Documentation / git-read-tree.txton commit repack: add --keep-pack option (ed7e5fc)
   1git-read-tree(1)
   2================
   3
   4NAME
   5----
   6git-read-tree - Reads tree information into the index
   7
   8
   9SYNOPSIS
  10--------
  11[verse]
  12'git read-tree' [[-m [--trivial] [--aggressive] | --reset | --prefix=<prefix>]
  13                [-u [--exclude-per-directory=<gitignore>] | -i]]
  14                [--index-output=<file>] [--no-sparse-checkout]
  15                (--empty | <tree-ish1> [<tree-ish2> [<tree-ish3>]])
  16
  17
  18DESCRIPTION
  19-----------
  20Reads the tree information given by <tree-ish> into the index,
  21but does not actually *update* any of the files it "caches". (see:
  22linkgit:git-checkout-index[1])
  23
  24Optionally, it can merge a tree into the index, perform a
  25fast-forward (i.e. 2-way) merge, or a 3-way merge, with the `-m`
  26flag.  When used with `-m`, the `-u` flag causes it to also update
  27the files in the work tree with the result of the merge.
  28
  29Trivial merges are done by 'git read-tree' itself.  Only conflicting paths
  30will be in unmerged state when 'git read-tree' returns.
  31
  32OPTIONS
  33-------
  34-m::
  35        Perform a merge, not just a read.  The command will
  36        refuse to run if your index file has unmerged entries,
  37        indicating that you have not finished previous merge you
  38        started.
  39
  40--reset::
  41        Same as -m, except that unmerged entries are discarded
  42        instead of failing.
  43
  44-u::
  45        After a successful merge, update the files in the work
  46        tree with the result of the merge.
  47
  48-i::
  49        Usually a merge requires the index file as well as the
  50        files in the working tree to be up to date with the
  51        current head commit, in order not to lose local
  52        changes.  This flag disables the check with the working
  53        tree and is meant to be used when creating a merge of
  54        trees that are not directly related to the current
  55        working tree status into a temporary index file.
  56
  57-n::
  58--dry-run::
  59        Check if the command would error out, without updating the index
  60        or the files in the working tree for real.
  61
  62-v::
  63        Show the progress of checking files out.
  64
  65--trivial::
  66        Restrict three-way merge by 'git read-tree' to happen
  67        only if there is no file-level merging required, instead
  68        of resolving merge for trivial cases and leaving
  69        conflicting files unresolved in the index.
  70
  71--aggressive::
  72        Usually a three-way merge by 'git read-tree' resolves
  73        the merge for really trivial cases and leaves other
  74        cases unresolved in the index, so that porcelains can
  75        implement different merge policies.  This flag makes the
  76        command resolve a few more cases internally:
  77+
  78* when one side removes a path and the other side leaves the path
  79  unmodified.  The resolution is to remove that path.
  80* when both sides remove a path.  The resolution is to remove that path.
  81* when both sides add a path identically.  The resolution
  82  is to add that path.
  83
  84--prefix=<prefix>::
  85        Keep the current index contents, and read the contents
  86        of the named tree-ish under the directory at `<prefix>`.
  87        The command will refuse to overwrite entries that already
  88        existed in the original index file.
  89
  90--exclude-per-directory=<gitignore>::
  91        When running the command with `-u` and `-m` options, the
  92        merge result may need to overwrite paths that are not
  93        tracked in the current branch.  The command usually
  94        refuses to proceed with the merge to avoid losing such a
  95        path.  However this safety valve sometimes gets in the
  96        way.  For example, it often happens that the other
  97        branch added a file that used to be a generated file in
  98        your branch, and the safety valve triggers when you try
  99        to switch to that branch after you ran `make` but before
 100        running `make clean` to remove the generated file.  This
 101        option tells the command to read per-directory exclude
 102        file (usually '.gitignore') and allows such an untracked
 103        but explicitly ignored file to be overwritten.
 104
 105--index-output=<file>::
 106        Instead of writing the results out to `$GIT_INDEX_FILE`,
 107        write the resulting index in the named file.  While the
 108        command is operating, the original index file is locked
 109        with the same mechanism as usual.  The file must allow
 110        to be rename(2)ed into from a temporary file that is
 111        created next to the usual index file; typically this
 112        means it needs to be on the same filesystem as the index
 113        file itself, and you need write permission to the
 114        directories the index file and index output file are
 115        located in.
