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 [--trivial] [--aggressive] | --reset | --prefix=<prefix>] [-u | -i]] [--exclude-per-directory=<gitignore>] [--index-output=<file>] <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: 18linkgit: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-v:: 54 Show the progress of checking files out. 55 56--trivial:: 57 Restrict three-way merge by 'git-read-tree' to happen 58 only if there is no file-level merging required, instead 59 of resolving merge for trivial cases and leaving 60 conflicting files unresolved in the index. 61 62--aggressive:: 63 Usually a three-way merge by 'git-read-tree' resolves 64 the merge for really trivial cases and leaves other 65 cases unresolved in the index, so that Porcelains can 66 implement different merge policies. This flag makes the 67 command to resolve a few more cases internally: 68+ 69* when one side removes a path and the other side leaves the path 70 unmodified. The resolution is to remove that path. 71* when both sides remove a path. The resolution is to remove that path. 72* when both sides adds a path identically. The resolution 73 is to add that path. 74 75--prefix=<prefix>/:: 76 Keep the current index contents, and read the contents 77 of named tree-ish under directory at `<prefix>`. The 78 original index file cannot have anything at the path 79 `<prefix>` itself, and have nothing in `<prefix>/` 80 directory. Note that the `<prefix>/` value must end 81 with a slash. 82 83--exclude-per-directory=<gitignore>:: 84 When running the command with `-u` and `-m` options, the 85 merge result may need to overwrite paths that are not 86 tracked in the current branch. The command usually 87 refuses to proceed with the merge to avoid losing such a 88 path. However this safety valve sometimes gets in the 89 way. For example, it often happens that the other 90 branch added a file that used to be a generated file in 91 your branch, and the safety valve triggers when you try 92 to switch to that branch after you ran `make` but before 93 running `make clean` to remove the generated file. This 94 option tells the command to read per-directory exclude 95 file (usually '.gitignore') and allows such an untracked 96 but explicitly ignored file to be overwritten. 97 98--index-output=<file>:: 99 Instead of writing the results out to `$GIT_INDEX_FILE`, 100 write the resulting index in the named file. While the 101 command is operating, the original index file is locked 102 with the same mechanism as usual. The file must allow 103 to be rename(2)ed into from a temporary file that is 104 created next to the usual index file; typically this 105 means it needs to be on the same filesystem as the index 106 file itself, and you need write permission to the 107 directories the index file and index output file are 108 located in. 109 110<tree-ish#>:: 111 The id of the tree object(s) to be read/merged. 112 113 114Merging 115------- 116If `-m` is specified, 'git-read-tree' can perform 3 kinds of 117merge, a single tree merge if only 1 tree is given, a 118fast-forward merge with 2 trees, or a 3-way merge if 3 trees are 119provided. 120 121 122Single Tree Merge 123~~~~~~~~~~~~~~~~~ 124If only 1 tree is specified, 'git-read-tree' operates as if the user did not 125specify `-m`, except that if the original index has an entry for a 126given pathname, and the contents of the path matches with the tree 127being read, the stat info from the index is used. (In other words, the 128index's stat()s take precedence over the merged tree's). 129 130That means that if you do a `git read-tree -m <newtree>` followed by a 131`git checkout-index -f -u -a`, the 'git-checkout-index' only checks out 132the stuff that really changed. 133 134This is used to avoid unnecessary false hits when 'git-diff-files' is 135run after 'git-read-tree'. 136 137 138Two Tree Merge 139~~~~~~~~~~~~~~ 140 141Typically, this is invoked as `git read-tree -m $H $M`, where $H 142is the head commit of the current repository, and $M is the head 143of a foreign tree, which is simply ahead of $H (i.e. we are in a 144fast forward situation). 145 146When two trees are specified, the user is telling 'git-read-tree' 147the following: 148 149 1. The current index and work tree is derived from $H, but 150 the user may have local changes in them since $H; 151 152 2. The user wants to fast-forward to $M. 153 154In this case, the `git read-tree -m $H $M` command makes sure 155that no local change is lost as the result of this "merge". 