1A tutorial introduction to git 2============================== 3 4This tutorial explains how to import a new project into git, make 5changes to it, and share changes with other developers. 6 7First, note that you can get documentation for a command such as "git 8diff" with: 9 10------------------------------------------------ 11$ man git-diff 12------------------------------------------------ 13 14It is a good idea to introduce yourself to git before doing any 15operation. The easiest way to do so is: 16 17------------------------------------------------ 18$ cat >~/.gitconfig <<\EOF 19[user] 20 name = Your Name Comes Here 21 email = you@yourdomain.example.com 22EOF 23------------------------------------------------ 24 25 26Importing a new project 27----------------------- 28 29Assume you have a tarball project.tar.gz with your initial work. You 30can place it under git revision control as follows. 31 32------------------------------------------------ 33$ tar xzf project.tar.gz 34$ cd project 35$ git init-db 36------------------------------------------------ 37 38Git will reply 39 40------------------------------------------------ 41Initialized empty Git repository in .git/ 42------------------------------------------------ 43 44You've now initialized the working directory--you may notice a new 45directory created, named ".git". Tell git that you want it to track 46every file under the current directory with (notice the dot '.' 47that means the current directory): 48 49------------------------------------------------ 50$ git add . 51------------------------------------------------ 52 53Finally, 54 55------------------------------------------------ 56$ git commit 57------------------------------------------------ 58 59will prompt you for a commit message, then record the current state 60of all the files to the repository. 61 62Try modifying some files, then run 63 64------------------------------------------------ 65$ git diff 66------------------------------------------------ 67 68to review your changes. When you're done, tell git that you 69want the updated contents of these files in the commit and then 70make a commit, like this: 71 72------------------------------------------------ 73$ git add file1 file... 74$ git commit 75------------------------------------------------ 76 77This will again prompt your for a message describing the change, and then 78record the new versions of the files you listed. It is cumbersome 79to list all files and you can say `git commit -a` (which stands for 'all') 80instead of running `git add` beforehand. 81 82------------------------------------------------ 83$ git commit -a 84------------------------------------------------ 85 86A note on commit messages: Though not required, it's a good idea to 87begin the commit message with a single short (less than 50 character) 88line summarizing the change, followed by a blank line and then a more 89thorough description. Tools that turn commits into email, for 90example, use the first line on the Subject: line and the rest of the 91commit in the body. 92 93 94Git tracks content not files 95---------------------------- 96 97With git you have to explicitly "add" all the changed _content_ you 98want to commit together. This can be done in a few different ways: 99 1001) By using 'git add <file_spec>...' 101 102 This can be performed multiple times before a commit. Note that this 103 is not only for adding new files. Even modified files must be 104 added to the set of changes about to be committed. The "git status" 105 command gives you a summary of what is included so far for the 106 next commit. When done you should use the 'git commit' command to 107 make it real. 108 109 Note: don't forget to 'add' a file again if you modified it after the 110 first 'add' and before 'commit'. Otherwise only the previous added 111 state of that file will be committed. This is because git tracks 112 content, so what you're really 'add'ing to the commit is the *content* 113 of the file in the state it is in when you 'add' it. 114 1152) By using 'git commit -a' directly 116 117 This is a quick way to automatically 'add' the content from all files 118 that were modified since the previous commit, and perform the actual 119 commit without having to separately 'add' them beforehand. This will 120 not add content from new files i.e. files that were never added before. 121 Those files still have to be added explicitly before performing a 122 commit. 123 124But here's a twist. If you do 'git commit <file1> <file2> ...' then only 125the changes belonging to those explicitly specified files will be 126committed, entirely bypassing the current "added" changes. Those "added" 127changes will still remain available for a subsequent commit though. 128 129However, for normal usage you only have to remember 'git add' + 'git commit' 130and/or 'git commit -a'. 