Documentation / git-read-tree.txton commit Merge git://git.kernel.org/pub/scm/gitk/gitk (7fb23e6)
   1git-read-tree(1)
   2================
   3
   4NAME
   5----
   6git-read-tree - Reads tree information into the index
   7
   8
   9SYNOPSIS
  10--------
  11'git-read-tree' (<tree-ish> | [[-m [--aggressive]| --reset] [-u | -i]] <tree-ish1> [<tree-ish2> [<tree-ish3>]])
  12
  13
  14DESCRIPTION
  15-----------
  16Reads the tree information given by <tree-ish> into the index,
  17but does not actually *update* any of the files it "caches". (see:
  18gitlink:git-checkout-index[1])
  19
  20Optionally, it can merge a tree into the index, perform a
  21fast-forward (i.e. 2-way) merge, or a 3-way merge, with the `-m`
  22flag.  When used with `-m`, the `-u` flag causes it to also update
  23the files in the work tree with the result of the merge.
  24
  25Trivial merges are done by `git-read-tree` itself.  Only conflicting paths
  26will be in unmerged state when `git-read-tree` returns.
  27
  28OPTIONS
  29-------
  30-m::
  31        Perform a merge, not just a read.  The command will
  32        refuse to run if your index file has unmerged entries,
  33        indicating that you have not finished previous merge you
  34        started.
  35
  36--reset::
  37        Same as -m, except that unmerged entries are discarded
  38        instead of failing.
  39
  40-u::
  41        After a successful merge, update the files in the work
  42        tree with the result of the merge.
  43
  44-i::
  45        Usually a merge requires the index file as well as the
  46        files in the working tree are up to date with the
  47        current head commit, in order not to lose local
  48        changes.  This flag disables the check with the working
  49        tree and is meant to be used when creating a merge of
  50        trees that are not directly related to the current
  51        working tree status into a temporary index file.
  52
  53--aggressive::
  54        Usually a three-way merge by `git-read-tree` resolves
  55        the merge for really trivial cases and leaves other
  56        cases unresolved in the index, so that Porcelains can
  57        implement different merge policies.  This flag makes the
  58        command to resolve a few more cases internally:
  59+
  60* when one side removes a path and the other side leaves the path
  61  unmodified.  The resolution is to remove that path.
  62* when both sides remove a path.  The resolution is to remove that path.
  63* when both sides adds a path identically.  The resolution
  64  is to add that path.
  65
  66<tree-ish#>::
  67        The id of the tree object(s) to be read/merged.
  68
  69
  70Merging
  71-------
  72If `-m` is specified, `git-read-tree` can perform 3 kinds of
  73merge, a single tree merge if only 1 tree is given, a
  74fast-forward merge with 2 trees, or a 3-way merge if 3 trees are
  75provided.
  76
  77
  78Single Tree Merge
  79~~~~~~~~~~~~~~~~~
  80If only 1 tree is specified, git-read-tree operates as if the user did not
  81specify `-m`, except that if the original index has an entry for a
  82given pathname, and the contents of the path matches with the tree
  83being read, the stat info from the index is used. (In other words, the
  84index's stat()s take precedence over the merged tree's).
  85
  86That means that if you do a `git-read-tree -m <newtree>` followed by a
  87`git-checkout-index -f -u -a`, the `git-checkout-index` only checks out
  88the stuff that really changed.
  89
  90This is used to avoid unnecessary false hits when `git-diff-files` is
  91run after `git-read-tree`.
  92
  93
  94Two Tree Merge
  95~~~~~~~~~~~~~~
  96
  97Typically, this is invoked as `git-read-tree -m $H $M`, where $H
  98is the head commit of the current repository, and $M is the head
  99of a foreign tree, which is simply ahead of $H (i.e. we are in a
 100fast forward situation).
 101
 102When two trees are specified, the user is telling git-read-tree
 103the following:
 104
 105     1. The current index and work tree is derived from $H, but
 106        the user may have local changes in them since $H;
 107
 108     2. The user wants to fast-forward to $M.
 109
 110In this case, the `git-read-tree -m $H $M` command makes sure
 111that no local change is lost as the result of this "merge".
