1Git 2.13 Release Notes 2====================== 3 4Backward compatibility notes. 5 6 * Use of an empty string as a pathspec element that is used for 7 'everything matches' is still warned and Git asks users to use a 8 more explicit '.' for that instead. The hope is that existing 9 users will not mind this change, and eventually the warning can be 10 turned into a hard error, upgrading the deprecation into removal of 11 this (mis)feature. That is not scheduled to happen in the upcoming 12 release (yet). 13 14 * The historical argument order "git merge <msg> HEAD <commit>..." 15 has been deprecated for quite some time, and is now removed. 16 17 * The default location "~/.git-credential-cache/socket" for the 18 socket used to communicate with the credential-cache daemon has 19 been moved to "~/.cache/git/credential/socket". 20 21 * Git now avoids blindly falling back to ".git" when the setup 22 sequence said we are _not_ in Git repository. A corner case that 23 happens to work right now may be broken by a call to die("BUG"). 24 We've tried hard to locate such cases and fixed them, but there 25 might still be cases that need to be addressed--bug reports are 26 greatly appreciated. 27 28 29Updates since v2.12 30------------------- 31 32UI, Workflows & Features 33 34 * "git describe" and "git name-rev" have been taught to take more 35 than one refname patterns to restrict the set of refs to base their 36 naming output on, and also learned to take negative patterns to 37 name refs not to be used for naming via their "--exclude" option. 38 39 * Deletion of a branch "foo/bar" could remove .git/refs/heads/foo 40 once there no longer is any other branch whose name begins with 41 "foo/", but we didn't do so so far. Now we do. 42 43 * When "git merge" detects a path that is renamed in one history 44 while the other history deleted (or modified) it, it now reports 45 both paths to help the user understand what is going on in the two 46 histories being merged. 47 48 * The <url> part in "http.<url>.<variable>" configuration variable 49 can now be spelled with '*' that serves as wildcard. 50 E.g. "http.https://*.example.com.proxy" can be used to specify the 51 proxy used for https://a.example.com, https://b.example.com, etc., 52 i.e. any host in the example.com domain. 53 54 * "git tag" did not leave useful message when adding a new entry to 55 reflog; this was left unnoticed for a long time because refs/tags/* 56 doesn't keep reflog by default. 57 58 * The "negative" pathspec feature was somewhat more cumbersome to use 59 than necessary in that its short-hand used "!" which needed to be 60 escaped from shells, and it required "exclude from what?" specified. 61 62 * The command line options for ssh invocation needs to be tweaked for 63 some implementations of SSH (e.g. PuTTY plink wants "-P <port>" 64 while OpenSSH wants "-p <port>" to specify port to connect to), and 65 the variant was guessed when GIT_SSH environment variable is used 66 to specify it. The logic to guess now applies to the command 67 specified by the newer GIT_SSH_COMMAND and also core.sshcommand 68 configuration variable, and comes with an escape hatch for users to 69 deal with misdetected cases. 70 71 * The "--git-path", "--git-common-dir", and "--shared-index-path" 72 options of "git rev-parse" did not produce usable output. They are 73 now updated to show the path to the correct file, relative to where 74 the caller is. 75 76 * "git diff -W" has been taught to handle the case where a new 77 function is added at the end of the file better. 78 79 * "git update-ref -d" and other operations to delete references did 80 not leave any entry in HEAD's reflog when the reference being 81 deleted was the current branch. This is not a problem in practice 82 because you do not want to delete the branch you are currently on, 83 but caused renaming of the current branch to something else not to 84 be logged in a useful way. 85 86 * "Cc:" on the trailer part does not have to conform to RFC strictly, 87 unlike in the e-mail header. "git send-email" has been updated to 88 ignore anything after '>' when picking addresses, to allow non-address 89 cruft like " # stable 4.4" after the address. 