Documentation / gitcore-tutorial.txton commit Documentation: complicate example of "man git-command" (3861cd5)
   1gitcore-tutorial(7)
   2===================
   3
   4NAME
   5----
   6gitcore-tutorial - A git core tutorial for developers
   7
   8SYNOPSIS
   9--------
  10git *
  11
  12DESCRIPTION
  13-----------
  14
  15This tutorial explains how to use the "core" git programs to set up and
  16work with a git repository.
  17
  18If you just need to use git as a revision control system you may prefer
  19to start with "A Tutorial Introduction to GIT" (linkgit:gittutorial[7]) or
  20link:user-manual.html[the GIT User Manual].
  21
  22However, an understanding of these low-level tools can be helpful if
  23you want to understand git's internals.
  24
  25The core git is often called "plumbing", with the prettier user
  26interfaces on top of it called "porcelain". You may not want to use the
  27plumbing directly very often, but it can be good to know what the
  28plumbing does for when the porcelain isn't flushing.
  29
  30[NOTE]
  31Deeper technical details are often marked as Notes, which you can
  32skip on your first reading.
  33
  34
  35Creating a git repository
  36-------------------------
  37
  38Creating a new git repository couldn't be easier: all git repositories start
  39out empty, and the only thing you need to do is find yourself a
  40subdirectory that you want to use as a working tree - either an empty
  41one for a totally new project, or an existing working tree that you want
  42to import into git.
  43
  44For our first example, we're going to start a totally new repository from
  45scratch, with no pre-existing files, and we'll call it `git-tutorial`.
  46To start up, create a subdirectory for it, change into that
  47subdirectory, and initialize the git infrastructure with `git-init`:
  48
  49------------------------------------------------
  50$ mkdir git-tutorial
  51$ cd git-tutorial
  52$ git-init
  53------------------------------------------------
  54
  55to which git will reply
  56
  57----------------
  58Initialized empty Git repository in .git/
  59----------------
  60
  61which is just git's way of saying that you haven't been doing anything
  62strange, and that it will have created a local `.git` directory setup for
  63your new project. You will now have a `.git` directory, and you can
  64inspect that with `ls`. For your new empty project, it should show you
  65three entries, among other things:
  66
  67 - a file called `HEAD`, that has `ref: refs/heads/master` in it.
  68   This is similar to a symbolic link and points at
  69   `refs/heads/master` relative to the `HEAD` file.
  70+
  71Don't worry about the fact that the file that the `HEAD` link points to
  72doesn't even exist yet -- you haven't created the commit that will
  73start your `HEAD` development branch yet.
  74
  75 - a subdirectory called `objects`, which will contain all the
  76   objects of your project. You should never have any real reason to
  77   look at the objects directly, but you might want to know that these
  78   objects are what contains all the real 'data' in your repository.
  79
  80 - a subdirectory called `refs`, which contains references to objects.
  81
  82In particular, the `refs` subdirectory will contain two other
  83subdirectories, named `heads` and `tags` respectively. They do
  84exactly what their names imply: they contain references to any number
  85of different 'heads' of development (aka 'branches'), and to any
  86'tags' that you have created to name specific versions in your
  87repository.
  88
  89One note: the special `master` head is the default branch, which is
  90why the `.git/HEAD` file was created points to it even if it
  91doesn't yet exist. Basically, the `HEAD` link is supposed to always
  92point to the branch you are working on right now, and you always
  93start out expecting to work on the `master` branch.
  94
  95However, this is only a convention, and you can name your branches
  96anything you want, and don't have to ever even 'have' a `master`
  97branch. A number of the git tools will assume that `.git/HEAD` is
  98valid, though.
  99
 100[NOTE]
 101An 'object' is identified by its 160-bit SHA1 hash, aka 'object name',
 102and a reference to an object is always the 40-byte hex
 103representation of that SHA1 name. The files in the `refs`
 104subdirectory are expected to contain these hex references
 105(usually with a final `\'\n\'` at the end), and you should thus
 106expect to see a number of 41-byte files containing these
 107references in these `refs` subdirectories when you actually start
 108populating your tree.
 109
 110[NOTE]
 111An advanced user may want to take a look at linkgit:gitrepository-layout[5]
 112after finishing this tutorial.
 113
 114You have now created your first git repository. Of course, since it's
 115empty, that's not very useful, so let's start populating it with data.
 116
 117
 118Populating a git repository
 119---------------------------
 120
 121We'll keep this simple and stupid, so we'll start off with populating a
 122few trivial files just to get a feel for it.
 123
 124Start off with just creating any random files that you want to maintain
 125in your git repository. We'll start off with a few bad examples, just to
 126get a feel for how this works:
 127
 128------------------------------------------------
 129$ echo "Hello World" >hello
 130$ echo "Silly example" >example
 131------------------------------------------------
 132
 133you have now created two files in your working tree (aka 'working directory'),
 134but to actually check in your hard work, you will have to go through two steps:
 135
 136 - fill in the 'index' file (aka 'cache') with the information about your
 137   working tree state.
 138
 139 - commit that index file as an object.
 140
 141The first step is trivial: when you want to tell git about any changes
 142to your working tree, you use the `git-update-index` program. That
 143program normally just takes a list of filenames you want to update, but
 144to avoid trivial mistakes, it refuses to add new entries to the index
 145(or remove existing ones) unless you explicitly tell it that you're
 146adding a new entry with the `\--add` flag (or removing an entry with the
 147`\--remove`) flag.
 148
 149So to populate the index with the two files you just created, you can do
 150
 151------------------------------------------------
 152$ git-update-index --add hello example
 153------------------------------------------------
 154
 155and you have now told git to track those two files.
 156
 157In fact, as you did that, if you now look into your object directory,
 158you'll notice that git will have added two new objects to the object
 159database. If you did exactly the steps above, you should now be able to do
 160
 161
 162----------------
 163$ ls .git/objects/??/*
 164----------------
 165
 166and see two files:
 167
 168----------------
 169.git/objects/55/7db03de997c86a4a028e1ebd3a1ceb225be238
 170.git/objects/f2/4c74a2e500f5ee1332c86b94199f52b1d1d962
 171----------------
 172
 173which correspond with the objects with names of `557db...` and
 174`f24c7...` respectively.
 175
 176If you want to, you can use `git-cat-file` to look at those objects, but
 177you'll have to use the object name, not the filename of the object:
 178
 179----------------
 180$ git-cat-file -t 557db03de997c86a4a028e1ebd3a1ceb225be238
 181----------------
 182
 183where the `-t` tells `git-cat-file` to tell you what the "type" of the
 184object is. git will tell you that you have a "blob" object (i.e., just a
 185regular file), and you can see the contents with
 186
 187----------------
 188$ git-cat-file "blob" 557db03
 189----------------
 190
 191which will print out "Hello World". The object `557db03` is nothing
 192more than the contents of your file `hello`.
 193
 194[NOTE]
 195Don't confuse that object with the file `hello` itself. The
 196object is literally just those specific *contents* of the file, and
 197however much you later change the contents in file `hello`, the object
 198we just looked at will never change. Objects are immutable.
 199
 200[NOTE]
 201The second example demonstrates that you can
 202abbreviate the object name to only the first several
 203hexadecimal digits in most places.
 204
 205Anyway, as we mentioned previously, you normally never actually take a
 206look at the objects themselves, and typing long 40-character hex
 207names is not something you'd normally want to do. The above digression
 208was just to show that `git-update-index` did something magical, and
 209actually saved away the contents of your files into the git object
 210database.
