Documentation / git-read-tree.txton commit git-fetch-pack: Do not use git-rev-list (69779a5)
   1git-read-tree(1)
   2================
   3
   4NAME
   5----
   6git-read-tree - Reads tree information into the directory cache
   7
   8
   9SYNOPSIS
  10--------
  11'git-read-tree' (<tree-ish> | [-m [-u|-i]] <tree-ish1> [<tree-ish2> [<tree-ish3>]])
  12
  13
  14DESCRIPTION
  15-----------
  16Reads the tree information given by <tree-ish> into the directory cache,
  17but does not actually *update* any of the files it "caches". (see:
  18git-checkout-index)
  19
  20Optionally, it can merge a tree into the cache, 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.
  32
  33-u::
  34        After a successful merge, update the files in the work
  35        tree with the result of the merge.
  36
  37-i::
  38        Usually a merge requires the index file as well as the
  39        files in the working tree are up to date with the
  40        current head commit, in order not to lose local
  41        changes.  This flag disables the check with the working
  42        tree and is meant to be used when creating a merge of
  43        trees that are not directly related to the current
  44        working tree status into a temporary index file.
  45
  46
  47<tree-ish#>::
  48        The id of the tree object(s) to be read/merged.
  49
  50
  51Merging
  52-------
  53If '-m' is specified, "git-read-tree" can perform 3 kinds of
  54merge, a single tree merge if only 1 tree is given, a
  55fast-forward merge with 2 trees, or a 3-way merge if 3 trees are
  56provided.
  57
  58
  59Single Tree Merge
  60~~~~~~~~~~~~~~~~~
  61If only 1 tree is specified, git-read-tree operates as if the user did not
  62specify '-m', except that if the original cache has an entry for a
  63given pathname, and the contents of the path matches with the tree
  64being read, the stat info from the cache is used. (In other words, the
  65cache's stat()s take precedence over the merged tree's).
  66
  67That means that if you do a "git-read-tree -m <newtree>" followed by a
  68"git-checkout-index -f -u -a", the "git-checkout-index" only checks out
  69the stuff that really changed.
  70
  71This is used to avoid unnecessary false hits when "git-diff-files" is
  72run after git-read-tree.
  73
  74
  75Two Tree Merge
  76~~~~~~~~~~~~~~
  77
  78Typically, this is invoked as "git-read-tree -m $H $M", where $H
  79is the head commit of the current repository, and $M is the head
  80of a foreign tree, which is simply ahead of $H (i.e. we are in a
  81fast forward situation).
  82
  83When two trees are specified, the user is telling git-read-tree
  84the following:
  85
  86     1. The current index and work tree is derived from $H, but
  87        the user may have local changes in them since $H;
  88
  89     2. The user wants to fast-forward to $M.
  90
  91In this case, the "git-read-tree -m $H $M" command makes sure
  92that no local change is lost as the result of this "merge".
  93Here are the "carry forward" rules:
  94
  95        I (index)           H        M        Result
  96       -------------------------------------------------------
  97      0 nothing             nothing  nothing  (does not happen)
  98      1 nothing             nothing  exists   use M
  99      2 nothing             exists   nothing  remove path from cache
 100      3 nothing             exists   exists   use M
 101
 102        clean I==H  I==M
 103       ------------------
 104      4 yes   N/A   N/A     nothing  nothing  keep index
 105      5 no    N/A   N/A     nothing  nothing  keep index
 106
 107      6 yes   N/A   yes     nothing  exists   keep index
 108      7 no    N/A   yes     nothing  exists   keep index
 109      8 yes   N/A   no      nothing  exists   fail
 110      9 no    N/A   no      nothing  exists   fail
 111
 112     10 yes   yes   N/A     exists   nothing  remove path from cache
 113     11 no    yes   N/A     exists   nothing  fail
 114     12 yes   no    N/A     exists   nothing  fail
 115     13 no    no    N/A     exists   nothing  fail
 116
 117        clean (H=M)
 118       ------
 119     14 yes                 exists   exists   keep index
 120     15 no                  exists   exists   keep index
 121
 122        clean I==H  I==M (H!=M)
 123       ------------------
 124     16 yes   no    no      exists   exists   fail
 125     17 no    no    no      exists   exists   fail
 126     18 yes   no    yes     exists   exists   keep index
 127     19 no    no    yes     exists   exists   keep index
 128     20 yes   yes   no      exists   exists   use M
 129     21 no    yes   no      exists   exists   fail
 130
 131In all "keep index" cases, the cache entry stays as in the
 132original index file.  If the entry were not up to date,
 133git-read-tree keeps the copy in the work tree intact when
 134operating under the -u flag.
 135
 136When this form of git-read-tree returns successfully, you can
 137see what "local changes" you made are carried forward by running
 138"git-diff-index --cached $M".  Note that this does not
 139necessarily match "git-diff-index --cached $H" would have
 140produced before such a two tree merge.  This is because of cases
 14118 and 19 --- if you already had the changes in $M (e.g. maybe
 142you picked it up via e-mail in a patch form), "git-diff-index
 143--cached $H" would have told you about the change before this
 144merge, but it would not show in "git-diff-index --cached $M"
 145output after two-tree merge.
