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 (<tree-ish> | [[-m [--trivial] [--aggressive] | --reset |
10 --prefix=<prefix>] [-u | -i]] [--exclude-per-directory=<gitignore>]
11 [--index-output=<file>] <tree-ish1> [<tree-ish2> [<tree-ish3>]])
12
14 Reads the tree information given by <tree-ish> into the index, but does
15 not actually update any of the files it "caches". (see: git-checkout-
16 index(1))
17 Optionally, it can merge a tree into the index, perform a fast-forward
19 (i.e. 2-way) merge, or a 3-way merge, with the -m flag. When used with
20 -m, the -u flag causes it to also update the files in the work tree
21 with the result of the merge.
22 Trivial merges are done by git-read-tree itself. Only conflicting paths
24 will be in unmerged state when git-read-tree returns.
25
27 -m
28 Perform a merge, not just a read. The command will refuse to run if
29 your index file has unmerged entries, indicating that you have not
30 finished previous merge you started.
31 --reset
33 Same as -m, except that unmerged entries are discarded instead of
34 failing.
35 -u
37 After a successful merge, update the files in the work tree with
38 the result of the merge.
39 -i
41 Usually a merge requires the index file as well as the files in the
42 working tree are up to date with the current head commit, in order
43 not to lose local changes. This flag disables the check with the
44 working tree and is meant to be used when creating a merge of trees
45 that are not directly related to the current working tree status
46 into a temporary index file.
47 --trivial
49 Restrict three-way merge by git-read-tree to happen only if there
50 is no file-level merging required, instead of resolving merge for
51 trivial cases and leaving conflicting files unresolved in the
52 index.
53 --aggressive
55 Usually a three-way merge by git-read-tree resolves the merge for
56 really trivial cases and leaves other cases unresolved in the
57 index, so that Porcelains can implement different merge policies.
58 This flag makes the command to resolve a few more cases internally:
59 · when one side removes a path and the other side leaves the path
62 unmodified. The resolution is to remove that path.
63 · when both sides remove a path. The resolution is to remove that
65 path.
66 · when both sides adds a path identically. The resolution is to
68 add that path.
69 --prefix=<prefix>/
71 Keep the current index contents, and read the contents of named
72 tree-ish under directory at <prefix>. The original index file
73 cannot have anything at the path <prefix> itself, and have nothing
74 in <prefix>/ directory. Note that the <prefix>/ value must end with
75 a slash.
76 --exclude-per-directory=<gitignore>
78 When running the command with -u and -m options, the merge result
79 may need to overwrite paths that are not tracked in the current
80 branch. The command usually refuses to proceed with the merge to
81 avoid losing such a path. However this safety valve sometimes gets
82 in the way. For example, it often happens that the other branch
83 added a file that used to be a generated file in your branch, and
84 the safety valve triggers when you try to switch to that branch
85 after you ran make but before running make clean to remove the
86 generated file. This option tells the command to read per-directory
87 exclude file (usually .gitignore) and allows such an untracked but
88 explicitly ignored file to be overwritten.
89 --index-output=<file>
91 Instead of writing the results out to $GIT_INDEX_FILE, write the
92 resulting index in the named file. While the command is operating,
93 the original index file is locked with the same mechanism as usual.
94 The file must allow to be rename(2)ed into from a temporary file
95 that is created next to the usual index file; typically this means
96 it needs to be on the same filesystem as the index file itself, and
97 you need write permission to the directories the index file and
98 index output file are located in.
99 <tree-ish#>
101 The id of the tree object(s) to be read/merged.
102
104 If -m is specified, git-read-tree can perform 3 kinds of merge, a
105 single tree merge if only 1 tree is given, a fast-forward merge with 2
106 trees, or a 3-way merge if 3 trees are provided.
107 Single Tree Merge
109 If only 1 tree is specified, git-read-tree operates as if the user did
110 not specify -m, except that if the original index has an entry for a
111 given pathname, and the contents of the path matches with the tree
112 being read, the stat info from the index is used. (In other words, the
113 index´s stat()s take precedence over the merged tree´s).