90 91 * When "git submodule init" decides that the submodule in the working 92 tree is its upstream, it now gives a warning as it is not a very 93 common setup. 94 95 * "git stash push" takes a pathspec so that the local changes can be 96 stashed away only partially. 97 98 * Documentation for "git ls-files" did not refer to core.quotePath. 99 100 * The experimental "split index" feature has gained a few 101 configuration variables to make it easier to use. 102 103 * From a working tree of a repository, a new option of "rev-parse" 104 lets you ask if the repository is used as a submodule of another 105 project, and where the root level of the working tree of that 106 project (i.e. your superproject) is. 107 108 * The pathspec mechanism learned to further limit the paths that 109 match the pattern to those that have specified attributes attached 110 via the gitattributes mechanism. 111 112 * Our source code has used the SHA1_HEADER cpp macro after "#include" 113 in the C code to switch among the SHA-1 implementations. Instead, 114 list the exact header file names and switch among implementations 115 using "#ifdef BLK_SHA1/#include "block-sha1/sha1.h"/.../#endif"; 116 this helps some IDE tools. 117 118 * The start-up sequence of "git" needs to figure out some configured 119 settings before it finds and set itself up in the location of the 120 repository and was quite messy due to its "chicken-and-egg" nature. 121 The code has been restructured. 122 123 * The command line prompt (in contrib/) learned a new 'tag' style 124 that can be specified with GIT_PS1_DESCRIBE_STYLE, to describe a 125 detached HEAD with "git describe --tags". 126 127 * The configuration file learned a new "includeIf.<condition>.path" 128 that includes the contents of the given path only when the 129 condition holds. This allows you to say "include this work-related 130 bit only in the repositories under my ~/work/ directory". 131 132 * Recent update to "rebase -i" started showing a message that is not 133 a warning with "warning:" prefix by mistake. This has been fixed. 134 135 * Recently we started passing the "--push-options" through the 136 external remote helper interface; now the "smart HTTP" remote 137 helper understands what to do with the passed information. 138 139 * "git describe --dirty" dies when it cannot be determined if the 140 state in the working tree matches that of HEAD (e.g. broken 141 repository or broken submodule). The command learned a new option 142 "git describe --broken" to give "$name-broken" (where $name is the 143 description of HEAD) in such a case. 144 145 * "git checkout" is taught the "--recurse-submodules" option. 146 147 * Recent enhancement to "git stash push" command to support pathspec 148 to allow only a subset of working tree changes to be stashed away 149 was found to be too chatty and exposed the internal implementation 150 detail (e.g. when it uses reset to match the index to HEAD before 151 doing other things, output from reset seeped out). These, and 152 other chattyness has been fixed. 153 154 * "git merge <message> HEAD <commit>" syntax that has been deprecated 155 since October 2007 has been removed. 156 157 * The refs completion for large number of refs has been sped up, 158 partly by giving up disambiguating ambiguous refs and partly by 159 eliminating most of the shell processing between 'git for-each-ref' 160 and 'ls-remote' and Bash's completion facility. 161 162 * On many keyboards, typing "@{" involves holding down SHIFT key and 163 one can easily end up with "@{Up..." when typing "@{upstream}". As 164 the upstream/push keywords do not appear anywhere else in the syntax, 165 we can safely accept them case insensitively without introducing 166 ambiguity or confusion to solve this. 167 168 * "git tag/branch/for-each-ref" family of commands long allowed to 169 filter the refs by "--contains X" (show only the refs that are 170 descendants of X), "--merged X" (show only the refs that are 171 ancestors of X), "--no-merged X" (show only the refs that are not 172 ancestors of X). One curious omission, "--no-contains X" (show 173 only the refs that are not descendants of X) has been added to 174 them. 175 176 * The default behaviour of "git log" in an interactive session has 177 been changed to enable "--decorate". 