131 132 133Viewing the changelog 134--------------------- 135 136At any point you can view the history of your changes using 137 138------------------------------------------------ 139$ git log 140------------------------------------------------ 141 142If you also want to see complete diffs at each step, use 143 144------------------------------------------------ 145$ git log -p 146------------------------------------------------ 147 148Often the overview of the change is useful to get a feel of 149each step 150 151------------------------------------------------ 152$ git log --stat --summary 153------------------------------------------------ 154 155Managing branches 156----------------- 157 158A single git repository can maintain multiple branches of 159development. To create a new branch named "experimental", use 160 161------------------------------------------------ 162$ git branch experimental 163------------------------------------------------ 164 165If you now run 166 167------------------------------------------------ 168$ git branch 169------------------------------------------------ 170 171you'll get a list of all existing branches: 172 173------------------------------------------------ 174 experimental 175* master 176------------------------------------------------ 177 178The "experimental" branch is the one you just created, and the 179"master" branch is a default branch that was created for you 180automatically. The asterisk marks the branch you are currently on; 181type 182 183------------------------------------------------ 184$ git checkout experimental 185------------------------------------------------ 186 187to switch to the experimental branch. Now edit a file, commit the 188change, and switch back to the master branch: 189 190------------------------------------------------ 191(edit file) 192$ git commit -a 193$ git checkout master 194------------------------------------------------ 195 196Check that the change you made is no longer visible, since it was 197made on the experimental branch and you're back on the master branch. 198 199You can make a different change on the master branch: 200 201------------------------------------------------ 202(edit file) 203$ git commit -a 204------------------------------------------------ 205 206at this point the two branches have diverged, with different changes 207made in each. To merge the changes made in experimental into master, run 208 209------------------------------------------------ 210$ git pull . experimental 211------------------------------------------------ 212 213If the changes don't conflict, you're done. If there are conflicts, 214markers will be left in the problematic files showing the conflict; 215 216------------------------------------------------ 217$ git diff 218------------------------------------------------ 219 220will show this. Once you've edited the files to resolve the 221conflicts, 222 223------------------------------------------------ 224$ git commit -a 225------------------------------------------------ 226 227will commit the result of the merge. Finally, 228 229------------------------------------------------ 230$ gitk 231------------------------------------------------ 232 233will show a nice graphical representation of the resulting history. 234 235If you develop on a branch crazy-idea, then regret it, you can always 236delete the branch with 237 238------------------------------------- 239$ git branch -D crazy-idea 240------------------------------------- 241 242Branches are cheap and easy, so this is a good way to try something 243out. 244 245Using git for collaboration 246--------------------------- 247 248Suppose that Alice has started a new project with a git repository in 249/home/alice/project, and that Bob, who has a home directory on the 250same machine, wants to contribute. 251 252Bob begins with: 253 254------------------------------------------------ 255$ git clone /home/alice/project myrepo 256------------------------------------------------ 257 258This creates a new directory "myrepo" containing a clone of Alice's 259repository. The clone is on an equal footing with the original 260project, possessing its own copy of the original project's history. 261 262Bob then makes some changes and commits them: 263 264------------------------------------------------ 265(edit files) 266$ git commit -a 267(repeat as necessary) 268------------------------------------------------ 269 270When he's ready, he tells Alice to pull changes from the repository 271at /home/bob/myrepo. She does this with: 272 273------------------------------------------------ 274$ cd /home/alice/project 275$ git pull /home/bob/myrepo master 276------------------------------------------------ 277 278This merges the changes from Bob's "master" branch into Alice's 279current branch. If Alice has made her own changes in the meantime, 280then she may need to manually fix any conflicts. (Note that the 281"master" argument in the above command is actually unnecessary, as it 282is the default.) 283 284The "pull" command thus performs two operations: it fetches changes 285from a remote branch, then merges them into the current branch. 286 287You can perform the first operation alone using the "git fetch" 288command. For example, Alice could create a temporary branch just to 289track Bob's changes, without merging them with her own, using: 290 291------------------------------------- 292$ git fetch /home/bob/myrepo master:bob-incoming 293------------------------------------- 294 295which fetches the changes from Bob's master branch into a new branch 296named bob-incoming. Then 297 298------------------------------------- 299$ git log -p master..bob-incoming 300------------------------------------- 301 302shows a list of all the changes that Bob made since he branched from 303Alice's master branch. 