7 git-read-tree - Reads tree information into the directory cache
12 'git-read-tree' (<tree-ish> | [-m [-u]] <tree-ish1> [<tree-ish2> [<tree-ish3>]])
17 Reads the tree information given by <tree-ish> into the directory cache,
18 but does not actually *update* any of the files it "caches". (see:
21 Optionally, it can merge a tree into the cache, perform a
22 fast-forward (i.e. 2-way) merge, or a 3-way merge, with the -m
23 flag. When used with -m, the -u flag causes it to also update
24 the files in the work tree with the result of the merge.
26 Trivial merges are done by "git-read-tree" itself. Only conflicting paths
27 will be in unmerged state when "git-read-tree" returns.
32 Perform a merge, not just a read.
35 After a successful merge, update the files in the work
36 tree with the result of the merge.
39 The id of the tree object(s) to be read/merged.
44 If '-m' is specified, "git-read-tree" can perform 3 kinds of
45 merge, a single tree merge if only 1 tree is given, a
46 fast-forward merge with 2 trees, or a 3-way merge if 3 trees are
52 If only 1 tree is specified, git-read-tree operates as if the user did not
53 specify '-m', except that if the original cache has an entry for a
54 given pathname, and the contents of the path matches with the tree
55 being read, the stat info from the cache is used. (In other words, the
56 cache's stat()s take precedence over the merged tree's).
58 That means that if you do a "git-read-tree -m <newtree>" followed by a
59 "git-checkout-index -f -u -a", the "git-checkout-index" only checks out
60 the stuff that really changed.
62 This is used to avoid unnecessary false hits when "git-diff-files" is
63 run after git-read-tree.
69 Typically, this is invoked as "git-read-tree -m $H $M", where $H
70 is the head commit of the current repository, and $M is the head
71 of a foreign tree, which is simply ahead of $H (i.e. we are in a
72 fast forward situation).
74 When two trees are specified, the user is telling git-read-tree
77 (1) The current index and work tree is derived from $H, but
78 the user may have local changes in them since $H;
80 (2) The user wants to fast-forward to $M.
82 In this case, the "git-read-tree -m $H $M" command makes sure
83 that no local change is lost as the result of this "merge".
84 Here are the "carry forward" rules:
87 -------------------------------------------------------
88 0 nothing nothing nothing (does not happen)
89 1 nothing nothing exists use M
90 2 nothing exists nothing remove path from cache
91 3 nothing exists exists use M
95 4 yes N/A N/A nothing nothing keep index
96 5 no N/A N/A nothing nothing keep index
98 6 yes N/A yes nothing exists keep index
99 7 no N/A yes nothing exists keep index
100 8 yes N/A no nothing exists fail
101 9 no N/A no nothing exists fail
103 10 yes yes N/A exists nothing remove path from cache
104 11 no yes N/A exists nothing fail
105 12 yes no N/A exists nothing fail
106 13 no no N/A exists nothing fail
110 14 yes exists exists keep index
111 15 no exists exists keep index
113 clean I==H I==M (H!=M)
115 16 yes no no exists exists fail
116 17 no no no exists exists fail
117 18 yes no yes exists exists keep index
118 19 no no yes exists exists keep index
119 20 yes yes no exists exists use M
120 21 no yes no exists exists fail
122 In all "keep index" cases, the cache entry stays as in the
123 original index file. If the entry were not up to date,
124 git-read-tree keeps the copy in the work tree intact when
125 operating under the -u flag.
127 When this form of git-read-tree returns successfully, you can
128 see what "local changes" you made are carried forward by running
129 "git-diff-index --cached $M". Note that this does not
130 necessarily match "git-diff-index --cached $H" would have
131 produced before such a two tree merge. This is because of cases
132 18 and 19 --- if you already had the changes in $M (e.g. maybe
133 you picked it up via e-mail in a patch form), "git-diff-index
134 --cached $H" would have told you about the change before this
135 merge, but it would not show in "git-diff-index --cached $M"
136 output after two-tree merge.
141 Each "index" entry has two bits worth of "stage" state. stage 0 is the
142 normal one, and is the only one you'd see in any kind of normal use.