156Here are the "carry forward" rules: 157 158 I (index) H M Result 159 ------------------------------------------------------- 160 0 nothing nothing nothing (does not happen) 161 1 nothing nothing exists use M 162 2 nothing exists nothing remove path from index 163 3 nothing exists exists, use M if "initial checkout" 164 H == M keep index otherwise 165 exists fail 166 H != M 167 168 clean I==H I==M 169 ------------------ 170 4 yes N/A N/A nothing nothing keep index 171 5 no N/A N/A nothing nothing keep index 172 173 6 yes N/A yes nothing exists keep index 174 7 no N/A yes nothing exists keep index 175 8 yes N/A no nothing exists fail 176 9 no N/A no nothing exists fail 177 178 10 yes yes N/A exists nothing remove path from index 179 11 no yes N/A exists nothing fail 180 12 yes no N/A exists nothing fail 181 13 no no N/A exists nothing fail 182 183 clean (H=M) 184 ------ 185 14 yes exists exists keep index 186 15 no exists exists keep index 187 188 clean I==H I==M (H!=M) 189 ------------------ 190 16 yes no no exists exists fail 191 17 no no no exists exists fail 192 18 yes no yes exists exists keep index 193 19 no no yes exists exists keep index 194 20 yes yes no exists exists use M 195 21 no yes no exists exists fail 196 197In all "keep index" cases, the index entry stays as in the 198original index file. If the entry were not up to date, 199'git-read-tree' keeps the copy in the work tree intact when 200operating under the -u flag. 201 202When this form of 'git-read-tree' returns successfully, you can 203see what "local changes" you made are carried forward by running 204`git diff-index --cached $M`. Note that this does not 205necessarily match `git diff-index --cached $H` would have 206produced before such a two tree merge. This is because of cases 20718 and 19 --- if you already had the changes in $M (e.g. maybe 208you picked it up via e-mail in a patch form), `git diff-index 209--cached $H` would have told you about the change before this 210merge, but it would not show in `git diff-index --cached $M` 211output after two-tree merge. 212 213Case #3 is slightly tricky and needs explanation. The result from this 214rule logically should be to remove the path if the user staged the removal 215of the path and then swiching to a new branch. That however will prevent 216the initial checkout from happening, so the rule is modified to use M (new 217tree) only when the contents of the index is empty. Otherwise the removal 218of the path is kept as long as $H and $M are the same. 219 2203-Way Merge 221~~~~~~~~~~~ 222Each "index" entry has two bits worth of "stage" state. stage 0 is the 223normal one, and is the only one you'd see in any kind of normal use. 224 225However, when you do 'git-read-tree' with three trees, the "stage" 226starts out at 1. 227 228This means that you can do 229 230---------------- 231$ git read-tree -m <tree1> <tree2> <tree3> 232---------------- 233 234and you will end up with an index with all of the <tree1> entries in 235"stage1", all of the <tree2> entries in "stage2" and all of the 236<tree3> entries in "stage3". When performing a merge of another 237branch into the current branch, we use the common ancestor tree 238as <tree1>, the current branch head as <tree2>, and the other 239branch head as <tree3>. 240 241Furthermore, 'git-read-tree' has special-case logic that says: if you see 242a file that matches in all respects in the following states, it 243"collapses" back to "stage0": 244 245 - stage 2 and 3 are the same; take one or the other (it makes no 246 difference - the same work has been done on our branch in 247 stage 2 and their branch in stage 3) 248 249 - stage 1 and stage 2 are the same and stage 3 is different; take 250 stage 3 (our branch in stage 2 did not do anything since the 251 ancestor in stage 1 while their branch in stage 3 worked on 252 it) 253 254 - stage 1 and stage 3 are the same and stage 2 is different take 255 stage 2 (we did something while they did nothing) 256 257The 'git-write-tree' command refuses to write a nonsensical tree, and it 258will complain about unmerged entries if it sees a single entry that is not 259stage 0. 260 261OK, this all sounds like a collection of totally nonsensical rules, 262but it's actually exactly what you want in order to do a fast 263merge. The different stages represent the "result tree" (stage 0, aka 264"merged"), the original tree (stage 1, aka "orig"), and the two trees 265you are trying to merge (stage 2 and 3 respectively). 