1GIT-READ-TREE(1) Git Manual GIT-READ-TREE(1)
2
6 git-read-tree - Reads tree information into the index
7
9 git read-tree [[-m [--trivial] [--aggressive] | --reset |
10 --prefix=<prefix>] [-u [--exclude-per-directory=<gitignore>] | -i]]
11 [--index-output=<file>] [--no-sparse-checkout] (--empty | <tree-ish1>
12 [<tree-ish2> [<tree-ish3>]])
13
15 Reads the tree information given by <tree-ish> into the index, but does
16 not actually update any of the files it "caches". (see: git-checkout-
17 index(1))
18 Optionally, it can merge a tree into the index, perform a fast-forward
20 (i.e. 2-way) merge, or a 3-way merge, with the -m flag. When used with
21 -m, the -u flag causes it to also update the files in the work tree
22 with the result of the merge.
23 Trivial merges are done by git read-tree itself. Only conflicting paths
25 will be in unmerged state when git read-tree returns.
26
28 -m
29 Perform a merge, not just a read. The command will refuse to run if
30 your index file has unmerged entries, indicating that you have not
31 finished previous merge you started.
32 --reset
34 Same as -m, except that unmerged entries are discarded instead of
35 failing.
36 -u
38 After a successful merge, update the files in the work tree with
39 the result of the merge.
40 -i
42 Usually a merge requires the index file as well as the files in the
43 working tree are up to date with the current head commit, in order
44 not to lose local changes. This flag disables the check with the
45 working tree and is meant to be used when creating a merge of trees
46 that are not directly related to the current working tree status
47 into a temporary index file.
48 -v
50 Show the progress of checking files out.
51 --trivial
53 Restrict three-way merge by git read-tree to happen only if there
54 is no file-level merging required, instead of resolving merge for
55 trivial cases and leaving conflicting files unresolved in the
56 index.
57 --aggressive
59 Usually a three-way merge by git read-tree resolves the merge for
60 really trivial cases and leaves other cases unresolved in the
61 index, so that Porcelains can implement different merge policies.
62 This flag makes the command to resolve a few more cases internally:
63 · when one side removes a path and the other side leaves the path
65 unmodified. The resolution is to remove that path.
66 · when both sides remove a path. The resolution is to remove that
68 path.
69 · when both sides adds a path identically. The resolution is to
71 add that path.
72 --prefix=<prefix>/
74 Keep the current index contents, and read the contents of named
75 tree-ish under directory at <prefix>. The original index file
76 cannot have anything at the path <prefix> itself, and have nothing
77 in <prefix>/ directory. Note that the <prefix>/ value must end with
78 a slash.
79 --exclude-per-directory=<gitignore>
81 When running the command with -u and -m options, the merge result
82 may need to overwrite paths that are not tracked in the current
83 branch. The command usually refuses to proceed with the merge to
84 avoid losing such a path. However this safety valve sometimes gets
85 in the way. For example, it often happens that the other branch
86 added a file that used to be a generated file in your branch, and
87 the safety valve triggers when you try to switch to that branch
88 after you ran make but before running make clean to remove the
89 generated file. This option tells the command to read per-directory
90 exclude file (usually .gitignore) and allows such an untracked but
91 explicitly ignored file to be overwritten.
92 --index-output=<file>
94 Instead of writing the results out to $GIT_INDEX_FILE, write the
95 resulting index in the named file. While the command is operating,
96 the original index file is locked with the same mechanism as usual.
97 The file must allow to be rename(2)ed into from a temporary file
98 that is created next to the usual index file; typically this means
99 it needs to be on the same filesystem as the index file itself, and
100 you need write permission to the directories the index file and
101 index output file are located in.
102 --no-sparse-checkout
104 Disable sparse checkout support even if core.sparseCheckout is
105 true.
106 --empty
108 Instead of reading tree object(s) into the index, just empty it.
109 <tree-ish#>
111 The id of the tree object(s) to be read/merged.
112
114 If -m is specified, git read-tree can perform 3 kinds of merge, a
115 single tree merge if only 1 tree is given, a fast-forward merge with 2
116 trees, or a 3-way merge if 3 trees are provided.
117 Single Tree Merge
119 If only 1 tree is specified, git read-tree operates as if the user did
120 not specify -m, except that if the original index has an entry for a
121 given pathname, and the contents of the path match with the tree being
122 read, the stat info from the index is used. (In other words, the
123 index’s stat()s take precedence over the merged tree’s).
124 That means that if you do a git read-tree -m <newtree> followed by a
126 git checkout-index -f -u -a, the git checkout-index only checks out the
127 stuff that really changed.
128 This is used to avoid unnecessary false hits when git diff-files is run
130 after git read-tree.
131 Two Tree Merge
133 Typically, this is invoked as git read-tree -m $H $M, where $H is the
134 head commit of the current repository, and $M is the head of a foreign
135 tree, which is simply ahead of $H (i.e. we are in a fast-forward
136 situation).
137 When two trees are specified, the user is telling git read-tree the
139 following:
140 1. The current index and work tree is derived from $H, but the user
142 may have local changes in them since $H.
