1Git 2.11 Release Notes 2====================== 3 4Backward compatibility notes. 5 6 * An empty string used as a pathspec element has always meant 7 'everything matches', but it is too easy to write a script that 8 finds a path to remove in $path and run 'git rm "$paht"', which 9 ends up removing everything. This release starts warning about the 10 use of an empty string that is used for 'everything matches' and 11 asks users to use a more explicit '.' for that instead. 12 13 The hope is that existing users will not mind this change, and 14 eventually the warning can be turned into a hard error, upgrading 15 the deprecation into removal of this (mis)feature. 16 17 18 * The historical argument order "git merge <msg> HEAD <commit>..." 19 has been deprecated for quite some time, and will be removed in the 20 next release (not this one). 21 22 23Updates since v2.10 24------------------- 25 26UI, Workflows & Features 27 28 * Comes with new version of git-gui, now at its 0.21.0 tag. 29 30 * "git format-patch --cover-letter HEAD^" to format a single patch 31 with a separate cover letter now numbers the output as [PATCH 0/1] 32 and [PATCH 1/1] by default. 33 34 * An incoming "git push" that attempts to push too many bytes can now 35 be rejected by setting a new configuration variable at the receiving 36 end. 37 38 * "git nosuchcommand --help" said "No manual entry for gitnosuchcommand", 39 which was not intuitive, given that "git nosuchcommand" said "git: 40 'nosuchcommand' is not a git command". 41 42 * "git clone --resurse-submodules --reference $path $URL" is a way to 43 reduce network transfer cost by borrowing objects in an existing 44 $path repository when cloning the superproject from $URL; it 45 learned to also peek into $path for presense of corresponding 46 repositories of submodules and borrow objects from there when able. 47 48 * The "git diff --submodule={short,log}" mechanism has been enhanced 49 to allow "--submodule=diff" to show the patch between the submodule 50 commits bound to the superproject. 51 52 * Even though "git hash-objects", which is a tool to take an 53 on-filesystem data stream and put it into the Git object store, 54 allowed to perform the "outside-world-to-Git" conversions (e.g. 55 end-of-line conversions and application of the clean-filter), and 56 it had the feature on by default from very early days, its reverse 57 operation "git cat-file", which takes an object from the Git object 58 store and externalize for the consumption by the outside world, 59 lacked an equivalent mechanism to run the "Git-to-outside-world" 60 conversion. The command learned the "--filters" option to do so. 61 62 * Output from "git diff" can be made easier to read by selecting 63 which lines are common and which lines are added/deleted 64 intelligently when the lines before and after the changed section 65 are the same. A command line option is added to help with the 66 experiment to find a good heuristics. 67 68 * In some projects, it is common to use "[RFC PATCH]" as the subject 69 prefix for a patch meant for discussion rather than application. A 70 new option "--rfc" was a short-hand for "--subject-prefix=RFC PATCH" 71 to help the participants of such projects. 72 73 * "git add --chmod=+x <pathspec>" added recently only toggled the 74 executable bit for paths that are either new or modified. This has 75 been corrected to flip the executable bit for all paths that match 76 the given pathspec. 77 78 * When "git format-patch --stdout" output is placed as an in-body 79 header and it uses the RFC2822 header folding, "git am" failed to 80 put the header line back into a single logical line. The 81 underlying "git mailinfo" was taught to handle this properly. 82 83 * "gitweb" can spawn "highlight" to show blob contents with 84 (programming) language-specific syntax highlighting, but only 85 when the language is known. "highlight" can however be told 86 to make the guess itself by giving it "--force" option, which 87 has been enabled. 88 89 * "git gui" l10n to Portuguese. 90 91 * When given an abbreviated object name that is not (or more 92 realistically, "no longer") unique, we gave a fatal error 93 "ambiguous argument". This error is now accompanied by hints that 94 lists the objects that begins with the given prefix. During the 95 course of development of this new feature, numerous minor bugs were 96 uncovered and corrected, the most notable one of which is that we 97 gave "short SHA1 xxxx is ambiguous." twice without good reason. 98 99 * "git log rev^..rev" is an often-used revision range specification 100 to show what was done on a side branch merged at rev. This has 101 gained a short-hand "rev^-1". In general "rev^-$n" is the same as 102 "^rev^$n rev", i.e. what has happened on other branches while the 103 history leading to nth parent was looking the other way. 104 105 * In recent versions of cURL, GSSAPI credential delegation is 106 disabled by default due to CVE-2011-2192; introduce a configuration 107 to selectively allow enabling this. 108 (merge 26a7b23429 ps/http-gssapi-cred-delegation later to maint). 109 110 * "git mergetool" learned to honor "-O<orderfile>" to control the 111 order of paths to present to the end user. 112 113 * "git diff/log --ws-error-highlight=<kind>" lacked the corresponding 114 configuration variable to set it by default. 115 116 * "git ls-files" learned "--recurse-submodules" option that can be 117 used to get a listing of tracked files across submodules (i.e. this 118 only works with "--cached" option, not for listing untracked or 119 ignored files). This would be a useful tool to sit on the upstream 120 side of a pipe that is read with xargs to work on all working tree 121 files from the top-level superproject. 122 123 * A new credential helper that talks via "libsecret" with 124 implementations of XDG Secret Service API has been added to 125 contrib/credential/. 126 127 * The GPG verification status shown in "%G?" pretty format specifier 128 was not rich enough to differentiate a signature made by an expired 129 key, a signature made by a revoked key, etc. New output letters 130 have been assigned to express them. 131 132 * In addition to purely abbreviated commit object names, "gitweb" 133 learned to turn "git describe" output (e.g. v2.9.3-599-g2376d31787) 134 into clickable links in its output. 