266 267The order of stages 1, 2 and 3 (hence the order of three 268<tree-ish> command line arguments) are significant when you 269start a 3-way merge with an index file that is already 270populated. Here is an outline of how the algorithm works: 271 272- if a file exists in identical format in all three trees, it will 273 automatically collapse to "merged" state by 'git-read-tree'. 274 275- a file that has _any_ difference what-so-ever in the three trees 276 will stay as separate entries in the index. It's up to "porcelain 277 policy" to determine how to remove the non-0 stages, and insert a 278 merged version. 279 280- the index file saves and restores with all this information, so you 281 can merge things incrementally, but as long as it has entries in 282 stages 1/2/3 (i.e., "unmerged entries") you can't write the result. So 283 now the merge algorithm ends up being really simple: 284 285 * you walk the index in order, and ignore all entries of stage 0, 286 since they've already been done. 287 288 * if you find a "stage1", but no matching "stage2" or "stage3", you 289 know it's been removed from both trees (it only existed in the 290 original tree), and you remove that entry. 291 292 * if you find a matching "stage2" and "stage3" tree, you remove one 293 of them, and turn the other into a "stage0" entry. Remove any 294 matching "stage1" entry if it exists too. .. all the normal 295 trivial rules .. 296 297You would normally use 'git-merge-index' with supplied 298'git-merge-one-file' to do this last step. The script updates 299the files in the working tree as it merges each path and at the 300end of a successful merge. 301 302When you start a 3-way merge with an index file that is already 303populated, it is assumed that it represents the state of the 304files in your work tree, and you can even have files with 305changes unrecorded in the index file. It is further assumed 306that this state is "derived" from the stage 2 tree. The 3-way 307merge refuses to run if it finds an entry in the original index 308file that does not match stage 2. 309 310This is done to prevent you from losing your work-in-progress 311changes, and mixing your random changes in an unrelated merge 312commit. To illustrate, suppose you start from what has been 313committed last to your repository: 314 315---------------- 316$ JC=`git rev-parse --verify "HEAD^0"` 317$ git checkout-index -f -u -a $JC 318---------------- 319 320You do random edits, without running 'git-update-index'. And then 321you notice that the tip of your "upstream" tree has advanced 322since you pulled from him: 323 324---------------- 325$ git fetch git://.... linus 326$ LT=`cat .git/FETCH_HEAD` 327---------------- 328 329Your work tree is still based on your HEAD ($JC), but you have 330some edits since. Three-way merge makes sure that you have not 331added or modified index entries since $JC, and if you haven't, 332then does the right thing. So with the following sequence: 333 334---------------- 335$ git read-tree -m -u `git merge-base $JC $LT` $JC $LT 336$ git merge-index git-merge-one-file -a 337$ echo "Merge with Linus" | \ 338 git commit-tree `git write-tree` -p $JC -p $LT 339---------------- 340 341what you would commit is a pure merge between $JC and $LT without 342your work-in-progress changes, and your work tree would be 343updated to the result of the merge. 344 345However, if you have local changes in the working tree that 346would be overwritten by this merge, 'git-read-tree' will refuse 347to run to prevent your changes from being lost. 348 349In other words, there is no need to worry about what exists only 350in the working tree. When you have local changes in a part of 351the project that is not involved in the merge, your changes do 352not interfere with the merge, and are kept intact. When they 353*do* interfere, the merge does not even start ('git-read-tree' 354complains loudly and fails without modifying anything). In such 355a case, you can simply continue doing what you were in the 356middle of doing, and when your working tree is ready (i.e. you 357have finished your work-in-progress), attempt the merge again. 358 359 360SEE ALSO 361-------- 362linkgit:git-write-tree[1]; linkgit:git-ls-files[1]; 363linkgit:gitignore[5] 364 365 366Author 367------ 368Written by Linus Torvalds <torvalds@osdl.org> 369 370Documentation 371-------------- 372Documentation by David Greaves, Junio C Hamano and the git-list <git@vger.kernel.org>. 373 374GIT 375--- 376Part of the linkgit:git[1] suite