 211
 212Updating the index did something else too: it created a `.git/index`
 213file. This is the index that describes your current working tree, and
 214something you should be very aware of. Again, you normally never worry
 215about the index file itself, but you should be aware of the fact that
 216you have not actually really "checked in" your files into git so far,
 217you've only *told* git about them.
 218
 219However, since git knows about them, you can now start using some of the
 220most basic git commands to manipulate the files or look at their status.
 221
 222In particular, let's not even check in the two files into git yet, we'll
 223start off by adding another line to `hello` first:
 224
 225------------------------------------------------
 226$ echo "It's a new day for git" >>hello
 227------------------------------------------------
 228
 229and you can now, since you told git about the previous state of `hello`, ask
 230git what has changed in the tree compared to your old index, using the
 231`git-diff-files` command:
 232
 233------------
 234$ git-diff-files
 235------------
 236
 237Oops. That wasn't very readable. It just spit out its own internal
 238version of a `diff`, but that internal version really just tells you
 239that it has noticed that "hello" has been modified, and that the old object
 240contents it had have been replaced with something else.
 241
 242To make it readable, we can tell git-diff-files to output the
 243differences as a patch, using the `-p` flag:
 244
 245------------
 246$ git-diff-files -p
 247diff --git a/hello b/hello
 248index 557db03..263414f 100644
 249--- a/hello
 250+++ b/hello
 251@@ -1 +1,2 @@
 252 Hello World
 253+It's a new day for git
 254----
 255
 256i.e. the diff of the change we caused by adding another line to `hello`.
 257
 258In other words, `git-diff-files` always shows us the difference between
 259what is recorded in the index, and what is currently in the working
 260tree. That's very useful.
 261
 262A common shorthand for `git-diff-files -p` is to just write `git
 263diff`, which will do the same thing.
 264
 265------------
 266$ git diff
 267diff --git a/hello b/hello
 268index 557db03..263414f 100644
 269--- a/hello
 270+++ b/hello
 271@@ -1 +1,2 @@
 272 Hello World
 273+It's a new day for git
 274------------
 275
 276
 277Committing git state
 278--------------------
 279
 280Now, we want to go to the next stage in git, which is to take the files
 281that git knows about in the index, and commit them as a real tree. We do
 282that in two phases: creating a 'tree' object, and committing that 'tree'
 283object as a 'commit' object together with an explanation of what the
 284tree was all about, along with information of how we came to that state.
 285
 286Creating a tree object is trivial, and is done with `git-write-tree`.
 287There are no options or other input: git-write-tree will take the
 288current index state, and write an object that describes that whole
 289index. In other words, we're now tying together all the different
 290filenames with their contents (and their permissions), and we're
 291creating the equivalent of a git "directory" object:
 292
 293------------------------------------------------
 294$ git-write-tree
 295------------------------------------------------
 296
 297and this will just output the name of the resulting tree, in this case
 298(if you have done exactly as I've described) it should be
 299
 300----------------
 3018988da15d077d4829fc51d8544c097def6644dbb
 302----------------
 303
 304which is another incomprehensible object name. Again, if you want to,
 305you can use `git-cat-file -t 8988d\...` to see that this time the object
 306is not a "blob" object, but a "tree" object (you can also use
 307`git-cat-file` to actually output the raw object contents, but you'll see
 308mainly a binary mess, so that's less interesting).
 309
 310However -- normally you'd never use `git-write-tree` on its own, because
 311normally you always commit a tree into a commit object using the
 312`git-commit-tree` command. In fact, it's easier to not actually use
 313`git-write-tree` on its own at all, but to just pass its result in as an
 314argument to `git-commit-tree`.
 315
 316`git-commit-tree` normally takes several arguments -- it wants to know
 317what the 'parent' of a commit was, but since this is the first commit
 318ever in this new repository, and it has no parents, we only need to pass in
 319the object name of the tree. However, `git-commit-tree` also wants to get a
 320commit message on its standard input, and it will write out the resulting
 321object name for the commit to its standard output.
 322
 323And this is where we create the `.git/refs/heads/master` file
 324which is pointed at by `HEAD`. This file is supposed to contain
 325the reference to the top-of-tree of the master branch, and since
 326that's exactly what `git-commit-tree` spits out, we can do this
 327all with a sequence of simple shell commands:
 328
 329------------------------------------------------
 330$ tree=$(git-write-tree)
 331$ commit=$(echo 'Initial commit' | git-commit-tree $tree)
 332$ git-update-ref HEAD $commit
 333------------------------------------------------
 334
 335In this case this creates a totally new commit that is not related to
 336anything else. Normally you do this only *once* for a project ever, and
 337all later commits will be parented on top of an earlier commit.
 338
 339Again, normally you'd never actually do this by hand. There is a
 340helpful script called `git commit` that will do all of this for you. So
 341you could have just written `git commit`
 342instead, and it would have done the above magic scripting for you.
 343
 344
 345Making a change
 346---------------
 347
 348Remember how we did the `git-update-index` on file `hello` and then we
 349changed `hello` afterward, and could compare the new state of `hello` with the
 350state we saved in the index file?
 351
 352Further, remember how I said that `git-write-tree` writes the contents
 353of the *index* file to the tree, and thus what we just committed was in
 354fact the *original* contents of the file `hello`, not the new ones. We did
 355that on purpose, to show the difference between the index state, and the
 356state in the working tree, and how they don't have to match, even
 357when we commit things.
 358
 359As before, if we do `git-diff-files -p` in our git-tutorial project,
 360we'll still see the same difference we saw last time: the index file
 361hasn't changed by the act of committing anything. However, now that we
 362have committed something, we can also learn to use a new command:
 363`git-diff-index`.
 364
 365Unlike `git-diff-files`, which showed the difference between the index
 366file and the working tree, `git-diff-index` shows the differences
 367between a committed *tree* and either the index file or the working
 368tree. In other words, `git-diff-index` wants a tree to be diffed
 369against, and before we did the commit, we couldn't do that, because we
 370didn't have anything to diff against.
 371
 372But now we can do
 373
 374----------------
 375$ git-diff-index -p HEAD
 376----------------
 377
 378(where `-p` has the same meaning as it did in `git-diff-files`), and it
 379will show us the same difference, but for a totally different reason.
 380Now we're comparing the working tree not against the index file,
 381but against the tree we just wrote. It just so happens that those two
 382are obviously the same, so we get the same result.
 383
 384Again, because this is a common operation, you can also just shorthand
 385it with
 386
 387----------------
 388$ git diff HEAD
 389----------------
 390
 391which ends up doing the above for you.
 392
 393In other words, `git-diff-index` normally compares a tree against the
 394working tree, but when given the `\--cached` flag, it is told to
 395instead compare against just the index cache contents, and ignore the
 396current working tree state entirely. Since we just wrote the index
 397file to HEAD, doing `git-diff-index \--cached -p HEAD` should thus return
 398an empty set of differences, and that's exactly what it does.
 399
 400[NOTE]
 401================
 402`git-diff-index` really always uses the index for its
 403comparisons, and saying that it compares a tree against the working
 404tree is thus not strictly accurate. In particular, the list of
 405files to compare (the "meta-data") *always* comes from the index file,
 406regardless of whether the `\--cached` flag is used or not. The `\--cached`
 407flag really only determines whether the file *contents* to be compared
 408come from the working tree or not.
 409
 410This is not hard to understand, as soon as you realize that git simply
 411never knows (or cares) about files that it is not told about
 412explicitly. git will never go *looking* for files to compare, it
 413expects you to tell it what the files are, and that's what the index
 414is there for.
