git`. The source side of a builtin is42574258- a function called `cmd_<bla>`, typically defined in `builtin-<bla>.c`,4259 and declared in `builtin.h`,42604261- an entry in the `commands[]` array in `git.c`, and42624263- an entry in `BUILTIN_OBJECTS` in the `Makefile`.42644265Sometimes, more than one builtin is contained in one source file. For4266example, `cmd_whatchanged()` and `cmd_log()` both reside in `builtin-log.c`,4267since they share quite a bit of code. In that case, the commands which are4268_not_ named like the `.c` file in which they live have to be listed in4269`BUILT_INS` in the `Makefile`.42704271`git log` looks more complicated in C than it does in the original script,4272but that allows for a much greater flexibility and performance.42734274Here again it is a good point to take a pause.42754276Lesson three is: study the code. Really, it is the best way to learn about4277the organization of Git (after you know the basic concepts).42784279So, think about something which you are interested in, say, "how can I4280access a blob just knowing the object name of it?". The first step is to4281find a Git command with which you can do it. In this example, it is either4282`git show` or `git cat-file`.42834284For the sake of clarity, let's stay with `git cat-file`, because it42854286- is plumbing, and42874288- was around even in the initial commit (it literally went only through4289 some 20 revisions as `cat-file.c`, was renamed to `builtin-cat-file.c`4290 when made a builtin, and then saw less than 10 versions).42914292So, look into `builtin-cat-file.c`, search for `cmd_cat_file()` and look what4293it does.42944295------------------------------------------------------------------4296 git_config(git_default_config);4297 if (argc != 3)4298 usage("git cat-file [-t|-s|-e|-p|<type>] <sha1>");4299 if (get_sha1(argv[2], sha1))4300 die("Not a valid object name %s", argv[2]);4301------------------------------------------------------------------43024303Let's skip over the obvious details; the only really interesting part4304here is the call to `get_sha1()`. It tries to interpret `argv[2]` as an4305object name, and if it refers to an object which is present in the current4306repository, it writes the resulting SHA-1 into the variable `sha1`.43074308Two things are interesting here:43094310- `get_sha1()` returns 0 on _success_. This might surprise some new4311 Git hackers, but there is a long tradition in UNIX to return different4312 negative numbers in case of different errors--and 0 on success.43134314- the variable `sha1` in the function signature of `get_sha1()` is `unsigned4315 char *`, but is actually expected to be a pointer to `unsigned4316 char[20]`. This variable will contain the 160-bit SHA-1 of the given4317 commit. Note that whenever a SHA-1 is passed as `unsigned char *`, it4318 is the binary representation, as opposed to the ASCII representation in4319 hex characters, which is passed as `char *`.43204321You will see both of these things throughout the code.43224323Now, for the meat:43244325-----------------------------------------------------------------------------4326 case 0:4327 buf = read_object_with_reference(sha1, argv[1], &size, NULL);4328-----------------------------------------------------------------------------43294330This is how you read a blob (actually, not only a blob, but any type of4331object). To know how the function `read_object_with_reference()` actually4332works, find the source code for it (something like `git grep4333read_object_with | grep ":[a-z]"` in the git repository), and read4334the source.43354336To find out how the result can be used, just read on in `cmd_cat_file()`:43374338-----------------------------------4339 write_or_die(1, buf, size);4340-----------------------------------43414342Sometimes, you do not know where to look for a feature. In many such cases,4343it helps to search through the output of `git log`, and then `git show` the4344corresponding commit.43454346Example: If you know that there was some test case for `git bundle`, but4347do not remember where it was (yes, you _could_ `git grep bundle t/`, but that4348does not illustrate the point!):43494350------------------------4351$ git log --no-merges t/4352------------------------43534354In the pager (`less`), just search for "bundle", go a few lines back,4355and see that it is in commit 18449ab0... Now just copy this object name,4356and paste it into the command line43574358-------------------4359$ git show 18449ab04360-------------------43614362Voila.43634364Another example: Find out what to do in order to make some script a4365builtin:43664367-------------------------------------------------4368$ git log --no-merges --diff-filter=A builtin-*.c4369-------------------------------------------------43704371You see, Git is actually the best tool to find out about the source of Git4372itself!43734374[[glossary]]4375Git