114 That means that if you do a git-read-tree -m <newtree> followed by a
116 git-checkout-index -f -u -a, the git-checkout-index only checks out the
117 stuff that really changed.
118 This is used to avoid unnecessary false hits when git-diff-files is run
120 after git-read-tree.
121 Two Tree Merge
123 Typically, this is invoked as git-read-tree -m $H $M, where $H is the
124 head commit of the current repository, and $M is the head of a foreign
125 tree, which is simply ahead of $H (i.e. we are in a fast forward
126 situation).
127 When two trees are specified, the user is telling git-read-tree the
129 following:
130 1. The current index and work tree is derived from $H, but the user
133 may have local changes in them since $H;
134 2. The user wants to fast-forward to $M.
136 In this case, the git-read-tree -m $H $M command makes sure that no
137 local change is lost as the result of this "merge". Here are the "carry
138 forward" rules:
139 I (index) H M Result
142 -------------------------------------------------------
143 0 nothing nothing nothing (does not happen)
144 1 nothing nothing exists use M
145 2 nothing exists nothing remove path from index
146 3 nothing exists exists use M
147 clean I==H I==M
149 ------------------
150 4 yes N/A N/A nothing nothing keep index
151 5 no N/A N/A nothing nothing keep index
152 6 yes N/A yes nothing exists keep index
154 7 no N/A yes nothing exists keep index
155 8 yes N/A no nothing exists fail
156 9 no N/A no nothing exists fail
157 10 yes yes N/A exists nothing remove path from index
159 11 no yes N/A exists nothing fail
160 12 yes no N/A exists nothing fail
161 13 no no N/A exists nothing fail
162 clean (H=M)
164 ------
165 14 yes exists exists keep index
166 15 no exists exists keep index
167 clean I==H I==M (H!=M)
169 ------------------
170 16 yes no no exists exists fail
171 17 no no no exists exists fail
172 18 yes no yes exists exists keep index
173 19 no no yes exists exists keep index
174 20 yes yes no exists exists use M
175 21 no yes no exists exists fail
176 In all "keep index" cases, the index entry stays as in the original
177 index file. If the entry were not up to date, git-read-tree keeps the
178 copy in the work tree intact when operating under the -u flag.
179 When this form of git-read-tree returns successfully, you can see what
181 "local changes" you made are carried forward by running git-diff-index
182 --cached $M. Note that this does not necessarily match git-diff-index
183 --cached $H would have produced before such a two tree merge. This is
184 because of cases 18 and 19 --- if you already had the changes in $M
185 (e.g. maybe you picked it up via e-mail in a patch form),
186 git-diff-index --cached $H would have told you about the change before
187 this merge, but it would not show in git-diff-index --cached $M output
188 after two-tree merge.
189 3-Way Merge
191 Each "index" entry has two bits worth of "stage" state. stage 0 is the
192 normal one, and is the only one you´d see in any kind of normal use.
193 However, when you do git-read-tree with three trees, the "stage" starts
195 out at 1.
196 This means that you can do
198 $ git-read-tree -m <tree1> <tree2> <tree3>
202 and you will end up with an index with all of the <tree1> entries in
204 "stage1", all of the <tree2> entries in "stage2" and all of the <tree3>
205 entries in "stage3". When performing a merge of another branch into the
206 current branch, we use the common ancestor tree as <tree1>, the current
207 branch head as <tree2>, and the other branch head as <tree3>.
208 Furthermore, git-read-tree has special-case logic that says: if you see
210 a file that matches in all respects in the following states, it
211 "collapses" back to "stage0":
212 · stage 2 and 3 are the same; take one or the other (it makes no
215 difference - the same work has been done on our branch in stage 2
216 and their branch in stage 3)
217 · stage 1 and stage 2 are the same and stage 3 is different; take
219 stage 3 (our branch in stage 2 did not do anything since the
220 ancestor in stage 1 while their branch in stage 3 worked on it)
221 · stage 1 and stage 3 are the same and stage 2 is different take
223 stage 2 (we did something while they did nothing)
224 The git-write-tree command refuses to write a nonsensical tree, and it
225 will complain about unmerged entries if it sees a single entry that is
226 not stage 0.