 116
 117--[no-]recurse-submodules::
 118        Using --recurse-submodules will update the content of all initialized
 119        submodules according to the commit recorded in the superproject by
 120        calling read-tree recursively, also setting the submodules HEAD to be
 121        detached at that commit.
 122
 123--no-sparse-checkout::
 124        Disable sparse checkout support even if `core.sparseCheckout`
 125        is true.
 126
 127--empty::
 128        Instead of reading tree object(s) into the index, just empty
 129        it.
 130
 131<tree-ish#>::
 132        The id of the tree object(s) to be read/merged.
 133
 134
 135Merging
 136-------
 137If `-m` is specified, 'git read-tree' can perform 3 kinds of
 138merge, a single tree merge if only 1 tree is given, a
 139fast-forward merge with 2 trees, or a 3-way merge if 3 or more trees are
 140provided.
 141
 142
 143Single Tree Merge
 144~~~~~~~~~~~~~~~~~
 145If only 1 tree is specified, 'git read-tree' operates as if the user did not
 146specify `-m`, except that if the original index has an entry for a
 147given pathname, and the contents of the path match with the tree
 148being read, the stat info from the index is used. (In other words, the
 149index's stat()s take precedence over the merged tree's).
 150
 151That means that if you do a `git read-tree -m <newtree>` followed by a
 152`git checkout-index -f -u -a`, the 'git checkout-index' only checks out
 153the stuff that really changed.
 154
 155This is used to avoid unnecessary false hits when 'git diff-files' is
 156run after 'git read-tree'.
 157
 158
 159Two Tree Merge
 160~~~~~~~~~~~~~~
 161
 162Typically, this is invoked as `git read-tree -m $H $M`, where $H
 163is the head commit of the current repository, and $M is the head
 164of a foreign tree, which is simply ahead of $H (i.e. we are in a
 165fast-forward situation).
 166
 167When two trees are specified, the user is telling 'git read-tree'
 168the following:
 169
 170     1. The current index and work tree is derived from $H, but
 171        the user may have local changes in them since $H.
 172
 173     2. The user wants to fast-forward to $M.
 174
 175In this case, the `git read-tree -m $H $M` command makes sure
 176that no local change is lost as the result of this "merge".
 177Here are the "carry forward" rules, where "I" denotes the index,
 178"clean" means that index and work tree coincide, and "exists"/"nothing"
 179refer to the presence of a path in the specified commit:
 180
 181....
 182        I                   H        M        Result
 183       -------------------------------------------------------
 184     0  nothing             nothing  nothing  (does not happen)
 185     1  nothing             nothing  exists   use M
 186     2  nothing             exists   nothing  remove path from index
 187     3  nothing             exists   exists,  use M if "initial checkout",
 188                                     H == M   keep index otherwise
 189                                     exists,  fail
 190                                     H != M
 191
 192        clean I==H  I==M
 193       ------------------
 194     4  yes   N/A   N/A     nothing  nothing  keep index
 195     5  no    N/A   N/A     nothing  nothing  keep index
 196
 197     6  yes   N/A   yes     nothing  exists   keep index
 198     7  no    N/A   yes     nothing  exists   keep index
 199     8  yes   N/A   no      nothing  exists   fail
 200     9  no    N/A   no      nothing  exists   fail
 201
 202     10 yes   yes   N/A     exists   nothing  remove path from index
 203     11 no    yes   N/A     exists   nothing  fail
 204     12 yes   no    N/A     exists   nothing  fail
 205     13 no    no    N/A     exists   nothing  fail
 206
 207        clean (H==M)
 208       ------
 209     14 yes                 exists   exists   keep index
 210     15 no                  exists   exists   keep index
 211
 212        clean I==H  I==M (H!=M)
 213       ------------------
 214     16 yes   no    no      exists   exists   fail
 215     17 no    no    no      exists   exists   fail
 216     18 yes   no    yes     exists   exists   keep index
 217     19 no    no    yes     exists   exists   keep index
 218     20 yes   yes   no      exists   exists   use M
 219     21 no    yes   no      exists   exists   fail
 220....
 221
 222In all "keep index" cases, the index entry stays as in the
 223original index file.  If the entry is not up to date,
 224'git read-tree' keeps the copy in the work tree intact when
 225operating under the -u flag.