304 305After examining those changes, and possibly fixing things, Alice 306could pull the changes into her master branch: 307 308------------------------------------- 309$ git checkout master 310$ git pull . bob-incoming 311------------------------------------- 312 313The last command is a pull from the "bob-incoming" branch in Alice's 314own repository. 315 316Alice could also perform both steps at once with: 317 318------------------------------------- 319$ git pull /home/bob/myrepo master:bob-incoming 320------------------------------------- 321 322This is just like the "git pull /home/bob/myrepo master" that we saw 323before, except that it also stores the unmerged changes from bob's 324master branch in bob-incoming before merging them into Alice's 325current branch. Note that git pull always merges into the current 326branch, regardless of what else is given on the commandline. 327 328Later, Bob can update his repo with Alice's latest changes using 329 330------------------------------------- 331$ git pull 332------------------------------------- 333 334Note that he doesn't need to give the path to Alice's repository; 335when Bob cloned Alice's repository, git stored the location of her 336repository in the repository configuration, and that location is 337used for pulls: 338 339------------------------------------- 340$ git repo-config --get remote.origin.url 341/home/bob/myrepo 342------------------------------------- 343 344(The complete configuration created by git-clone is visible using 345"git repo-config -l", and the gitlink:git-repo-config[1] man page 346explains the meaning of each option.) 347 348Git also keeps a pristine copy of Alice's master branch under the 349name "origin/master": 350 351------------------------------------- 352$ git branch -r 353 origin/master 354------------------------------------- 355 356If Bob later decides to work from a different host, he can still 357perform clones and pulls using the ssh protocol: 358 359------------------------------------- 360$ git clone alice.org:/home/alice/project myrepo 361------------------------------------- 362 363Alternatively, git has a native protocol, or can use rsync or http; 364see gitlink:git-pull[1] for details. 365 366Git can also be used in a CVS-like mode, with a central repository 367that various users push changes to; see gitlink:git-push[1] and 368link:cvs-migration.html[git for CVS users]. 369 370Exploring history 371----------------- 372 373Git history is represented as a series of interrelated commits. We 374have already seen that the git log command can list those commits. 375Note that first line of each git log entry also gives a name for the 376commit: 377 378------------------------------------- 379$ git log 380commit c82a22c39cbc32576f64f5c6b3f24b99ea8149c7 381Author: Junio C Hamano <junkio@cox.net> 382Date: Tue May 16 17:18:22 2006 -0700 383 384 merge-base: Clarify the comments on post processing. 385------------------------------------- 386 387We can give this name to git show to see the details about this 388commit. 389 390------------------------------------- 391$ git show c82a22c39cbc32576f64f5c6b3f24b99ea8149c7 392------------------------------------- 393 394But there are other ways to refer to commits. You can use any initial 395part of the name that is long enough to uniquely identify the commit: 396 397------------------------------------- 398$ git show c82a22c39c # the first few characters of the name are 399 # usually enough 400$ git show HEAD # the tip of the current branch 401$ git show experimental # the tip of the "experimental" branch 402------------------------------------- 403 404Every commit has at least one "parent" commit, which points to the 405previous state of the project: 406 407------------------------------------- 408$ git show HEAD^ # to see the parent of HEAD 409$ git show HEAD^^ # to see the grandparent of HEAD 410$ git show HEAD~4 # to see the great-great grandparent of HEAD 411------------------------------------- 412 413Note that merge commits may have more than one parent: 414 415------------------------------------- 416$ git show HEAD^1 # show the first parent of HEAD (same as HEAD^) 417$ git show HEAD^2 # show the second parent of HEAD 418------------------------------------- 419 420You can also give commits names of your own; after running 421 422------------------------------------- 423$ git-tag v2.5 1b2e1d63ff 424------------------------------------- 425 426you can refer to 1b2e1d63ff by the name "v2.5". If you intend to 427share this name with other people (for example, to identify a release 428version), you should create a "tag" object, and perhaps sign it; see 429gitlink:git-tag[1] for details. 