 415================
 416
 417However, our next step is to commit the *change* we did, and again, to
 418understand what's going on, keep in mind the difference between "working
 419tree contents", "index file" and "committed tree". We have changes
 420in the working tree that we want to commit, and we always have to
 421work through the index file, so the first thing we need to do is to
 422update the index cache:
 423
 424------------------------------------------------
 425$ git-update-index hello
 426------------------------------------------------
 427
 428(note how we didn't need the `\--add` flag this time, since git knew
 429about the file already).
 430
 431Note what happens to the different `git-diff-\*` versions here. After
 432we've updated `hello` in the index, `git-diff-files -p` now shows no
 433differences, but `git-diff-index -p HEAD` still *does* show that the
 434current state is different from the state we committed. In fact, now
 435`git-diff-index` shows the same difference whether we use the `--cached`
 436flag or not, since now the index is coherent with the working tree.
 437
 438Now, since we've updated `hello` in the index, we can commit the new
 439version. We could do it by writing the tree by hand again, and
 440committing the tree (this time we'd have to use the `-p HEAD` flag to
 441tell commit that the HEAD was the *parent* of the new commit, and that
 442this wasn't an initial commit any more), but you've done that once
 443already, so let's just use the helpful script this time:
 444
 445------------------------------------------------
 446$ git commit
 447------------------------------------------------
 448
 449which starts an editor for you to write the commit message and tells you
 450a bit about what you have done.
 451
 452Write whatever message you want, and all the lines that start with '#'
 453will be pruned out, and the rest will be used as the commit message for
 454the change. If you decide you don't want to commit anything after all at
 455this point (you can continue to edit things and update the index), you
 456can just leave an empty message. Otherwise `git commit` will commit
 457the change for you.
 458
 459You've now made your first real git commit. And if you're interested in
 460looking at what `git commit` really does, feel free to investigate:
 461it's a few very simple shell scripts to generate the helpful (?) commit
 462message headers, and a few one-liners that actually do the
 463commit itself (`git-commit`).
 464
 465
 466Inspecting Changes
 467------------------
 468
 469While creating changes is useful, it's even more useful if you can tell
 470later what changed. The most useful command for this is another of the
 471`diff` family, namely `git-diff-tree`.
 472
 473`git-diff-tree` can be given two arbitrary trees, and it will tell you the
 474differences between them. Perhaps even more commonly, though, you can
 475give it just a single commit object, and it will figure out the parent
 476of that commit itself, and show the difference directly. Thus, to get
 477the same diff that we've already seen several times, we can now do
 478
 479----------------
 480$ git-diff-tree -p HEAD
 481----------------
 482
 483(again, `-p` means to show the difference as a human-readable patch),
 484and it will show what the last commit (in `HEAD`) actually changed.
 485
 486[NOTE]
 487============
 488Here is an ASCII art by Jon Loeliger that illustrates how
 489various diff-\* commands compare things.
 490
 491                      diff-tree
 492                       +----+
 493                       |    |
 494                       |    |
 495                       V    V
 496                    +-----------+
 497                    | Object DB |
 498                    |  Backing  |
 499                    |   Store   |
 500                    +-----------+
 501                      ^    ^
 502                      |    |
 503                      |    |  diff-index --cached
 504                      |    |
 505          diff-index  |    V
 506                      |  +-----------+
 507                      |  |   Index   |
 508                      |  |  "cache"  |
 509                      |  +-----------+
 510                      |    ^
 511                      |    |
 512                      |    |  diff-files
 513                      |    |
 514                      V    V
 515                    +-----------+
 516                    |  Working  |
 517                    | Directory |
 518                    +-----------+
 519============
 520
 521More interestingly, you can also give `git-diff-tree` the `--pretty` flag,
 522which tells it to also show the commit message and author and date of the
 523commit, and you can tell it to show a whole series of diffs.
 524Alternatively, you can tell it to be "silent", and not show the diffs at
 525all, but just show the actual commit message.
 526
 527In fact, together with the `git-rev-list` program (which generates a
 528list of revisions), `git-diff-tree` ends up being a veritable fount of
 529changes. A trivial (but very useful) script called `git-whatchanged` is
 530included with git which does exactly this, and shows a log of recent
 531activities.
 532
 533To see the whole history of our pitiful little git-tutorial project, you
 534can do
 535
 536----------------
 537$ git log
 538----------------
 539
 540which shows just the log messages, or if we want to see the log together
 541with the associated patches use the more complex (and much more
 542powerful)
 543
 544----------------
 545$ git-whatchanged -p
 546----------------
 547
 548and you will see exactly what has changed in the repository over its
 549short history.
 550
 551[NOTE]
 552When using the above two commands, the initial commit will be shown.
 553If this is a problem because it is huge, you can hide it by setting
 554the log.showroot configuration variable to false. Having this, you
 555can still show it for each command just adding the `\--root` option,
 556which is a flag for `git-diff-tree` accepted by both commands.
 557
 558With that, you should now be having some inkling of what git does, and
 559can explore on your own.
 560
 561[NOTE]
 562Most likely, you are not directly using the core
 563git Plumbing commands, but using Porcelain such as `git-add`, `git-rm'
 564and `git-commit'.
 565
 566
 567Tagging a version
 568-----------------
 569
 570In git, there are two kinds of tags, a "light" one, and an "annotated tag".
 571
 572A "light" tag is technically nothing more than a branch, except we put
 573it in the `.git/refs/tags/` subdirectory instead of calling it a `head`.
 574So the simplest form of tag involves nothing more than
 575
 576------------------------------------------------
 577$ git tag my-first-tag
 578------------------------------------------------
 579
 580which just writes the current `HEAD` into the `.git/refs/tags/my-first-tag`
 581file, after which point you can then use this symbolic name for that
 582particular state. You can, for example, do
 583
 584----------------
 585$ git diff my-first-tag
 586----------------
 587
 588to diff your current state against that tag which at this point will
 589obviously be an empty diff, but if you continue to develop and commit
 590stuff, you can use your tag as an "anchor-point" to see what has changed
 591since you tagged it.
 592
 593An "annotated tag" is actually a real git object, and contains not only a
 594pointer to the state you want to tag, but also a small tag name and
 595message, along with optionally a PGP signature that says that yes,
 596you really did
 597that tag. You create these annotated tags with either the `-a` or
 598`-s` flag to `git tag`:
 599
 600----------------
 601$ git tag -s <tagname>
 602----------------
 603
 604which will sign the current `HEAD` (but you can also give it another
 605argument that specifies the thing to tag, i.e., you could have tagged the
 606current `mybranch` point by using `git tag <tagname> mybranch`).
 607
 608You normally only do signed tags for major releases or things
 609like that, while the light-weight tags are useful for any marking you
 610want to do -- any time you decide that you want to remember a certain
 611point, just create a private tag for it, and you have a nice symbolic
 612name for the state at that point.
 613
 614
 615Copying repositories
 616--------------------
 617
 618git repositories are normally totally self-sufficient and relocatable.
 619Unlike CVS, for example, there is no separate notion of
 620"repository" and "working tree". A git repository normally *is* the
 621working tree, with the local git information hidden in the `.git`
 622subdirectory. There is nothing else. What you see is what you got.
 623
 624[NOTE]
 625You can tell git to split the git internal information from
 626the directory that it tracks, but we'll ignore that for now: it's not
 627how normal projects work, and it's really only meant for special uses.
 628So the mental model of "the git information is always tied directly to
 629the working tree that it describes" may not be technically 100%
 630accurate, but it's a good model for all normal use.