Glossary4376============43774378include::glossary-content.txt[]43794380[[git-quick-start]]4381Appendix A: Git Quick Reference4382===============================43834384This is a quick summary of the major commands; the previous chapters4385explain how these work in more detail.43864387[[quick-creating-a-new-repository]]4388Creating a new repository4389-------------------------43904391From a tarball:43924393-----------------------------------------------4394$ tar xzf project.tar.gz4395$ cd project4396$ git init4397Initialized empty Git repository in .git/4398$ git add .4399$ git commit4400-----------------------------------------------44014402From a remote repository:44034404-----------------------------------------------4405$ git clone git://example.com/pub/project.git4406$ cd project4407-----------------------------------------------44084409[[managing-branches]]4410Managing branches4411-----------------44124413-----------------------------------------------4414$ git branch # list all local branches in this repo4415$ git checkout test # switch working directory to branch "test"4416$ git branch new # create branch "new" starting at current HEAD4417$ git branch -d new # delete branch "new"4418-----------------------------------------------44194420Instead of basing a new branch on current HEAD (the default), use:44214422-----------------------------------------------4423$ git branch new test # branch named "test"4424$ git branch new v2.6.15 # tag named v2.6.154425$ git branch new HEAD^ # commit before the most recent4426$ git branch new HEAD^^ # commit before that4427$ git branch new test~10 # ten commits before tip of branch "test"4428-----------------------------------------------44294430Create and switch to a new branch at the same time:44314432-----------------------------------------------4433$ git checkout -b new v2.6.154434-----------------------------------------------44354436Update and examine branches from the repository you cloned from:44374438-----------------------------------------------4439$ git fetch # update4440$ git branch -r # list4441 origin/master4442 origin/next4443 ...4444$ git checkout -b masterwork origin/master4445-----------------------------------------------44464447Fetch a branch from a different repository, and give it a new4448name in your repository:44494450-----------------------------------------------4451$ git fetch git://example.com/project.git theirbranch:mybranch4452$ git fetch git://example.com/project.git v2.6.15:mybranch4453-----------------------------------------------44544455Keep a list of repositories you work with regularly:44564457-----------------------------------------------4458$ git remote add example git://example.com/project.git4459$ git remote # list remote repositories4460example4461origin4462$ git remote show example # get details4463* remote example4464 URL: git://example.com/project.git4465 Tracked remote branches4466 master4467 next4468 ...4469$ git fetch example # update branches from example4470$ git branch -r # list all remote branches4471-----------------------------------------------447244734474[[exploring-history]]4475Exploring history4476-----------------44774478-----------------------------------------------4479$ gitk # visualize and browse history4480$ git log # list all commits4481$ git log src/ # ...modifying src/4482$ git log v2.6.15..v2.6.16 # ...in v2.6.16, not in v2.6.154483$ git log master..test # ...in branch test, not in branch master4484$ git log test..master # ...in branch master, but not in test4485$ git log test...master # ...in one branch, not in both4486$ git log -S'foo()' # ...where difference contain "foo()"4487$ git log --since="2 weeks ago"4488$ git log -p # show patches as well4489$ git show # most recent commit4490$ git diff v2.6.15..v2.6.16 # diff between two tagged versions4491$ git diff v2.6.15..HEAD # diff with current head4492$ git grep "foo()" # search working directory for "foo()"4493$ git grep v2.6.15 "foo()" # search old tree for "foo()"4494$ git show v2.6.15:a.txt # look at old version of a.txt4495-----------------------------------------------44964497Search for regressions:44984499-----------------------------------------------4500$ git bisect start4501$ git bisect bad # current version is bad4502$ git bisect good v2.6.13-rc2 # last known good revision4503Bisecting: 675 revisions left to test after this4504 # test here, then:4505$ git bisect good # if this revision is good, or4506$ git bisect bad # if this revision is bad.4507 # repeat until done.4508-----------------------------------------------45094510[[making-changes]]4511Making changes4512--------------45134514Make sure git knows who to blame:45154516------------------------------------------------4517$ cat >>~/.gitconfig <<\EOF4518[user]4519 name = Your Name Comes Here4520 email = you@yourdomain.example.com4521EOF4522------------------------------------------------45234524Select file contents to include in the next commit, then make