144 However, when you do "git-read-tree" with three trees, the "stage"
147 This means that you can do
149 git-read-tree -m <tree1> <tree2> <tree3>
151 and you will end up with an index with all of the <tree1> entries in
152 "stage1", all of the <tree2> entries in "stage2" and all of the
153 <tree3> entries in "stage3".
155 Furthermore, "git-read-tree" has special-case logic that says: if you see
156 a file that matches in all respects in the following states, it
157 "collapses" back to "stage0":
159 - stage 2 and 3 are the same; take one or the other (it makes no
160 difference - the same work has been done on stage 2 and 3)
162 - stage 1 and stage 2 are the same and stage 3 is different; take
163 stage 3 (some work has been done on stage 3)
165 - stage 1 and stage 3 are the same and stage 2 is different take
166 stage 2 (some work has been done on stage 2)
168 The "git-write-tree" command refuses to write a nonsensical tree, and it
169 will complain about unmerged entries if it sees a single entry that is not
172 Ok, this all sounds like a collection of totally nonsensical rules,
173 but it's actually exactly what you want in order to do a fast
174 merge. The different stages represent the "result tree" (stage 0, aka
175 "merged"), the original tree (stage 1, aka "orig"), and the two trees
176 you are trying to merge (stage 2 and 3 respectively).
178 The order of stages 1, 2 and 3 (hence the order of three
179 <tree-ish> command line arguments) are significant when you
180 start a 3-way merge with an index file that is already
181 populated. Here is an outline of how the algorithm works:
183 - if a file exists in identical format in all three trees, it will
184 automatically collapse to "merged" state by git-read-tree.
186 - a file that has _any_ difference what-so-ever in the three trees
187 will stay as separate entries in the index. It's up to "porcelain
188 policy" to determine how to remove the non-0 stages, and insert a
191 - the index file saves and restores with all this information, so you
192 can merge things incrementally, but as long as it has entries in
193 stages 1/2/3 (ie "unmerged entries") you can't write the result. So
194 now the merge algorithm ends up being really simple:
196 * you walk the index in order, and ignore all entries of stage 0,
197 since they've already been done.
199 * if you find a "stage1", but no matching "stage2" or "stage3", you
200 know it's been removed from both trees (it only existed in the
201 original tree), and you remove that entry.
203 * if you find a matching "stage2" and "stage3" tree, you remove one
204 of them, and turn the other into a "stage0" entry. Remove any
205 matching "stage1" entry if it exists too. .. all the normal
208 You would normally use "git-merge-index" with supplied
209 "git-merge-one-file" to do this last step. The script
210 does not touch the files in the work tree, and the entire merge
211 happens in the index file. In other words, there is no need to
212 worry about what is in the working directory, since it is never
213 shown and never used.
215 When you start a 3-way merge with an index file that is already
216 populated, it is assumed that it represents the state of the
217 files in your work tree, and you can even have files with
218 changes unrecorded in the index file. It is further assumed
219 that this state is "derived" from the stage 2 tree. The 3-way
220 merge refuses to run if it finds an entry in the original index
221 file that does not match stage 2.
223 This is done to prevent you from losing your work-in-progress
224 changes. To illustrate, suppose you start from what has been
225 commited last to your repository:
228 $ git-checkout-index -f -u -a $JC
230 You do random edits, without running git-update-index. And then
231 you notice that the tip of your "upstream" tree has advanced
232 since you pulled from him:
234 $ git-fetch rsync://.... linus
235 $ LT=`cat .git/MERGE_HEAD`
237 Your work tree is still based on your HEAD ($JC), but you have
238 some edits since. Three-way merge makes sure that you have not
239 added or modified cache entries since $JC, and if you haven't,
240 then does the right thing. So with the following sequence:
242 $ git-read-tree -m -u `git-merge-base $JC $LT` $JC $LT
243 $ git-merge-index git-merge-one-file -a
244 $ echo "Merge with Linus" | \
245 git-commit-tree `git-write-tree` -p $JC -p $LT
247 what you would commit is a pure merge between $JC and LT without
248 your work-in-progress changes, and your work tree would be
249 updated to the result of the merge.
254 link:git-write-tree.html[git-write-tree]; link:git-ls-files.html[git-ls-files]
259 Written by Linus Torvalds <torvalds@osdl.org>
263 Documentation by David Greaves, Junio C Hamano and the git-list <git@vger.kernel.org>.
267 Part of the link:git.html[git] suite