 631
 632This has two implications:
 633
 634 - if you grow bored with the tutorial repository you created (or you've
 635   made a mistake and want to start all over), you can just do simple
 636+
 637----------------
 638$ rm -rf git-tutorial
 639----------------
 640+
 641and it will be gone. There's no external repository, and there's no
 642history outside the project you created.
 643
 644 - if you want to move or duplicate a git repository, you can do so. There
 645   is `git clone` command, but if all you want to do is just to
 646   create a copy of your repository (with all the full history that
 647   went along with it), you can do so with a regular
 648   `cp -a git-tutorial new-git-tutorial`.
 649+
 650Note that when you've moved or copied a git repository, your git index
 651file (which caches various information, notably some of the "stat"
 652information for the files involved) will likely need to be refreshed.
 653So after you do a `cp -a` to create a new copy, you'll want to do
 654+
 655----------------
 656$ git-update-index --refresh
 657----------------
 658+
 659in the new repository to make sure that the index file is up-to-date.
 660
 661Note that the second point is true even across machines. You can
 662duplicate a remote git repository with *any* regular copy mechanism, be it
 663`scp`, `rsync` or `wget`.
 664
 665When copying a remote repository, you'll want to at a minimum update the
 666index cache when you do this, and especially with other peoples'
 667repositories you often want to make sure that the index cache is in some
 668known state (you don't know *what* they've done and not yet checked in),
 669so usually you'll precede the `git-update-index` with a
 670
 671----------------
 672$ git-read-tree --reset HEAD
 673$ git-update-index --refresh
 674----------------
 675
 676which will force a total index re-build from the tree pointed to by `HEAD`.
 677It resets the index contents to `HEAD`, and then the `git-update-index`
 678makes sure to match up all index entries with the checked-out files.
 679If the original repository had uncommitted changes in its
 680working tree, `git-update-index --refresh` notices them and
 681tells you they need to be updated.
 682
 683The above can also be written as simply
 684
 685----------------
 686$ git reset
 687----------------
 688
 689and in fact a lot of the common git command combinations can be scripted
 690with the `git xyz` interfaces.  You can learn things by just looking
 691at what the various git scripts do.  For example, `git reset` used to be
 692the above two lines implemented in `git-reset`, but some things like
 693`git status` and `git commit` are slightly more complex scripts around
 694the basic git commands.
 695
 696Many (most?) public remote repositories will not contain any of
 697the checked out files or even an index file, and will *only* contain the
 698actual core git files. Such a repository usually doesn't even have the
 699`.git` subdirectory, but has all the git files directly in the
 700repository.
 701
 702To create your own local live copy of such a "raw" git repository, you'd
 703first create your own subdirectory for the project, and then copy the
 704raw repository contents into the `.git` directory. For example, to
 705create your own copy of the git repository, you'd do the following
 706
 707----------------
 708$ mkdir my-git
 709$ cd my-git
 710$ rsync -rL rsync://rsync.kernel.org/pub/scm/git/git.git/ .git
 711----------------
 712
 713followed by
 714
 715----------------
 716$ git-read-tree HEAD
 717----------------
 718
 719to populate the index. However, now you have populated the index, and
 720you have all the git internal files, but you will notice that you don't
 721actually have any of the working tree files to work on. To get
 722those, you'd check them out with
 723
 724----------------
 725$ git-checkout-index -u -a
 726----------------
 727
 728where the `-u` flag means that you want the checkout to keep the index
 729up-to-date (so that you don't have to refresh it afterward), and the
 730`-a` flag means "check out all files" (if you have a stale copy or an
 731older version of a checked out tree you may also need to add the `-f`
 732flag first, to tell git-checkout-index to *force* overwriting of any old
 733files).
 734
 735Again, this can all be simplified with
 736
 737----------------
 738$ git clone rsync://rsync.kernel.org/pub/scm/git/git.git/ my-git
 739$ cd my-git
 740$ git checkout
 741----------------
 742
 743which will end up doing all of the above for you.
 744
 745You have now successfully copied somebody else's (mine) remote
 746repository, and checked it out.
 747
 748
 749Creating a new branch
 750---------------------
 751
 752Branches in git are really nothing more than pointers into the git
 753object database from within the `.git/refs/` subdirectory, and as we
 754already discussed, the `HEAD` branch is nothing but a symlink to one of
 755these object pointers.
 756
 757You can at any time create a new branch by just picking an arbitrary
 758point in the project history, and just writing the SHA1 name of that
 759object into a file under `.git/refs/heads/`. You can use any filename you
 760want (and indeed, subdirectories), but the convention is that the
 761"normal" branch is called `master`. That's just a convention, though,
 762and nothing enforces it.
 763
 764To show that as an example, let's go back to the git-tutorial repository we
 765used earlier, and create a branch in it. You do that by simply just
 766saying that you want to check out a new branch:
 767
 768------------
 769$ git checkout -b mybranch
 770------------
 771
 772will create a new branch based at the current `HEAD` position, and switch
 773to it.
 774
 775[NOTE]
 776================================================
 777If you make the decision to start your new branch at some
 778other point in the history than the current `HEAD`, you can do so by
 779just telling `git checkout` what the base of the checkout would be.
 780In other words, if you have an earlier tag or branch, you'd just do
 781
 782------------
 783$ git checkout -b mybranch earlier-commit
 784------------
 785
 786and it would create the new branch `mybranch` at the earlier commit,
 787and check out the state at that time.
 788================================================
 789
 790You can always just jump back to your original `master` branch by doing
 791
 792------------
 793$ git checkout master
 794------------
 795
 796(or any other branch-name, for that matter) and if you forget which
 797branch you happen to be on, a simple
 798
 799------------
 800$ cat .git/HEAD
 801------------
 802
 803will tell you where it's pointing.  To get the list of branches
 804you have, you can say
 805
 806------------
 807$ git branch
 808------------
 809
 810which used to be nothing more than a simple script around `ls .git/refs/heads`.
 811There will be an asterisk in front of the branch you are currently on.
 812
 813Sometimes you may wish to create a new branch _without_ actually
 814checking it out and switching to it. If so, just use the command
 815
 816------------
 817$ git branch <branchname> [startingpoint]
 818------------
 819
 820which will simply _create_ the branch, but will not do anything further.
 821You can then later -- once you decide that you want to actually develop
 822on that branch -- switch to that branch with a regular `git checkout`
 823with the branchname as the argument.
 824
 825
 826Merging two branches
 827--------------------
 828
 829One of the ideas of having a branch is that you do some (possibly
 830experimental) work in it, and eventually merge it back to the main
 831branch. So assuming you created the above `mybranch` that started out
 832being the same as the original `master` branch, let's make sure we're in
 833that branch, and do some work there.
 834
 835------------------------------------------------
 836$ git checkout mybranch
 837$ echo "Work, work, work" >>hello
 838$ git commit -m "Some work." -i hello
 839------------------------------------------------
 840
 841Here, we just added another line to `hello`, and we used a shorthand for
 842doing both `git-update-index hello` and `git commit` by just giving the
 843filename directly to `git commit`, with an `-i` flag (it tells
 844git to 'include' that file in addition to what you have done to
 845the index file so far when making the commit).  The `-m` flag is to give the
 846commit log message from the command line.