 226
 227When this form of 'git read-tree' returns successfully, you can
 228see which of the "local changes" that you made were carried forward by running
 229`git diff-index --cached $M`.  Note that this does not
 230necessarily match what `git diff-index --cached $H` would have
 231produced before such a two tree merge.  This is because of cases
 23218 and 19 --- if you already had the changes in $M (e.g. maybe
 233you picked it up via e-mail in a patch form), `git diff-index
 234--cached $H` would have told you about the change before this
 235merge, but it would not show in `git diff-index --cached $M`
 236output after the two-tree merge.
 237
 238Case 3 is slightly tricky and needs explanation.  The result from this
 239rule logically should be to remove the path if the user staged the removal
 240of the path and then switching to a new branch.  That however will prevent
 241the initial checkout from happening, so the rule is modified to use M (new
 242tree) only when the content of the index is empty.  Otherwise the removal
 243of the path is kept as long as $H and $M are the same.
 244
 2453-Way Merge
 246~~~~~~~~~~~
 247Each "index" entry has two bits worth of "stage" state. stage 0 is the
 248normal one, and is the only one you'd see in any kind of normal use.
 249
 250However, when you do 'git read-tree' with three trees, the "stage"
 251starts out at 1.
 252
 253This means that you can do
 254
 255----------------
 256$ git read-tree -m <tree1> <tree2> <tree3>
 257----------------
 258
 259and you will end up with an index with all of the <tree1> entries in
 260"stage1", all of the <tree2> entries in "stage2" and all of the
 261<tree3> entries in "stage3".  When performing a merge of another
 262branch into the current branch, we use the common ancestor tree
 263as <tree1>, the current branch head as <tree2>, and the other
 264branch head as <tree3>.
 265
 266Furthermore, 'git read-tree' has special-case logic that says: if you see
 267a file that matches in all respects in the following states, it
 268"collapses" back to "stage0":
 269
 270   - stage 2 and 3 are the same; take one or the other (it makes no
 271     difference - the same work has been done on our branch in
 272     stage 2 and their branch in stage 3)
 273
 274   - stage 1 and stage 2 are the same and stage 3 is different; take
 275     stage 3 (our branch in stage 2 did not do anything since the
 276     ancestor in stage 1 while their branch in stage 3 worked on
 277     it)
 278
 279   - stage 1 and stage 3 are the same and stage 2 is different take
 280     stage 2 (we did something while they did nothing)
 281
 282The 'git write-tree' command refuses to write a nonsensical tree, and it
 283will complain about unmerged entries if it sees a single entry that is not
 284stage 0.
 285
 286OK, this all sounds like a collection of totally nonsensical rules,
 287but it's actually exactly what you want in order to do a fast
 288merge. The different stages represent the "result tree" (stage 0, aka
 289"merged"), the original tree (stage 1, aka "orig"), and the two trees
 290you are trying to merge (stage 2 and 3 respectively).
 291
 292The order of stages 1, 2 and 3 (hence the order of three
 293<tree-ish> command-line arguments) are significant when you
 294start a 3-way merge with an index file that is already
 295populated.  Here is an outline of how the algorithm works:
 296
 297- if a file exists in identical format in all three trees, it will
 298  automatically collapse to "merged" state by 'git read-tree'.
 299
 300- a file that has _any_ difference what-so-ever in the three trees
 301  will stay as separate entries in the index. It's up to "porcelain
 302  policy" to determine how to remove the non-0 stages, and insert a
 303  merged version.
 304
 305- the index file saves and restores with all this information, so you
 306  can merge things incrementally, but as long as it has entries in
 307  stages 1/2/3 (i.e., "unmerged entries") you can't write the result. So
 308  now the merge algorithm ends up being really simple:
 309
 310  * you walk the index in order, and ignore all entries of stage 0,
 311    since they've already been done.
 312
 313  * if you find a "stage1", but no matching "stage2" or "stage3", you
 314    know it's been removed from both trees (it only existed in the
 315    original tree), and you remove that entry.
 316
 317  * if you find a matching "stage2" and "stage3" tree, you remove one
 318    of them, and turn the other into a "stage0" entry. Remove any
 319    matching "stage1" entry if it exists too.  .. all the normal
 320    trivial rules ..
 321
 322You would normally use 'git merge-index' with supplied
 323'git merge-one-file' to do this last step.  The script updates
 324the files in the working tree as it merges each path and at the
 325end of a successful merge.