227 OK, this all sounds like a collection of totally nonsensical rules, but
229 it´s actually exactly what you want in order to do a fast merge. The
230 different stages represent the "result tree" (stage 0, aka "merged"),
231 the original tree (stage 1, aka "orig"), and the two trees you are
232 trying to merge (stage 2 and 3 respectively).
233 The order of stages 1, 2 and 3 (hence the order of three <tree-ish>
235 command line arguments) are significant when you start a 3-way merge
236 with an index file that is already populated. Here is an outline of how
237 the algorithm works:
238 · if a file exists in identical format in all three trees, it will
241 automatically collapse to "merged" state by git-read-tree.
242 · a file that has any difference what-so-ever in the three trees will
244 stay as separate entries in the index. It´s up to "porcelain
245 policy" to determine how to remove the non-0 stages, and insert a
246 merged version.
247 · the index file saves and restores with all this information, so you
249 can merge things incrementally, but as long as it has entries in
250 stages 1/2/3 (i.e., "unmerged entries") you can´t write the result.
251 So now the merge algorithm ends up being really simple:
252 · you walk the index in order, and ignore all entries of stage 0,
255 since they´ve already been done.
256 · if you find a "stage1", but no matching "stage2" or "stage3",
258 you know it´s been removed from both trees (it only existed in
259 the original tree), and you remove that entry.
260 · if you find a matching "stage2" and "stage3" tree, you remove
262 one of them, and turn the other into a "stage0" entry. Remove
263 any matching "stage1" entry if it exists too. .. all the normal
264 trivial rules ..
265 You would normally use git-merge-index with supplied git-merge-one-file
266 to do this last step. The script updates the files in the working tree
267 as it merges each path and at the end of a successful merge.
268 When you start a 3-way merge with an index file that is already
270 populated, it is assumed that it represents the state of the files in
271 your work tree, and you can even have files with changes unrecorded in
272 the index file. It is further assumed that this state is "derived" from
273 the stage 2 tree. The 3-way merge refuses to run if it finds an entry
274 in the original index file that does not match stage 2.
275 This is done to prevent you from losing your work-in-progress changes,
277 and mixing your random changes in an unrelated merge commit. To
278 illustrate, suppose you start from what has been committed last to your
279 repository:
280 $ JC=`git-rev-parse --verify "HEAD^0"`
284 $ git-checkout-index -f -u -a $JC
285 You do random edits, without running git-update-index. And then you
287 notice that the tip of your "upstream" tree has advanced since you
288 pulled from him:
289 $ git-fetch git://.... linus
293 $ LT=`cat .git/FETCH_HEAD`
294 Your work tree is still based on your HEAD ($JC), but you have some
296 edits since. Three-way merge makes sure that you have not added or
297 modified index entries since $JC, and if you haven´t, then does the
298 right thing. So with the following sequence:
299 $ git-read-tree -m -u `git-merge-base $JC $LT` $JC $LT
303 $ git-merge-index git-merge-one-file -a
304 $ echo "Merge with Linus" | \
305 git-commit-tree `git-write-tree` -p $JC -p $LT
306 what you would commit is a pure merge between $JC and $LT without your
308 work-in-progress changes, and your work tree would be updated to the
309 result of the merge.
310 However, if you have local changes in the working tree that would be
312 overwritten by this merge,git-read-tree will refuse to run to prevent
313 your changes from being lost.
314 In other words, there is no need to worry about what exists only in the
316 working tree. When you have local changes in a part of the project that
317 is not involved in the merge, your changes do not interfere with the
318 merge, and are kept intact. When they do interfere, the merge does not
319 even start (git-read-tree complains loudly and fails without modifying
320 anything). In such a case, you can simply continue doing what you were
321 in the middle of doing, and when your working tree is ready (i.e. you
322 have finished your work-in-progress), attempt the merge again.
323
325 git-write-tree(1); git-ls-files(1); gitignore(5)
326
328 Written by Linus Torvalds <torvalds@osdl.org>
329
331 Documentation by David Greaves, Junio C Hamano and the git-list
332 <git@vger.kernel.org>.
333
335 Part of the git(7) suite
336Git 1.5.3.3 10/09/2007 GIT-READ-TREE(1)