 847
 848Now, to make it a bit more interesting, let's assume that somebody else
 849does some work in the original branch, and simulate that by going back
 850to the master branch, and editing the same file differently there:
 851
 852------------
 853$ git checkout master
 854------------
 855
 856Here, take a moment to look at the contents of `hello`, and notice how they
 857don't contain the work we just did in `mybranch` -- because that work
 858hasn't happened in the `master` branch at all. Then do
 859
 860------------
 861$ echo "Play, play, play" >>hello
 862$ echo "Lots of fun" >>example
 863$ git commit -m "Some fun." -i hello example
 864------------
 865
 866since the master branch is obviously in a much better mood.
 867
 868Now, you've got two branches, and you decide that you want to merge the
 869work done. Before we do that, let's introduce a cool graphical tool that
 870helps you view what's going on:
 871
 872----------------
 873$ gitk --all
 874----------------
 875
 876will show you graphically both of your branches (that's what the `\--all`
 877means: normally it will just show you your current `HEAD`) and their
 878histories. You can also see exactly how they came to be from a common
 879source.
 880
 881Anyway, let's exit `gitk` (`^Q` or the File menu), and decide that we want
 882to merge the work we did on the `mybranch` branch into the `master`
 883branch (which is currently our `HEAD` too). To do that, there's a nice
 884script called `git merge`, which wants to know which branches you want
 885to resolve and what the merge is all about:
 886
 887------------
 888$ git merge -m "Merge work in mybranch" mybranch
 889------------
 890
 891where the first argument is going to be used as the commit message if
 892the merge can be resolved automatically.
 893
 894Now, in this case we've intentionally created a situation where the
 895merge will need to be fixed up by hand, though, so git will do as much
 896of it as it can automatically (which in this case is just merge the `example`
 897file, which had no differences in the `mybranch` branch), and say:
 898
 899----------------
 900        Auto-merging hello
 901        CONFLICT (content): Merge conflict in hello
 902        Automatic merge failed; fix up by hand
 903----------------
 904
 905It tells you that it did an "Automatic merge", which
 906failed due to conflicts in `hello`.
 907
 908Not to worry. It left the (trivial) conflict in `hello` in the same form you
 909should already be well used to if you've ever used CVS, so let's just
 910open `hello` in our editor (whatever that may be), and fix it up somehow.
 911I'd suggest just making it so that `hello` contains all four lines:
 912
 913------------
 914Hello World
 915It's a new day for git
 916Play, play, play
 917Work, work, work
 918------------
 919
 920and once you're happy with your manual merge, just do a
 921
 922------------
 923$ git commit -i hello
 924------------
 925
 926which will very loudly warn you that you're now committing a merge
 927(which is correct, so never mind), and you can write a small merge
 928message about your adventures in git-merge-land.
 929
 930After you're done, start up `gitk \--all` to see graphically what the
 931history looks like. Notice that `mybranch` still exists, and you can
 932switch to it, and continue to work with it if you want to. The
 933`mybranch` branch will not contain the merge, but next time you merge it
 934from the `master` branch, git will know how you merged it, so you'll not
 935have to do _that_ merge again.
 936
 937Another useful tool, especially if you do not always work in X-Window
 938environment, is `git show-branch`.
 939
 940------------------------------------------------
 941$ git-show-branch --topo-order --more=1 master mybranch
 942* [master] Merge work in mybranch
 943 ! [mybranch] Some work.
 944--
 945-  [master] Merge work in mybranch
 946*+ [mybranch] Some work.
 947*  [master^] Some fun.
 948------------------------------------------------
 949
 950The first two lines indicate that it is showing the two branches
 951and the first line of the commit log message from their
 952top-of-the-tree commits, you are currently on `master` branch
 953(notice the asterisk `\*` character), and the first column for
 954the later output lines is used to show commits contained in the
 955`master` branch, and the second column for the `mybranch`
 956branch. Three commits are shown along with their log messages.
 957All of them have non blank characters in the first column (`*`
 958shows an ordinary commit on the current branch, `-` is a merge commit), which
 959means they are now part of the `master` branch. Only the "Some
 960work" commit has the plus `+` character in the second column,
 961because `mybranch` has not been merged to incorporate these
 962commits from the master branch.  The string inside brackets
 963before the commit log message is a short name you can use to
 964name the commit.  In the above example, 'master' and 'mybranch'
 965are branch heads.  'master^' is the first parent of 'master'
 966branch head.  Please see 'git-rev-parse' documentation if you
 967see more complex cases.
 968
 969[NOTE]
 970Without the '--more=1' option, 'git-show-branch' would not output the
 971'[master^]' commit, as '[mybranch]' commit is a common ancestor of
 972both 'master' and 'mybranch' tips.  Please see 'git-show-branch'
 973documentation for details.
 974
 975[NOTE]
 976If there were more commits on the 'master' branch after the merge, the
 977merge commit itself would not be shown by 'git-show-branch' by
 978default.  You would need to provide '--sparse' option to make the
 979merge commit visible in this case.
 980
 981Now, let's pretend you are the one who did all the work in
 982`mybranch`, and the fruit of your hard work has finally been merged
 983to the `master` branch. Let's go back to `mybranch`, and run
 984`git merge` to get the "upstream changes" back to your branch.
 985
 986------------
 987$ git checkout mybranch
 988$ git merge -m "Merge upstream changes." master
 989------------
 990
 991This outputs something like this (the actual commit object names
 992would be different)
 993
 994----------------
 995Updating from ae3a2da... to a80b4aa....
 996Fast forward
 997 example |    1 +
 998 hello   |    1 +
 999 2 files changed, 2 insertions(+), 0 deletions(-)
1000----------------
1001
1002Because your branch did not contain anything more than what are
1003already merged into the `master` branch, the merge operation did
1004not actually do a merge. Instead, it just updated the top of
1005the tree of your branch to that of the `master` branch. This is
1006often called 'fast forward' merge.
1007
1008You can run `gitk \--all` again to see how the commit ancestry
1009looks like, or run `show-branch`, which tells you this.
1010
1011------------------------------------------------
1012$ git show-branch master mybranch
1013! [master] Merge work in mybranch
1014 * [mybranch] Merge work in mybranch
1015--
1016-- [master] Merge work in mybranch
1017------------------------------------------------
1018
1019
1020Merging external work
1021---------------------
1022
1023It's usually much more common that you merge with somebody else than
1024merging with your own branches, so it's worth pointing out that git
1025makes that very easy too, and in fact, it's not that different from
1026doing a `git merge`. In fact, a remote merge ends up being nothing
1027more than "fetch the work from a remote repository into a temporary tag"
1028followed by a `git merge`.
1029
1030Fetching from a remote repository is done by, unsurprisingly,
1031`git fetch`:
1032
1033----------------
1034$ git fetch <remote-repository>
1035----------------
1036
1037One of the following transports can be used to name the
1038repository to download from:
1039
1040Rsync::
1041        `rsync://remote.machine/path/to/repo.git/`
1042+
1043Rsync transport is usable for both uploading and downloading,
1044but is completely unaware of what git does, and can produce
1045unexpected results when you download from the public repository
1046while the repository owner is uploading into it via `rsync`
1047transport.  Most notably, it could update the files under
1048`refs/` which holds the object name of the topmost commits
1049before uploading the files in `objects/` -- the downloader would
1050obtain head commit object name while that object itself is still
1051not available in the repository.  For this reason, it is
1052considered deprecated.
1053
1054SSH::
1055        `remote.machine:/path/to/repo.git/` or
1056+
1057`ssh://remote.machine/path/to/repo.git/`
1058+
1059This transport can be used for both uploading and downloading,
1060and requires you to have a log-in privilege over `ssh` to the
1061remote machine.  It finds out the set of objects the other side
1062lacks by exchanging the head commits both ends have and
1063transfers (close to) minimum set of objects.  It is by far the
1064most efficient way to exchange git objects between repositories.