178 179 180Performance, Internal Implementation, Development Support etc. 181 182 * The code to list branches in "git branch" has been consolidated 183 with the more generic ref-filter API. 184 185 * Resource usage while enumerating refs from alternate object store 186 has been optimized to help receiving end of "push" that hosts a 187 repository with many "forks". 188 189 * The gitattributes machinery is being taught to work better in a 190 multi-threaded environment. 191 192 * "git rebase -i" starts using the recently updated "sequencer" code. 193 194 * Code and design clean-up for the refs API. 195 196 * The preload-index code has been taught not to bother with the index 197 entries that are paths that are not checked out by "sparse checkout". 198 199 * Some warning() messages from "git clean" were updated to show the 200 errno from failed system calls. 201 202 * The "parse_config_key()" API function has been cleaned up. 203 204 * A test that creates a confusing branch whose name is HEAD has been 205 corrected not to do so. 206 207 * The code that parses header fields in the commit object has been 208 updated for (micro)performance and code hygiene. 209 210 * An helper function to make it easier to append the result from 211 real_path() to a strbuf has been added. 212 213 * Reduce authentication round-trip over HTTP when the server supports 214 just a single authentication method. This also improves the 215 behaviour when Git is misconfigured to enable http.emptyAuth 216 against a server that does not authenticate without a username 217 (i.e. not using Kerberos etc., which makes http.emptyAuth 218 pointless). 219 220 * Windows port wants to use OpenSSL's implementation of SHA-1 221 routines, so let them. 222 223 * The t/perf performance test suite was not prepared to test not so 224 old versions of Git, but now it covers versions of Git that are not 225 so ancient. 226 227 * Add 32-bit Linux variant to the set of platforms to be tested with 228 Travis CI. 229 230 * "git branch --list" takes the "--abbrev" and "--no-abbrev" options 231 to control the output of the object name in its "-v"(erbose) 232 output, but a recent update started ignoring them; fix it before 233 the breakage reaches to any released version. 234 235 * Picking two versions of Git and running tests to make sure the 236 older one and the newer one interoperate happily has now become 237 possible. 238 239 * "uchar [40]" to "struct object_id" conversion continues. 240 241 * "git tag --contains" used to (ab)use the object bits to keep track 242 of the state of object reachability without clearing them after 243 use; this has been cleaned up and made to use the newer commit-slab 244 facility. 245 246 * The "debug" helper used in the test framework learned to run 247 a command under "gdb" interactively. 248 249 * The "detect attempt to create collisions" variant of SHA-1 250 implementation by Marc Stevens (CWI) and Dan Shumow (Microsoft) 251 has been integrated and made the default. 252 253 * The test framework learned to detect unterminated here documents. 254 255 * The name-hash used for detecting paths that are different only in 256 cases (which matter on case insensitive filesystems) has been 257 optimized to take advantage of multi-threading when it makes sense. 258 259 * An earlier version of sha1dc/sha1.c that was merged to 'master' 260 compiled incorrectly on Windows, which has been fixed. 261 262 * "what URL do we want to update this submodule?" and "are we 263 interested in this submodule?" are split into two distinct 264 concepts, and then the way used to express the latter got extended, 265 paving a way to make it easier to manage a project with many 266 submodules and make it possible to later extend use of multiple 267 worktrees for a project with submodules. 268 269 * Some debugging output from "git describe" were marked for l10n, 270 but some weren't. Mark missing ones for l10n. 271 272 * Define a new task in .travis.yml that triggers a test session on 273 Windows run elsewhere. 274 275Also contains various documentation updates and code clean-ups. 276 277 278Fixes since v2.12 279----------------- 280 281Unless otherwise noted, all the fixes since v2.12 in the maintenance 282track are contained in this release (see the maintenance releases' 283notes for details). 