430 431Any git command that needs to know a commit can take any of these 432names. For example: 433 434------------------------------------- 435$ git diff v2.5 HEAD # compare the current HEAD to v2.5 436$ git branch stable v2.5 # start a new branch named "stable" based 437 # at v2.5 438$ git reset --hard HEAD^ # reset your current branch and working 439 # directory to its state at HEAD^ 440------------------------------------- 441 442Be careful with that last command: in addition to losing any changes 443in the working directory, it will also remove all later commits from 444this branch. If this branch is the only branch containing those 445commits, they will be lost. (Also, don't use "git reset" on a 446publicly-visible branch that other developers pull from, as git will 447be confused by history that disappears in this way.) 448 449The git grep command can search for strings in any version of your 450project, so 451 452------------------------------------- 453$ git grep "hello" v2.5 454------------------------------------- 455 456searches for all occurrences of "hello" in v2.5. 457 458If you leave out the commit name, git grep will search any of the 459files it manages in your current directory. So 460 461------------------------------------- 462$ git grep "hello" 463------------------------------------- 464 465is a quick way to search just the files that are tracked by git. 466 467Many git commands also take sets of commits, which can be specified 468in a number of ways. Here are some examples with git log: 469 470------------------------------------- 471$ git log v2.5..v2.6 # commits between v2.5 and v2.6 472$ git log v2.5.. # commits since v2.5 473$ git log --since="2 weeks ago" # commits from the last 2 weeks 474$ git log v2.5.. Makefile # commits since v2.5 which modify 475 # Makefile 476------------------------------------- 477 478You can also give git log a "range" of commits where the first is not 479necessarily an ancestor of the second; for example, if the tips of 480the branches "stable-release" and "master" diverged from a common 481commit some time ago, then 482 483------------------------------------- 484$ git log stable..experimental 485------------------------------------- 486 487will list commits made in the experimental branch but not in the 488stable branch, while 489 490------------------------------------- 491$ git log experimental..stable 492------------------------------------- 493 494will show the list of commits made on the stable branch but not 495the experimental branch. 496 497The "git log" command has a weakness: it must present commits in a 498list. When the history has lines of development that diverged and 499then merged back together, the order in which "git log" presents 500those commits is meaningless. 501 502Most projects with multiple contributors (such as the linux kernel, 503or git itself) have frequent merges, and gitk does a better job of 504visualizing their history. For example, 505 506------------------------------------- 507$ gitk --since="2 weeks ago" drivers/ 508------------------------------------- 509 510allows you to browse any commits from the last 2 weeks of commits 511that modified files under the "drivers" directory. (Note: you can 512adjust gitk's fonts by holding down the control key while pressing 513"-" or "+".) 514 515Finally, most commands that take filenames will optionally allow you 516to precede any filename by a commit, to specify a particular version 517of the file: 518 519------------------------------------- 520$ git diff v2.5:Makefile HEAD:Makefile.in 521------------------------------------- 522 523You can also use "git cat-file -p" to see any such file: 524 525------------------------------------- 526$ git cat-file -p v2.5:Makefile 527------------------------------------- 528 529Next Steps 530---------- 531 532This tutorial should be enough to perform basic distributed revision 533control for your projects. However, to fully understand the depth 534and power of git you need to understand two simple ideas on which it 535is based: 536 537 * The object database is the rather elegant system used to 538 store the history of your project--files, directories, and 539 commits. 540 541 * The index file is a cache of the state of a directory tree, 542 used to create commits, check out working directories, and 543 hold the various trees involved in a merge. 544 545link:tutorial-2.html[Part two of this tutorial] explains the object 546database, the index file, and a few other odds and ends that you'll 547need to make the most of git. 548 549If you don't want to consider with that right away, a few other 550digressions that may be interesting at this point are: 551 552 * gitlink:git-format-patch[1], gitlink:git-am[1]: These convert 553 series of git commits into emailed patches, and vice versa, 554 useful for projects such as the linux kernel which rely heavily 555 on emailed patches. 556 557 * gitlink:git-bisect[1]: When there is a regression in your 558 project, one way to track down the bug is by searching through 559 the history to find the exact commit that's to blame. Git bisect 560 can help you perform a binary search for that commit. It is 561 smart enough to perform a close-to-optimal search even in the 562 case of complex non-linear history with lots of merged branches. 563 564 * link:everyday.html[Everyday GIT with 20 Commands Or So] 565 566 * link:cvs-migration.html[git for CVS users].