 112Here are the "carry forward" rules:
 113
 114        I (index)           H        M        Result
 115       -------------------------------------------------------
 116      0 nothing             nothing  nothing  (does not happen)
 117      1 nothing             nothing  exists   use M
 118      2 nothing             exists   nothing  remove path from index
 119      3 nothing             exists   exists   use M
 120
 121        clean I==H  I==M
 122       ------------------
 123      4 yes   N/A   N/A     nothing  nothing  keep index
 124      5 no    N/A   N/A     nothing  nothing  keep index
 125
 126      6 yes   N/A   yes     nothing  exists   keep index
 127      7 no    N/A   yes     nothing  exists   keep index
 128      8 yes   N/A   no      nothing  exists   fail
 129      9 no    N/A   no      nothing  exists   fail
 130
 131     10 yes   yes   N/A     exists   nothing  remove path from index
 132     11 no    yes   N/A     exists   nothing  fail
 133     12 yes   no    N/A     exists   nothing  fail
 134     13 no    no    N/A     exists   nothing  fail
 135
 136        clean (H=M)
 137       ------
 138     14 yes                 exists   exists   keep index
 139     15 no                  exists   exists   keep index
 140
 141        clean I==H  I==M (H!=M)
 142       ------------------
 143     16 yes   no    no      exists   exists   fail
 144     17 no    no    no      exists   exists   fail
 145     18 yes   no    yes     exists   exists   keep index
 146     19 no    no    yes     exists   exists   keep index
 147     20 yes   yes   no      exists   exists   use M
 148     21 no    yes   no      exists   exists   fail
 149
 150In all "keep index" cases, the index entry stays as in the
 151original index file.  If the entry were not up to date,
 152git-read-tree keeps the copy in the work tree intact when
 153operating under the -u flag.
 154
 155When this form of git-read-tree returns successfully, you can
 156see what "local changes" you made are carried forward by running
 157`git-diff-index --cached $M`.  Note that this does not
 158necessarily match `git-diff-index --cached $H` would have
 159produced before such a two tree merge.  This is because of cases
 16018 and 19 --- if you already had the changes in $M (e.g. maybe
 161you picked it up via e-mail in a patch form), `git-diff-index
 162--cached $H` would have told you about the change before this
 163merge, but it would not show in `git-diff-index --cached $M`
 164output after two-tree merge.
 165
 166
 1673-Way Merge
 168~~~~~~~~~~~
 169Each "index" entry has two bits worth of "stage" state. stage 0 is the
 170normal one, and is the only one you'd see in any kind of normal use.
 171
 172However, when you do `git-read-tree` with three trees, the "stage"
 173starts out at 1.
 174
 175This means that you can do
 176
 177----------------
 178$ git-read-tree -m <tree1> <tree2> <tree3>
 179----------------
 180
 181and you will end up with an index with all of the <tree1> entries in
 182"stage1", all of the <tree2> entries in "stage2" and all of the
 183<tree3> entries in "stage3".  When performing a merge of another
 184branch into the current branch, we use the common ancestor tree
 185as <tree1>, the current branch head as <tree2>, and the other
 186branch head as <tree3>.
 187
 188Furthermore, `git-read-tree` has special-case logic that says: if you see
 189a file that matches in all respects in the following states, it
 190"collapses" back to "stage0":
 191
 192   - stage 2 and 3 are the same; take one or the other (it makes no
 193     difference - the same work has been done on our branch in
 194     stage 2 and their branch in stage 3)
 195
 196   - stage 1 and stage 2 are the same and stage 3 is different; take
 197     stage 3 (our branch in stage 2 did not do anything since the
 198     ancestor in stage 1 while their branch in stage 3 worked on
 199     it)
 200
 201   - stage 1 and stage 3 are the same and stage 2 is different take
 202     stage 2 (we did something while they did nothing)
 203
 204The `git-write-tree` command refuses to write a nonsensical tree, and it
 205will complain about unmerged entries if it sees a single entry that is not
 206stage 0.
 207
 208OK, this all sounds like a collection of totally nonsensical rules,
 209but it's actually exactly what you want in order to do a fast
 210merge. The different stages represent the "result tree" (stage 0, aka
 211"merged"), the original tree (stage 1, aka "orig"), and the two trees
 212you are trying to merge (stage 2 and 3 respectively).