284 285 * "git repack --depth=<n>" for a long time busted the specified depth 286 when reusing delta from existing packs. This has been corrected. 287 288 * The code to parse the command line "git grep <patterns>... <rev> 289 [[--] <pathspec>...]" has been cleaned up, and a handful of bugs 290 have been fixed (e.g. we used to check "--" if it is a rev). 291 292 * "git ls-remote" and "git archive --remote" are designed to work 293 without being in a directory under Git's control. However, recent 294 updates revealed that we randomly look into a directory called 295 .git/ without actually doing necessary set-up when working in a 296 repository. Stop doing so. 297 298 * "git show-branch" expected there were only very short branch names 299 in the repository and used a fixed-length buffer to hold them 300 without checking for overflow. 301 302 * A caller of tempfile API that uses stdio interface to write to 303 files may ignore errors while writing, which is detected when 304 tempfile is closed (with a call to ferror()). By that time, the 305 original errno that may have told us what went wrong is likely to 306 be long gone and was overwritten by an irrelevant value. 307 close_tempfile() now resets errno to EIO to make errno at least 308 predictable. 309 310 * "git remote rm X", when a branch has remote X configured as the 311 value of its branch.*.remote, tried to remove branch.*.remote and 312 branch.*.merge and failed if either is unset. 313 314 * A "gc.log" file left by a backgrounded "gc --auto" disables further 315 automatic gc; it has been taught to run at least once a day (by 316 default) by ignoring a stale "gc.log" file that is too old. 317 318 * The code to parse "git -c VAR=VAL cmd" and set configuration 319 variable for the duration of cmd had two small bugs, which have 320 been fixed. 321 322 * user.email that consists of only cruft chars should consistently 323 error out, but didn't. 324 325 * "git upload-pack", which is a counter-part of "git fetch", did not 326 report a request for a ref that was not advertised as invalid. 327 This is generally not a problem (because "git fetch" will stop 328 before making such a request), but is the right thing to do. 329 330 * A leak in a codepath to read from a packed object in (rare) cases 331 has been plugged. 332 333 * When a redirected http transport gets an error during the 334 redirected request, we ignored the error we got from the server, 335 and ended up giving a not-so-useful error message. 336 337 * The patch subcommand of "git add -i" was meant to have paths 338 selection prompt just like other subcommand, unlike "git add -p" 339 directly jumps to hunk selection. Recently, this was broken and 340 "add -i" lost the paths selection dialog, but it now has been 341 fixed. 342 343 * Git v2.12 was shipped with an embarrassing breakage where various 344 operations that verify paths given from the user stopped dying when 345 seeing an issue, and instead later triggering segfault. 346 347 * There is no need for Python only to give a few messages to the 348 standard error stream, but we somehow did. 349 350 * The code to parse "git log -L..." command line was buggy when there 351 are many ranges specified with -L; overrun of the allocated buffer 352 has been fixed. 353 354 * The command-line parsing of "git log -L" copied internal data 355 structures using incorrect size on ILP32 systems. 356 357 * "git diff --quiet" relies on the size field in diff_filespec to be 358 correctly populated, but diff_populate_filespec() helper function 359 made an incorrect short-cut when asked only to populate the size 360 field for paths that need to go through convert_to_git() (e.g. CRLF 361 conversion). 362 363 * A few tests were run conditionally under (rare) conditions where 364 they cannot be run (like running cvs tests under 'root' account). 365 366 * "git branch @" created refs/heads/@ as a branch, and in general the 367 code that handled @{-1} and @{upstream} was a bit too loose in 368 disambiguating. 369 370 * "git fetch" that requests a commit by object name, when the other 371 side does not allow such an request, failed without much 372 explanation. 