143 2. The user wants to fast-forward to $M.
145 In this case, the git read-tree -m $H $M command makes sure that no
147 local change is lost as the result of this "merge". Here are the "carry
148 forward" rules, where "I" denotes the index, "clean" means that index
149 and work tree coincide, and "exists"/"nothing" refer to the presence of
150 a path in the specified commit:
151 I H M Result
153 -------------------------------------------------------
154 0 nothing nothing nothing (does not happen)
155 1 nothing nothing exists use M
156 2 nothing exists nothing remove path from index
157 3 nothing exists exists, use M if "initial checkout",
158 H == M keep index otherwise
159 exists, fail
160 H != M
161 clean I==H I==M
163 ------------------
164 4 yes N/A N/A nothing nothing keep index
165 5 no N/A N/A nothing nothing keep index
166 6 yes N/A yes nothing exists keep index
168 7 no N/A yes nothing exists keep index
169 8 yes N/A no nothing exists fail
170 9 no N/A no nothing exists fail
171 10 yes yes N/A exists nothing remove path from index
173 11 no yes N/A exists nothing fail
174 12 yes no N/A exists nothing fail
175 13 no no N/A exists nothing fail
176 clean (H==M)
178 ------
179 14 yes exists exists keep index
180 15 no exists exists keep index
181 clean I==H I==M (H!=M)
183 ------------------
184 16 yes no no exists exists fail
185 17 no no no exists exists fail
186 18 yes no yes exists exists keep index
187 19 no no yes exists exists keep index
188 20 yes yes no exists exists use M
189 21 no yes no exists exists fail
190 In all "keep index" cases, the index entry stays as in the original
192 index file. If the entry is not up to date, git read-tree keeps the
193 copy in the work tree intact when operating under the -u flag.
194 When this form of git read-tree returns successfully, you can see which
196 of the "local changes" that you made were carried forward by running
197 git diff-index --cached $M. Note that this does not necessarily match
198 what git diff-index --cached $H would have produced before such a two
199 tree merge. This is because of cases 18 and 19 --- if you already had
200 the changes in $M (e.g. maybe you picked it up via e-mail in a patch
201 form), git diff-index --cached $H would have told you about the change
202 before this merge, but it would not show in git diff-index --cached $M
203 output after the two-tree merge.
204 Case 3 is slightly tricky and needs explanation. The result from this
206 rule logically should be to remove the path if the user staged the
207 removal of the path and then switching to a new branch. That however
208 will prevent the initial checkout from happening, so the rule is
209 modified to use M (new tree) only when the content of the index is
210 empty. Otherwise the removal of the path is kept as long as $H and $M
211 are the same.
212 3-Way Merge
214 Each "index" entry has two bits worth of "stage" state. stage 0 is the
215 normal one, and is the only one you’d see in any kind of normal use.
216 However, when you do git read-tree with three trees, the "stage" starts
218 out at 1.
219 This means that you can do
221 $ git read-tree -m <tree1> <tree2> <tree3>
223 and you will end up with an index with all of the <tree1> entries in
226 "stage1", all of the <tree2> entries in "stage2" and all of the <tree3>
227 entries in "stage3". When performing a merge of another branch into the
228 current branch, we use the common ancestor tree as <tree1>, the current
229 branch head as <tree2>, and the other branch head as <tree3>.
230 Furthermore, git read-tree has special-case logic that says: if you see
232 a file that matches in all respects in the following states, it
233 "collapses" back to "stage0":
234 · stage 2 and 3 are the same; take one or the other (it makes no
236 difference - the same work has been done on our branch in stage 2
237 and their branch in stage 3)
238 · stage 1 and stage 2 are the same and stage 3 is different; take
240 stage 3 (our branch in stage 2 did not do anything since the
241 ancestor in stage 1 while their branch in stage 3 worked on it)
242 · stage 1 and stage 3 are the same and stage 2 is different take
244 stage 2 (we did something while they did nothing)
245 The git write-tree command refuses to write a nonsensical tree, and it
247 will complain about unmerged entries if it sees a single entry that is
248 not stage 0.
249 OK, this all sounds like a collection of totally nonsensical rules, but
251 it’s actually exactly what you want in order to do a fast merge. The
252 different stages represent the "result tree" (stage 0, aka "merged"),
253 the original tree (stage 1, aka "orig"), and the two trees you are
254 trying to merge (stage 2 and 3 respectively).
255 The order of stages 1, 2 and 3 (hence the order of three <tree-ish>
257 command line arguments) are significant when you start a 3-way merge
258 with an index file that is already populated. Here is an outline of how
259 the algorithm works:
260 · if a file exists in identical format in all three trees, it will
262 automatically collapse to "merged" state by git read-tree.
263 · a file that has any difference what-so-ever in the three trees will
265 stay as separate entries in the index. It’s up to "porcelain
266 policy" to determine how to remove the non-0 stages, and insert a
267 merged version.