1065
1066Local directory::
1067        `/path/to/repo.git/`
1068+
1069This transport is the same as SSH transport but uses `sh` to run
1070both ends on the local machine instead of running other end on
1071the remote machine via `ssh`.
1072
1073git Native::
1074        `git://remote.machine/path/to/repo.git/`
1075+
1076This transport was designed for anonymous downloading.  Like SSH
1077transport, it finds out the set of objects the downstream side
1078lacks and transfers (close to) minimum set of objects.
1079
1080HTTP(S)::
1081        `http://remote.machine/path/to/repo.git/`
1082+
1083Downloader from http and https URL
1084first obtains the topmost commit object name from the remote site
1085by looking at the specified refname under `repo.git/refs/` directory,
1086and then tries to obtain the
1087commit object by downloading from `repo.git/objects/xx/xxx\...`
1088using the object name of that commit object.  Then it reads the
1089commit object to find out its parent commits and the associate
1090tree object; it repeats this process until it gets all the
1091necessary objects.  Because of this behavior, they are
1092sometimes also called 'commit walkers'.
1093+
1094The 'commit walkers' are sometimes also called 'dumb
1095transports', because they do not require any git aware smart
1096server like git Native transport does.  Any stock HTTP server
1097that does not even support directory index would suffice.  But
1098you must prepare your repository with `git-update-server-info`
1099to help dumb transport downloaders.
1100
1101Once you fetch from the remote repository, you `merge` that
1102with your current branch.
1103
1104However -- it's such a common thing to `fetch` and then
1105immediately `merge`, that it's called `git pull`, and you can
1106simply do
1107
1108----------------
1109$ git pull <remote-repository>
1110----------------
1111
1112and optionally give a branch-name for the remote end as a second
1113argument.
1114
1115[NOTE]
1116You could do without using any branches at all, by
1117keeping as many local repositories as you would like to have
1118branches, and merging between them with `git pull`, just like
1119you merge between branches. The advantage of this approach is
1120that it lets you keep a set of files for each `branch` checked
1121out and you may find it easier to switch back and forth if you
1122juggle multiple lines of development simultaneously. Of
1123course, you will pay the price of more disk usage to hold
1124multiple working trees, but disk space is cheap these days.
1125
1126It is likely that you will be pulling from the same remote
1127repository from time to time. As a short hand, you can store
1128the remote repository URL in the local repository's config file
1129like this:
1130
1131------------------------------------------------
1132$ git config remote.linus.url http://www.kernel.org/pub/scm/git/git.git/
1133------------------------------------------------
1134
1135and use the "linus" keyword with `git pull` instead of the full URL.
1136
1137Examples.
1138
1139. `git pull linus`
1140. `git pull linus tag v0.99.1`
1141
1142the above are equivalent to:
1143
1144. `git pull http://www.kernel.org/pub/scm/git/git.git/ HEAD`
1145. `git pull http://www.kernel.org/pub/scm/git/git.git/ tag v0.99.1`
1146
1147
1148How does the merge work?
1149------------------------
1150
1151We said this tutorial shows what plumbing does to help you cope
1152with the porcelain that isn't flushing, but we so far did not
1153talk about how the merge really works.  If you are following
1154this tutorial the first time, I'd suggest to skip to "Publishing
1155your work" section and come back here later.
1156
1157OK, still with me?  To give us an example to look at, let's go
1158back to the earlier repository with "hello" and "example" file,
1159and bring ourselves back to the pre-merge state:
1160
1161------------
1162$ git show-branch --more=2 master mybranch
1163! [master] Merge work in mybranch
1164 * [mybranch] Merge work in mybranch
1165--
1166-- [master] Merge work in mybranch
1167+* [master^2] Some work.
1168+* [master^] Some fun.
1169------------
1170
1171Remember, before running `git merge`, our `master` head was at
1172"Some fun." commit, while our `mybranch` head was at "Some
1173work." commit.
1174
1175------------
1176$ git checkout mybranch
1177$ git reset --hard master^2
1178$ git checkout master
1179$ git reset --hard master^
1180------------
1181
1182After rewinding, the commit structure should look like this:
1183
1184------------
1185$ git show-branch
1186* [master] Some fun.
1187 ! [mybranch] Some work.
1188--
1189 + [mybranch] Some work.
1190*  [master] Some fun.
1191*+ [mybranch^] New day.
1192------------
1193
1194Now we are ready to experiment with the merge by hand.
1195
1196`git merge` command, when merging two branches, uses 3-way merge
1197algorithm.  First, it finds the common ancestor between them.
1198The command it uses is `git-merge-base`:
1199
1200------------
1201$ mb=$(git-merge-base HEAD mybranch)
1202------------
1203
1204The command writes the commit object name of the common ancestor
1205to the standard output, so we captured its output to a variable,
1206because we will be using it in the next step.  By the way, the common
1207ancestor commit is the "New day." commit in this case.  You can
1208tell it by:
1209
1210------------
1211$ git-name-rev $mb
1212my-first-tag
1213------------
1214
1215After finding out a common ancestor commit, the second step is
1216this:
1217
1218------------
1219$ git-read-tree -m -u $mb HEAD mybranch
1220------------
1221
1222This is the same `git-read-tree` command we have already seen,
1223but it takes three trees, unlike previous examples.  This reads
1224the contents of each tree into different 'stage' in the index
1225file (the first tree goes to stage 1, the second to stage 2,
1226etc.).  After reading three trees into three stages, the paths
1227that are the same in all three stages are 'collapsed' into stage
12280.  Also paths that are the same in two of three stages are
1229collapsed into stage 0, taking the SHA1 from either stage 2 or
1230stage 3, whichever is different from stage 1 (i.e. only one side
1231changed from the common ancestor).
1232
1233After 'collapsing' operation, paths that are different in three
1234trees are left in non-zero stages.  At this point, you can
1235inspect the index file with this command:
1236
1237------------
1238$ git-ls-files --stage
1239100644 7f8b141b65fdcee47321e399a2598a235a032422 0       example
1240100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1       hello
1241100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2       hello
1242100644 cc44c73eb783565da5831b4d820c962954019b69 3       hello
1243------------
1244
1245In our example of only two files, we did not have unchanged
1246files so only 'example' resulted in collapsing, but in real-life
1247large projects, only small number of files change in one commit,
1248and this 'collapsing' tends to trivially merge most of the paths
1249fairly quickly, leaving only a handful the real changes in non-zero
1250stages.
1251
1252To look at only non-zero stages, use `\--unmerged` flag:
1253
1254------------
1255$ git-ls-files --unmerged
1256100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1       hello
1257100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2       hello
1258100644 cc44c73eb783565da5831b4d820c962954019b69 3       hello
1259------------
1260
1261The next step of merging is to merge these three versions of the
1262file, using 3-way merge.  This is done by giving
1263`git-merge-one-file` command as one of the arguments to
1264`git-merge-index` command:
1265
1266------------
1267$ git-merge-index git-merge-one-file hello
1268Auto-merging hello.
1269merge: warning: conflicts during merge
1270ERROR: Merge conflict in hello.
1271fatal: merge program failed
1272------------
1273
1274`git-merge-one-file` script is called with parameters to
1275describe those three versions, and is responsible to leave the
1276merge results in the working tree.