373 374 * "git filter-branch --prune-empty" drops a single-parent commit that 375 becomes a no-op, but did not drop a root commit whose tree is empty. 376 377 * Recent versions of Git treats http alternates (used in dumb http 378 transport) just like HTTP redirects and requires the client to 379 enable following it, due to security concerns. But we forgot to 380 give a warning when we decide not to honor the alternates. 381 382 * "git push" had a handful of codepaths that could lead to a deadlock 383 when unexpected error happened, which has been fixed. 384 385 * "Dumb http" transport used to misparse a nonsense http-alternates 386 response, which has been fixed. 387 388 * "git add -p <pathspec>" unnecessarily expanded the pathspec to a 389 list of individual files that matches the pathspec by running "git 390 ls-files <pathspec>", before feeding it to "git diff-index" to see 391 which paths have changes, because historically the pathspec 392 language supported by "diff-index" was weaker. These days they are 393 equivalent and there is no reason to internally expand it. This 394 helps both performance and avoids command line argument limit on 395 some platforms. 396 (merge 7288e12cce jk/add-i-use-pathspecs later to maint). 397 398 * "git status --porcelain" is supposed to give a stable output, but a 399 few strings were left as translatable by mistake. 400 401 * "git revert -m 0 $merge_commit" complained that reverting a merge 402 needs to say relative to which parent the reversion needs to 403 happen, as if "-m 0" weren't given. The correct diagnosis is that 404 "-m 0" does not refer to the first parent ("-m 1" does). This has 405 been fixed. 406 407 * Code to read submodule.<name>.ignore config did not state the 408 variable name correctly when giving an error message diagnosing 409 misconfiguration. 410 411 * Fix for NO_PTHREADS build. 412 413 * Fix for potential segv introduced in v2.11.0 and later (also 414 v2.10.2) to "git log --pickaxe-regex -S". 415 416 * A few unterminated here documents in tests were fixed, which in 417 turn revealed incorrect expectations the tests make. These tests 418 have been updated. 419 420 * Fix for NO_PTHREADS option. 421 (merge 2225e1ea20 bw/grep-recurse-submodules later to maint). 422 423 * Git now avoids blindly falling back to ".git" when the setup 424 sequence said we are _not_ in Git repository. A corner case that 425 happens to work right now may be broken by a call to die("BUG"). 426 (merge b1ef400eec jk/no-looking-at-dotgit-outside-repo-final later to maint). 427 428 * A few commands that recently learned the "--recurse-submodule" 429 option misbehaved when started from a subdirectory of the 430 superproject. 431 (merge b2dfeb7c00 bw/recurse-submodules-relative-fix later to maint). 432 433 * FreeBSD implementation of getcwd(3) behaved differently when an 434 intermediate directory is unreadable/unsearchable depending on the 435 length of the buffer provided, which our strbuf_getcwd() was not 436 aware of. strbuf_getcwd() has been taught to cope with it better. 437 (merge a54e938e5b rs/freebsd-getcwd-workaround later to maint). 438 439 * A recent update to "rebase -i" stopped running hooks for the "git 440 commit" command during "reword" action, which has been fixed. 441 442 * Removing an entry from a notes tree and then looking another note 443 entry from the resulting tree using the internal notes API 444 functions did not work as expected. No in-tree users of the API 445 has such access pattern, but it still is worth fixing. 446 447 * "git receive-pack" could have been forced to die by attempting 448 allocate an unreasonably large amount of memory with a crafted push 449 certificate; this has been fixed. 450 (merge f2214dede9 bc/push-cert-receive-fix later to maint). 451 452 * Other minor doc, test and build updates and code cleanups. 453 (merge df2a6e38b7 jk/pager-in-use later to maint). 454 (merge 75ec4a6cb0 ab/branch-list-doc later to maint). 455 (merge 3e5b36c637 sg/skip-prefix-in-prettify-refname later to maint). 456 (merge 2c5e2865cc jk/fast-import-cleanup later to maint). 457 (merge 4473060bc2 ab/test-readme-updates later to maint). 458 (merge 48a96972fd ab/doc-submitting later to maint). 459 (merge f5c2bc2b96 jk/make-coccicheck-detect-errors later to maint). 460 (merge c105f563d1 cc/untracked later to maint).