 146
 147
 1483-Way Merge
 149~~~~~~~~~~~
 150Each "index" entry has two bits worth of "stage" state. stage 0 is the
 151normal one, and is the only one you'd see in any kind of normal use.
 152
 153However, when you do "git-read-tree" with three trees, the "stage"
 154starts out at 1.
 155
 156This means that you can do
 157
 158        git-read-tree -m <tree1> <tree2> <tree3>
 159
 160and you will end up with an index with all of the <tree1> entries in
 161"stage1", all of the <tree2> entries in "stage2" and all of the
 162<tree3> entries in "stage3".
 163
 164Furthermore, "git-read-tree" has special-case logic that says: if you see
 165a file that matches in all respects in the following states, it
 166"collapses" back to "stage0":
 167
 168   - stage 2 and 3 are the same; take one or the other (it makes no
 169     difference - the same work has been done on stage 2 and 3)
 170
 171   - stage 1 and stage 2 are the same and stage 3 is different; take
 172     stage 3 (some work has been done on stage 3)
 173
 174   - stage 1 and stage 3 are the same and stage 2 is different take
 175     stage 2 (some work has been done on stage 2)
 176
 177The "git-write-tree" command refuses to write a nonsensical tree, and it
 178will complain about unmerged entries if it sees a single entry that is not
 179stage 0.
 180
 181Ok, this all sounds like a collection of totally nonsensical rules,
 182but it's actually exactly what you want in order to do a fast
 183merge. The different stages represent the "result tree" (stage 0, aka
 184"merged"), the original tree (stage 1, aka "orig"), and the two trees
 185you are trying to merge (stage 2 and 3 respectively).
 186
 187The order of stages 1, 2 and 3 (hence the order of three
 188<tree-ish> command line arguments) are significant when you
 189start a 3-way merge with an index file that is already
 190populated.  Here is an outline of how the algorithm works:
 191
 192- if a file exists in identical format in all three trees, it will
 193  automatically collapse to "merged" state by git-read-tree.
 194
 195- a file that has _any_ difference what-so-ever in the three trees
 196  will stay as separate entries in the index. It's up to "porcelain
 197  policy" to determine how to remove the non-0 stages, and insert a
 198  merged version.
 199
 200- the index file saves and restores with all this information, so you
 201  can merge things incrementally, but as long as it has entries in
 202  stages 1/2/3 (ie "unmerged entries") you can't write the result. So
 203  now the merge algorithm ends up being really simple:
 204
 205  * you walk the index in order, and ignore all entries of stage 0,
 206    since they've already been done.
 207
 208  * if you find a "stage1", but no matching "stage2" or "stage3", you
 209    know it's been removed from both trees (it only existed in the
 210    original tree), and you remove that entry.
 211
 212  * if you find a matching "stage2" and "stage3" tree, you remove one
 213    of them, and turn the other into a "stage0" entry. Remove any
 214    matching "stage1" entry if it exists too.  .. all the normal
 215    trivial rules ..
 216
 217You would normally use "git-merge-index" with supplied
 218"git-merge-one-file" to do this last step.  The script
 219does not touch the files in the work tree, and the entire merge
 220happens in the index file.  In other words, there is no need to
 221worry about what is in the working directory, since it is never
 222shown and never used.
 223
 224When you start a 3-way merge with an index file that is already
 225populated, it is assumed that it represents the state of the
 226files in your work tree, and you can even have files with
 227changes unrecorded in the index file.  It is further assumed
 228that this state is "derived" from the stage 2 tree.  The 3-way
 229merge refuses to run if it finds an entry in the original index
 230file that does not match stage 2.
 231
 232This is done to prevent you from losing your work-in-progress
 233changes.  To illustrate, suppose you start from what has been
 234commited last to your repository:
 235
 236    $ JC=`cat .git/HEAD`
 237    $ git-checkout-index -f -u -a $JC
 238
 239You do random edits, without running git-update-index.  And then
 240you notice that the tip of your "upstream" tree has advanced
 241since you pulled from him:
 242
 243    $ git-fetch rsync://.... linus
 244    $ LT=`cat .git/MERGE_HEAD`
 245
 246Your work tree is still based on your HEAD ($JC), but you have
 247some edits since.  Three-way merge makes sure that you have not
 248added or modified cache entries since $JC, and if you haven't,
 249then does the right thing.  So with the following sequence:
 250
 251    $ git-read-tree -m -u `git-merge-base $JC $LT` $JC $LT
 252    $ git-merge-index git-merge-one-file -a
 253    $ echo "Merge with Linus" | \
 254      git-commit-tree `git-write-tree` -p $JC -p $LT
 255
 256what you would commit is a pure merge between $JC and LT without
 257your work-in-progress changes, and your work tree would be
 258updated to the result of the merge.
 259
 260
 261See Also
 262--------
 263gitlink:git-write-tree[1]; gitlink:git-ls-files[1]
 264
 265
 266Author
 267------
 268Written by Linus Torvalds <torvalds@osdl.org>
 269
 270Documentation
 271--------------
 272Documentation by David Greaves, Junio C Hamano and the git-list <git@vger.kernel.org>.
 273
 274GIT
 275---
 276Part of the gitlink:git[7] suite
 277