the4525commit:45264527-----------------------------------------------4528$ git add a.txt # updated file4529$ git add b.txt # new file4530$ git rm c.txt # old file4531$ git commit4532-----------------------------------------------45334534Or, prepare and create the commit in one step:45354536-----------------------------------------------4537$ git commit d.txt # use latest content only of d.txt4538$ git commit -a # use latest content of all tracked files4539-----------------------------------------------45404541[[merging]]4542Merging4543-------45444545-----------------------------------------------4546$ git merge test # merge branch "test" into the current branch4547$ git pull git://example.com/project.git master4548 # fetch and merge in remote branch4549$ git pull . test # equivalent to git merge test4550-----------------------------------------------45514552[[sharing-your-changes]]4553Sharing your changes4554--------------------45554556Importing or exporting patches:45574558-----------------------------------------------4559$ git format-patch origin..HEAD # format a patch for each commit4560 # in HEAD but not in origin4561$ git am mbox # import patches from the mailbox "mbox"4562-----------------------------------------------45634564Fetch a branch in a different git repository, then merge into the4565current branch:45664567-----------------------------------------------4568$ git pull git://example.com/project.git theirbranch4569-----------------------------------------------45704571Store the fetched branch into a local branch before merging into the4572current branch:45734574-----------------------------------------------4575$ git pull git://example.com/project.git theirbranch:mybranch4576-----------------------------------------------45774578After creating commits on a local branch, update the remote4579branch with your commits:45804581-----------------------------------------------4582$ git push ssh://example.com/project.git mybranch:theirbranch4583-----------------------------------------------45844585When remote and local branch are both named "test":45864587-----------------------------------------------4588$ git push ssh://example.com/project.git test4589-----------------------------------------------45904591Shortcut version for a frequently used remote repository:45924593-----------------------------------------------4594$ git remote add example ssh://example.com/project.git4595$ git push example test4596-----------------------------------------------45974598[[repository-maintenance]]4599Repository maintenance4600----------------------46014602Check for corruption:46034604-----------------------------------------------4605$ git fsck4606-----------------------------------------------46074608Recompress, remove unused cruft:46094610-----------------------------------------------4611$ git gc4612-----------------------------------------------461346144615[[todo]]4616Appendix B: Notes and todo list for this manual4617===============================================46184619This is a work in progress.46204621The basic requirements:46224623- It must be readable in order, from beginning to end, by someone4624 intelligent with a basic grasp of the UNIX command line, but without4625 any special knowledge of git. If necessary, any other prerequisites4626 should be specifically mentioned as they arise.4627- Whenever possible, section headings should clearly describe the task4628 they explain how to do, in language that requires no more knowledge4629 than necessary: for example, "importing patches into a project" rather4630 than "the `git am` command"46314632Think about how to create a clear chapter dependency graph that will4633allow people to get to important topics without necessarily reading4634everything in between.46354636Scan Documentation/ for other stuff left out; in particular:46374638- howto's4639- some of technical/?4640- hooks4641- list of commands in linkgit:git[1]46424643Scan email archives for other stuff left out46444645Scan man pages to see if any assume more background than this manual4646provides.46474648Simplify beginning by suggesting disconnected head instead of4649temporary branch creation?46504651Add more good examples. Entire sections of just cookbook examples4652might be a good idea; maybe make an "advanced examples" section a4653standard end-of-chapter section?46544655Include cross-references to the glossary, where appropriate.46564657Document shallow clones? See draft 1.5.0 release notes for some4658documentation.46594660Add a section on working with other version control systems, including4661CVS, Subversion, and just imports of series of release tarballs.46624663More details on gitweb?46644665Write a chapter on using plumbing and writing scripts.46664667Alternates, clone -reference, etc.46684669More on recovery from repository corruption. See:4670 http://marc.theaimsgroup.com/?l=git&m=117263864820799&w=24671 http://marc.theaimsgroup.com/?l=git&m=117147855503798&w=2