 326
 327When you start a 3-way merge with an index file that is already
 328populated, it is assumed that it represents the state of the
 329files in your work tree, and you can even have files with
 330changes unrecorded in the index file.  It is further assumed
 331that this state is "derived" from the stage 2 tree.  The 3-way
 332merge refuses to run if it finds an entry in the original index
 333file that does not match stage 2.
 334
 335This is done to prevent you from losing your work-in-progress
 336changes, and mixing your random changes in an unrelated merge
 337commit.  To illustrate, suppose you start from what has been
 338committed last to your repository:
 339
 340----------------
 341$ JC=`git rev-parse --verify "HEAD^0"`
 342$ git checkout-index -f -u -a $JC
 343----------------
 344
 345You do random edits, without running 'git update-index'.  And then
 346you notice that the tip of your "upstream" tree has advanced
 347since you pulled from him:
 348
 349----------------
 350$ git fetch git://.... linus
 351$ LT=`git rev-parse FETCH_HEAD`
 352----------------
 353
 354Your work tree is still based on your HEAD ($JC), but you have
 355some edits since.  Three-way merge makes sure that you have not
 356added or modified index entries since $JC, and if you haven't,
 357then does the right thing.  So with the following sequence:
 358
 359----------------
 360$ git read-tree -m -u `git merge-base $JC $LT` $JC $LT
 361$ git merge-index git-merge-one-file -a
 362$ echo "Merge with Linus" | \
 363  git commit-tree `git write-tree` -p $JC -p $LT
 364----------------
 365
 366what you would commit is a pure merge between $JC and $LT without
 367your work-in-progress changes, and your work tree would be
 368updated to the result of the merge.
 369
 370However, if you have local changes in the working tree that
 371would be overwritten by this merge, 'git read-tree' will refuse
 372to run to prevent your changes from being lost.
 373
 374In other words, there is no need to worry about what exists only
 375in the working tree.  When you have local changes in a part of
 376the project that is not involved in the merge, your changes do
 377not interfere with the merge, and are kept intact.  When they
 378*do* interfere, the merge does not even start ('git read-tree'
 379complains loudly and fails without modifying anything).  In such
 380a case, you can simply continue doing what you were in the
 381middle of doing, and when your working tree is ready (i.e. you
 382have finished your work-in-progress), attempt the merge again.
 383
 384
 385Sparse checkout
 386---------------
 387
 388"Sparse checkout" allows populating the working directory sparsely.
 389It uses the skip-worktree bit (see linkgit:git-update-index[1]) to tell
 390Git whether a file in the working directory is worth looking at.
 391
 392'git read-tree' and other merge-based commands ('git merge', 'git
 393checkout'...) can help maintaining the skip-worktree bitmap and working
 394directory update. `$GIT_DIR/info/sparse-checkout` is used to
 395define the skip-worktree reference bitmap. When 'git read-tree' needs
 396to update the working directory, it resets the skip-worktree bit in the index
 397based on this file, which uses the same syntax as .gitignore files.
 398If an entry matches a pattern in this file, skip-worktree will not be
 399set on that entry. Otherwise, skip-worktree will be set.
 400
 401Then it compares the new skip-worktree value with the previous one. If
 402skip-worktree turns from set to unset, it will add the corresponding
 403file back. If it turns from unset to set, that file will be removed.
 404
 405While `$GIT_DIR/info/sparse-checkout` is usually used to specify what
 406files are in, you can also specify what files are _not_ in, using
 407negate patterns. For example, to remove the file `unwanted`:
 408
 409----------------
 410/*
 411!unwanted
 412----------------
 413
 414Another tricky thing is fully repopulating the working directory when you
 415no longer want sparse checkout. You cannot just disable "sparse
 416checkout" because skip-worktree bits are still in the index and your working
 417directory is still sparsely populated. You should re-populate the working
 418directory with the `$GIT_DIR/info/sparse-checkout` file content as
 419follows:
 420
 421----------------
 422/*
 423----------------
 424
 425Then you can disable sparse checkout. Sparse checkout support in 'git
 426read-tree' and similar commands is disabled by default. You need to
 427turn `core.sparseCheckout` on in order to have sparse checkout
 428support.
 429
 430
 431SEE ALSO
 432--------
 433linkgit:git-write-tree[1]; linkgit:git-ls-files[1];
 434linkgit:gitignore[5]
 435
 436GIT
 437---
 438Part of the linkgit:git[1] suite