1277It is a fairly straightforward shell script, and
1278eventually calls `merge` program from RCS suite to perform a
1279file-level 3-way merge.  In this case, `merge` detects
1280conflicts, and the merge result with conflict marks is left in
1281the working tree..  This can be seen if you run `ls-files
1282--stage` again at this point:
1283
1284------------
1285$ git-ls-files --stage
1286100644 7f8b141b65fdcee47321e399a2598a235a032422 0       example
1287100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1       hello
1288100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2       hello
1289100644 cc44c73eb783565da5831b4d820c962954019b69 3       hello
1290------------
1291
1292This is the state of the index file and the working file after
1293`git merge` returns control back to you, leaving the conflicting
1294merge for you to resolve.  Notice that the path `hello` is still
1295unmerged, and what you see with `git diff` at this point is
1296differences since stage 2 (i.e. your version).
1297
1298
1299Publishing your work
1300--------------------
1301
1302So, we can use somebody else's work from a remote repository, but
1303how can *you* prepare a repository to let other people pull from
1304it?
1305
1306You do your real work in your working tree that has your
1307primary repository hanging under it as its `.git` subdirectory.
1308You *could* make that repository accessible remotely and ask
1309people to pull from it, but in practice that is not the way
1310things are usually done. A recommended way is to have a public
1311repository, make it reachable by other people, and when the
1312changes you made in your primary working tree are in good shape,
1313update the public repository from it. This is often called
1314'pushing'.
1315
1316[NOTE]
1317This public repository could further be mirrored, and that is
1318how git repositories at `kernel.org` are managed.
1319
1320Publishing the changes from your local (private) repository to
1321your remote (public) repository requires a write privilege on
1322the remote machine. You need to have an SSH account there to
1323run a single command, `git-receive-pack`.
1324
1325First, you need to create an empty repository on the remote
1326machine that will house your public repository. This empty
1327repository will be populated and be kept up-to-date by pushing
1328into it later. Obviously, this repository creation needs to be
1329done only once.
1330
1331[NOTE]
1332`git push` uses a pair of programs,
1333`git-send-pack` on your local machine, and `git-receive-pack`
1334on the remote machine. The communication between the two over
1335the network internally uses an SSH connection.
1336
1337Your private repository's git directory is usually `.git`, but
1338your public repository is often named after the project name,
1339i.e. `<project>.git`. Let's create such a public repository for
1340project `my-git`. After logging into the remote machine, create
1341an empty directory:
1342
1343------------
1344$ mkdir my-git.git
1345------------
1346
1347Then, make that directory into a git repository by running
1348`git init`, but this time, since its name is not the usual
1349`.git`, we do things slightly differently:
1350
1351------------
1352$ GIT_DIR=my-git.git git-init
1353------------
1354
1355Make sure this directory is available for others you want your
1356changes to be pulled by via the transport of your choice. Also
1357you need to make sure that you have the `git-receive-pack`
1358program on the `$PATH`.
1359
1360[NOTE]
1361Many installations of sshd do not invoke your shell as the login
1362shell when you directly run programs; what this means is that if
1363your login shell is `bash`, only `.bashrc` is read and not
1364`.bash_profile`. As a workaround, make sure `.bashrc` sets up
1365`$PATH` so that you can run `git-receive-pack` program.
1366
1367[NOTE]
1368If you plan to publish this repository to be accessed over http,
1369you should do `chmod +x my-git.git/hooks/post-update` at this
1370point.  This makes sure that every time you push into this
1371repository, `git-update-server-info` is run.
1372
1373Your "public repository" is now ready to accept your changes.
1374Come back to the machine you have your private repository. From
1375there, run this command:
1376
1377------------
1378$ git push <public-host>:/path/to/my-git.git master
1379------------
1380
1381This synchronizes your public repository to match the named
1382branch head (i.e. `master` in this case) and objects reachable
1383from them in your current repository.
1384
1385As a real example, this is how I update my public git
1386repository. Kernel.org mirror network takes care of the
1387propagation to other publicly visible machines:
1388
1389------------
1390$ git push master.kernel.org:/pub/scm/git/git.git/
1391------------
1392
1393
1394Packing your repository
1395-----------------------
1396
1397Earlier, we saw that one file under `.git/objects/??/` directory
1398is stored for each git object you create. This representation
1399is efficient to create atomically and safely, but
1400not so convenient to transport over the network. Since git objects are
1401immutable once they are created, there is a way to optimize the
1402storage by "packing them together". The command
1403
1404------------
1405$ git repack
1406------------
1407
1408will do it for you. If you followed the tutorial examples, you
1409would have accumulated about 17 objects in `.git/objects/??/`
1410directories by now. `git repack` tells you how many objects it
1411packed, and stores the packed file in `.git/objects/pack`
1412directory.
1413
1414[NOTE]
1415You will see two files, `pack-\*.pack` and `pack-\*.idx`,
1416in `.git/objects/pack` directory. They are closely related to
1417each other, and if you ever copy them by hand to a different
1418repository for whatever reason, you should make sure you copy
1419them together. The former holds all the data from the objects
1420in the pack, and the latter holds the index for random
1421access.
1422
1423If you are paranoid, running `git-verify-pack` command would
1424detect if you have a corrupt pack, but do not worry too much.
1425Our programs are always perfect ;-).
1426
1427Once you have packed objects, you do not need to leave the
1428unpacked objects that are contained in the pack file anymore.
1429
1430------------
1431$ git prune-packed
1432------------
1433
1434would remove them for you.
1435
1436You can try running `find .git/objects -type f` before and after
1437you run `git prune-packed` if you are curious.  Also `git
1438count-objects` would tell you how many unpacked objects are in
1439your repository and how much space they are consuming.
1440
1441[NOTE]
1442`git pull` is slightly cumbersome for HTTP transport, as a
1443packed repository may contain relatively few objects in a
1444relatively large pack. If you expect many HTTP pulls from your
1445public repository you might want to repack & prune often, or
1446never.
1447
1448If you run `git repack` again at this point, it will say
1449"Nothing to pack". Once you continue your development and
1450accumulate the changes, running `git repack` again will create a
1451new pack, that contains objects created since you packed your
1452repository the last time. We recommend that you pack your project
1453soon after the initial import (unless you are starting your
1454project from scratch), and then run `git repack` every once in a
1455while, depending on how active your project is.
1456
1457When a repository is synchronized via `git push` and `git pull`
1458objects packed in the source repository are usually stored
1459unpacked in the destination, unless rsync transport is used.
1460While this allows you to use different packing strategies on
1461both ends, it also means you may need to repack both
1462repositories every once in a while.
1463
1464
1465Working with Others
1466-------------------
1467
1468Although git is a truly distributed system, it is often
1469convenient to organize your project with an informal hierarchy
1470of developers. Linux kernel development is run this way. There
1471is a nice illustration (page 17, "Merges to Mainline") in
1472link:http://www.xenotime.net/linux/mentor/linux-mentoring-2006.pdf[Randy Dunlap's presentation].
1473
1474It should be stressed that this hierarchy is purely *informal*.
1475There is nothing fundamental in git that enforces the "chain of
1476patch flow" this hierarchy implies. You do not have to pull
1477from only one remote repository.
1478
1479A recommended workflow for a "project lead" goes like this:
1480
14811. Prepare your primary repository on your local machine. Your
1482   work is done there.
1483
14842. Prepare a public repository accessible to others.
1485+
1486If other people are pulling from your repository over dumb
1487transport protocols (HTTP), you need to keep this repository
1488'dumb transport friendly'.  After `git init`,
1489`$GIT_DIR/hooks/post-update` copied from the standard templates
1490would contain a call to `git-update-server-info` but the
1491`post-update` hook itself is disabled by default -- enable it
1492with `chmod +x post-update`.  This makes sure `git-update-server-info`
1493keeps the necessary files up-to-date.