 213
 214The order of stages 1, 2 and 3 (hence the order of three
 215<tree-ish> command line arguments) are significant when you
 216start a 3-way merge with an index file that is already
 217populated.  Here is an outline of how the algorithm works:
 218
 219- if a file exists in identical format in all three trees, it will
 220  automatically collapse to "merged" state by git-read-tree.
 221
 222- a file that has _any_ difference what-so-ever in the three trees
 223  will stay as separate entries in the index. It's up to "porcelain
 224  policy" to determine how to remove the non-0 stages, and insert a
 225  merged version.
 226
 227- the index file saves and restores with all this information, so you
 228  can merge things incrementally, but as long as it has entries in
 229  stages 1/2/3 (i.e., "unmerged entries") you can't write the result. So
 230  now the merge algorithm ends up being really simple:
 231
 232  * you walk the index in order, and ignore all entries of stage 0,
 233    since they've already been done.
 234
 235  * if you find a "stage1", but no matching "stage2" or "stage3", you
 236    know it's been removed from both trees (it only existed in the
 237    original tree), and you remove that entry.
 238
 239  * if you find a matching "stage2" and "stage3" tree, you remove one
 240    of them, and turn the other into a "stage0" entry. Remove any
 241    matching "stage1" entry if it exists too.  .. all the normal
 242    trivial rules ..
 243
 244You would normally use `git-merge-index` with supplied
 245`git-merge-one-file` to do this last step.  The script updates
 246the files in the working tree as it merges each path and at the
 247end of a successful merge.
 248
 249When you start a 3-way merge with an index file that is already
 250populated, it is assumed that it represents the state of the
 251files in your work tree, and you can even have files with
 252changes unrecorded in the index file.  It is further assumed
 253that this state is "derived" from the stage 2 tree.  The 3-way
 254merge refuses to run if it finds an entry in the original index
 255file that does not match stage 2.
 256
 257This is done to prevent you from losing your work-in-progress
 258changes, and mixing your random changes in an unrelated merge
 259commit.  To illustrate, suppose you start from what has been
 260commited last to your repository:
 261
 262----------------
 263$ JC=`git-rev-parse --verify "HEAD^0"`
 264$ git-checkout-index -f -u -a $JC
 265----------------
 266
 267You do random edits, without running git-update-index.  And then
 268you notice that the tip of your "upstream" tree has advanced
 269since you pulled from him:
 270
 271----------------
 272$ git-fetch git://.... linus
 273$ LT=`cat .git/FETCH_HEAD`
 274----------------
 275
 276Your work tree is still based on your HEAD ($JC), but you have
 277some edits since.  Three-way merge makes sure that you have not
 278added or modified index entries since $JC, and if you haven't,
 279then does the right thing.  So with the following sequence:
 280
 281----------------
 282$ git-read-tree -m -u `git-merge-base $JC $LT` $JC $LT
 283$ git-merge-index git-merge-one-file -a
 284$ echo "Merge with Linus" | \
 285  git-commit-tree `git-write-tree` -p $JC -p $LT
 286----------------
 287
 288what you would commit is a pure merge between $JC and $LT without
 289your work-in-progress changes, and your work tree would be
 290updated to the result of the merge.
 291
 292However, if you have local changes in the working tree that
 293would be overwritten by this merge,`git-read-tree` will refuse
 294to run to prevent your changes from being lost.
 295
 296In other words, there is no need to worry about what exists only
 297in the working tree.  When you have local changes in a part of
 298the project that is not involved in the merge, your changes do
 299not interfere with the merge, and are kept intact.  When they
 300*do* interfere, the merge does not even start (`git-read-tree`
 301complains loudly and fails without modifying anything).  In such
 302a case, you can simply continue doing what you were in the
 303middle of doing, and when your working tree is ready (i.e. you
 304have finished your work-in-progress), attempt the merge again.
 305
 306
 307See Also
 308--------
 309gitlink:git-write-tree[1]; gitlink:git-ls-files[1]
 310
 311
 312Author
 313------
 314Written by Linus Torvalds <torvalds@osdl.org>
 315
 316Documentation
 317--------------
 318Documentation by David Greaves, Junio C Hamano and the git-list <git@vger.kernel.org>.
 319
 320GIT
 321---
 322Part of the gitlink:git[7] suite
 323