268 · the index file saves and restores with all this information, so you
270 can merge things incrementally, but as long as it has entries in
271 stages 1/2/3 (i.e., "unmerged entries") you can’t write the result.
272 So now the merge algorithm ends up being really simple:
273 · you walk the index in order, and ignore all entries of stage 0,
275 since they’ve already been done.
276 · if you find a "stage1", but no matching "stage2" or "stage3",
278 you know it’s been removed from both trees (it only existed in
279 the original tree), and you remove that entry.
280 · if you find a matching "stage2" and "stage3" tree, you remove
282 one of them, and turn the other into a "stage0" entry. Remove
283 any matching "stage1" entry if it exists too. .. all the normal
284 trivial rules ..
285 You would normally use git merge-index with supplied git merge-one-file
287 to do this last step. The script updates the files in the working tree
288 as it merges each path and at the end of a successful merge.
289 When you start a 3-way merge with an index file that is already
291 populated, it is assumed that it represents the state of the files in
292 your work tree, and you can even have files with changes unrecorded in
293 the index file. It is further assumed that this state is "derived" from
294 the stage 2 tree. The 3-way merge refuses to run if it finds an entry
295 in the original index file that does not match stage 2.
296 This is done to prevent you from losing your work-in-progress changes,
298 and mixing your random changes in an unrelated merge commit. To
299 illustrate, suppose you start from what has been committed last to your
300 repository:
301 $ JC=`git rev-parse --verify "HEAD^0"`
303 $ git checkout-index -f -u -a $JC
304 You do random edits, without running git update-index. And then you
307 notice that the tip of your "upstream" tree has advanced since you
308 pulled from him:
309 $ git fetch git://.... linus
311 $ LT=`cat .git/FETCH_HEAD`
312 Your work tree is still based on your HEAD ($JC), but you have some
315 edits since. Three-way merge makes sure that you have not added or
316 modified index entries since $JC, and if you haven’t, then does the
317 right thing. So with the following sequence:
318 $ git read-tree -m -u `git merge-base $JC $LT` $JC $LT
320 $ git merge-index git-merge-one-file -a
321 $ echo "Merge with Linus" | \
322 git commit-tree `git write-tree` -p $JC -p $LT
323 what you would commit is a pure merge between $JC and $LT without your
326 work-in-progress changes, and your work tree would be updated to the
327 result of the merge.
328 However, if you have local changes in the working tree that would be
330 overwritten by this merge, git read-tree will refuse to run to prevent
331 your changes from being lost.
332 In other words, there is no need to worry about what exists only in the
334 working tree. When you have local changes in a part of the project that
335 is not involved in the merge, your changes do not interfere with the
336 merge, and are kept intact. When they do interfere, the merge does not
337 even start (git read-tree complains loudly and fails without modifying
338 anything). In such a case, you can simply continue doing what you were
339 in the middle of doing, and when your working tree is ready (i.e. you
340 have finished your work-in-progress), attempt the merge again.
341
343 "Sparse checkout" allows to sparsely populate working directory. It
344 uses skip-worktree bit (see git-update-index(1)) to tell Git whether a
345 file on working directory is worth looking at.
346 "git read-tree" and other merge-based commands ("git merge", "git
348 checkout"...) can help maintaining skip-worktree bitmap and working
349 directory update. $GIT_DIR/info/sparse-checkout is used to define the
350 skip-worktree reference bitmap. When "git read-tree" needs to update
351 working directory, it will reset skip-worktree bit in index based on
352 this file, which uses the same syntax as .gitignore files. If an entry
353 matches a pattern in this file, skip-worktree will be set on that
354 entry. Otherwise, skip-worktree will be unset.
355 Then it compares the new skip-worktree value with the previous one. If
357 skip-worktree turns from unset to set, it will add the corresponding
358 file back. If it turns from set to unset, that file will be removed.
359 While $GIT_DIR/info/sparse-checkout is usually used to specify what
361 files are in. You can also specify what files are not in, using negate
362 patterns. For example, to remove file "unwanted":
363 *
365 !unwanted
366 Another tricky thing is fully repopulating working directory when you
369 no longer want sparse checkout. You cannot just disable "sparse
370 checkout" because skip-worktree are still in the index and you working
371 directory is still sparsely populated. You should re-populate working
372 directory with the $GIT_DIR/info/sparse-checkout file content as
373 follows:
374 *
376 Then you can disable sparse checkout. Sparse checkout support in "git
379 read-tree" and similar commands is disabled by default. You need to
380 turn core.sparseCheckout on in order to have sparse checkout support.
381
383 git-write-tree(1); git-ls-files(1); gitignore(5)
384
386 Written by Linus Torvalds <torvalds@osdl.org[1]>
387
389 Documentation by David Greaves, Junio C Hamano and the git-list
390 <git@vger.kernel.org[2]>.
391
393 Part of the git(1) suite
394
396 1. torvalds@osdl.org
397 mailto:torvalds@osdl.org
398 2. git@vger.kernel.org
400 mailto:git@vger.kernel.org
401Git 1.7.4.4 04/11/2011 GIT-READ-TREE(1)