1494
14953. Push into the public repository from your primary
1496   repository.
1497
14984. `git repack` the public repository. This establishes a big
1499   pack that contains the initial set of objects as the
1500   baseline, and possibly `git prune` if the transport
1501   used for pulling from your repository supports packed
1502   repositories.
1503
15045. Keep working in your primary repository. Your changes
1505   include modifications of your own, patches you receive via
1506   e-mails, and merges resulting from pulling the "public"
1507   repositories of your "subsystem maintainers".
1508+
1509You can repack this private repository whenever you feel like.
1510
15116. Push your changes to the public repository, and announce it
1512   to the public.
1513
15147. Every once in a while, "git repack" the public repository.
1515   Go back to step 5. and continue working.
1516
1517
1518A recommended work cycle for a "subsystem maintainer" who works
1519on that project and has an own "public repository" goes like this:
1520
15211. Prepare your work repository, by `git clone` the public
1522   repository of the "project lead". The URL used for the
1523   initial cloning is stored in the remote.origin.url
1524   configuration variable.
1525
15262. Prepare a public repository accessible to others, just like
1527   the "project lead" person does.
1528
15293. Copy over the packed files from "project lead" public
1530   repository to your public repository, unless the "project
1531   lead" repository lives on the same machine as yours.  In the
1532   latter case, you can use `objects/info/alternates` file to
1533   point at the repository you are borrowing from.
1534
15354. Push into the public repository from your primary
1536   repository. Run `git repack`, and possibly `git prune` if the
1537   transport used for pulling from your repository supports
1538   packed repositories.
1539
15405. Keep working in your primary repository. Your changes
1541   include modifications of your own, patches you receive via
1542   e-mails, and merges resulting from pulling the "public"
1543   repositories of your "project lead" and possibly your
1544   "sub-subsystem maintainers".
1545+
1546You can repack this private repository whenever you feel
1547like.
1548
15496. Push your changes to your public repository, and ask your
1550   "project lead" and possibly your "sub-subsystem
1551   maintainers" to pull from it.
1552
15537. Every once in a while, `git repack` the public repository.
1554   Go back to step 5. and continue working.
1555
1556
1557A recommended work cycle for an "individual developer" who does
1558not have a "public" repository is somewhat different. It goes
1559like this:
1560
15611. Prepare your work repository, by `git clone` the public
1562   repository of the "project lead" (or a "subsystem
1563   maintainer", if you work on a subsystem). The URL used for
1564   the initial cloning is stored in the remote.origin.url
1565   configuration variable.
1566
15672. Do your work in your repository on 'master' branch.
1568
15693. Run `git fetch origin` from the public repository of your
1570   upstream every once in a while. This does only the first
1571   half of `git pull` but does not merge. The head of the
1572   public repository is stored in `.git/refs/remotes/origin/master`.
1573
15744. Use `git cherry origin` to see which ones of your patches
1575   were accepted, and/or use `git rebase origin` to port your
1576   unmerged changes forward to the updated upstream.
1577
15785. Use `git format-patch origin` to prepare patches for e-mail
1579   submission to your upstream and send it out. Go back to
1580   step 2. and continue.
1581
1582
1583Working with Others, Shared Repository Style
1584--------------------------------------------
1585
1586If you are coming from CVS background, the style of cooperation
1587suggested in the previous section may be new to you. You do not
1588have to worry. git supports "shared public repository" style of
1589cooperation you are probably more familiar with as well.
1590
1591See linkgit:gitcvs-migration[7] for the details.
1592
1593Bundling your work together
1594---------------------------
1595
1596It is likely that you will be working on more than one thing at
1597a time.  It is easy to manage those more-or-less independent tasks
1598using branches with git.
1599
1600We have already seen how branches work previously,
1601with "fun and work" example using two branches.  The idea is the
1602same if there are more than two branches.  Let's say you started
1603out from "master" head, and have some new code in the "master"
1604branch, and two independent fixes in the "commit-fix" and
1605"diff-fix" branches:
1606
1607------------
1608$ git show-branch
1609! [commit-fix] Fix commit message normalization.
1610 ! [diff-fix] Fix rename detection.
1611  * [master] Release candidate #1
1612---
1613 +  [diff-fix] Fix rename detection.
1614 +  [diff-fix~1] Better common substring algorithm.
1615+   [commit-fix] Fix commit message normalization.
1616  * [master] Release candidate #1
1617++* [diff-fix~2] Pretty-print messages.
1618------------
1619
1620Both fixes are tested well, and at this point, you want to merge
1621in both of them.  You could merge in 'diff-fix' first and then
1622'commit-fix' next, like this:
1623
1624------------
1625$ git merge -m "Merge fix in diff-fix" diff-fix
1626$ git merge -m "Merge fix in commit-fix" commit-fix
1627------------
1628
1629Which would result in:
1630
1631------------
1632$ git show-branch
1633! [commit-fix] Fix commit message normalization.
1634 ! [diff-fix] Fix rename detection.
1635  * [master] Merge fix in commit-fix
1636---
1637  - [master] Merge fix in commit-fix
1638+ * [commit-fix] Fix commit message normalization.
1639  - [master~1] Merge fix in diff-fix
1640 +* [diff-fix] Fix rename detection.
1641 +* [diff-fix~1] Better common substring algorithm.
1642  * [master~2] Release candidate #1
1643++* [master~3] Pretty-print messages.
1644------------
1645
1646However, there is no particular reason to merge in one branch
1647first and the other next, when what you have are a set of truly
1648independent changes (if the order mattered, then they are not
1649independent by definition).  You could instead merge those two
1650branches into the current branch at once.  First let's undo what
1651we just did and start over.  We would want to get the master
1652branch before these two merges by resetting it to 'master~2':
1653
1654------------
1655$ git reset --hard master~2
1656------------
1657
1658You can make sure 'git show-branch' matches the state before
1659those two 'git merge' you just did.  Then, instead of running
1660two 'git merge' commands in a row, you would merge these two
1661branch heads (this is known as 'making an Octopus'):
1662
1663------------
1664$ git merge commit-fix diff-fix
1665$ git show-branch
1666! [commit-fix] Fix commit message normalization.
1667 ! [diff-fix] Fix rename detection.
1668  * [master] Octopus merge of branches 'diff-fix' and 'commit-fix'
1669---
1670  - [master] Octopus merge of branches 'diff-fix' and 'commit-fix'
1671+ * [commit-fix] Fix commit message normalization.
1672 +* [diff-fix] Fix rename detection.
1673 +* [diff-fix~1] Better common substring algorithm.
1674  * [master~1] Release candidate #1
1675++* [master~2] Pretty-print messages.
1676------------
1677
1678Note that you should not do Octopus because you can.  An octopus
1679is a valid thing to do and often makes it easier to view the
1680commit history if you are merging more than two independent
1681changes at the same time.  However, if you have merge conflicts
1682with any of the branches you are merging in and need to hand
1683resolve, that is an indication that the development happened in
1684those branches were not independent after all, and you should
1685merge two at a time, documenting how you resolved the conflicts,
1686and the reason why you preferred changes made in one side over
1687the other.  Otherwise it would make the project history harder
1688to follow, not easier.
1689
1690SEE ALSO
1691--------
1692linkgit:gittutorial[7], linkgit:gittutorial-2[7],
1693linkgit:giteveryday[7], linkgit:gitcvs-migration[7],
1694link:user-manual.html[The Git User's Manual]
1695
1696GIT
1697---
1698Part of the linkgit:git[1] suite.