135 136 137Performance, Internal Implementation, Development Support etc. 138 139 * The delta-base-cache mechanism has been a key to the performance in 140 a repository with a tightly packed packfile, but it did not scale 141 well even with a larger value of core.deltaBaseCacheLimit. 142 143 * Enhance "git status --porcelain" output by collecting more data on 144 the state of the index and the working tree files, which may 145 further be used to teach git-prompt (in contrib/) to make fewer 146 calls to git. 147 148 * Extract a small helper out of the function that reads the authors 149 script file "git am" internally uses. 150 (merge a77598e jc/am-read-author-file later to maint). 151 152 * Lifts calls to exit(2) and die() higher in the callchain in 153 sequencer.c files so that more helper functions in it can be used 154 by callers that want to handle error conditions themselves. 155 156 * "git am" has been taught to make an internal call to "git apply"'s 157 innards without spawning the latter as a separate process. 158 159 * The ref-store abstraction was introduced to the refs API so that we 160 can plug in different backends to store references. 161 162 * The "unsigned char sha1[20]" to "struct object_id" conversion 163 continues. Notable changes in this round includes that ce->sha1, 164 i.e. the object name recorded in the cache_entry, turns into an 165 object_id. 166 167 * JGit can show a fake ref "capabilities^{}" to "git fetch" when it 168 does not advertise any refs, but "git fetch" was not prepared to 169 see such an advertisement. When the other side disconnects without 170 giving any ref advertisement, we used to say "there may not be a 171 repository at that URL", but we may have seen other advertisement 172 like "shallow" and ".have" in which case we definitely know that a 173 repository is there. The code to detect this case has also been 174 updated. 175 176 * Some codepaths in "git pack-objects" were not ready to use an 177 existing pack bitmap; now they are and as the result they have 178 become faster. 179 180 * The codepath in "git fsck" to detect malformed tree objects has 181 been updated not to die but keep going after detecting them. 182 183 * We call "qsort(array, nelem, sizeof(array[0]), fn)", and most of 184 the time third parameter is redundant. A new QSORT() macro lets us 185 omit it. 186 187 * "git pack-objects" in a repository with many packfiles used to 188 spend a lot of time looking for/at objects in them; the accesses to 189 the packfiles are now optimized by checking the most-recently-used 190 packfile first. 191 (merge c9af708b1a jk/pack-objects-optim-mru later to maint). 192 193 * Codepaths involved in interacting alternate object store have 194 been cleaned up. 195 196 * In order for the receiving end of "git push" to inspect the 197 received history and decide to reject the push, the objects sent 198 from the sending end need to be made available to the hook and 199 the mechanism for the connectivity check, and this was done 200 traditionally by storing the objects in the receiving repository 201 and letting "git gc" to expire it. Instead, store the newly 202 received objects in a temporary area, and make them available by 203 reusing the alternate object store mechanism to them only while we 204 decide if we accept the check, and once we decide, either migrate 205 them to the repository or purge them immediately. 206 207 * The require_clean_work_tree() helper was recreated in C when "git 208 pull" was rewritten from shell; the helper is now made available to 209 other callers in preparation for upcoming "rebase -i" work. 210 211 * "git upload-pack" had its code cleaned-up and performance improved 212 by reducing use of timestamp-ordered commit-list, which was 213 replaced with a priority queue. 214 215 216Also contains various documentation updates and code clean-ups. 217 218 219Fixes since v2.10 220----------------- 221 222Unless otherwise noted, all the fixes since v2.9 in the maintenance 223track are contained in this release (see the maintenance releases' 224notes for details). 225 226 * Clarify various ways to specify the "revision ranges" in the 227 documentation. 228 229 * "diff-highlight" script (in contrib/) learned to work better with 230 "git log -p --graph" output. 231 232 * The test framework left the number of tests and success/failure 233 count in the t/test-results directory, keyed by the name of the 234 test script plus the process ID. The latter however turned out not 235 to serve any useful purpose. The process ID part of the filename 236 has been removed. 237 238 * Having a submodule whose ".git" repository is somehow corrupt 239 caused a few commands that recurse into submodules loop forever. 240 241 * "git symbolic-ref -d HEAD" happily removes the symbolic ref, but 242 the resulting repository becomes an invalid one. Teach the command 243 to forbid removal of HEAD. 244 245 * A test spawned a short-lived background process, which sometimes 246 prevented the test directory from getting removed at the end of the 247 script on some platforms. 248 249 * Update a few tests that used to use GIT_CURL_VERBOSE to use the 250 newer GIT_TRACE_CURL. 251 252 * "git pack-objects --include-tag" was taught that when we know that 253 we are sending an object C, we want a tag B that directly points at 254 C but also a tag A that points at the tag B. We used to miss the 255 intermediate tag B in some cases. 256 257 * Update Japanese translation for "git-gui". 258 259 * "git fetch http::/site/path" did not die correctly and segfaulted 260 instead. 261 262 * "git commit-tree" stopped reading commit.gpgsign configuration 263 variable that was meant for Porcelain "git commit" in Git 2.9; we 264 forgot to update "git gui" to look at the configuration to match 265 this change. 266 267 * "git add --chmod=+x" added recently lacked documentation, which has 268 been corrected. 269 270 * "git log --cherry-pick" used to include merge commits as candidates 271 to be matched up with other commits, resulting a lot of wasted time. 272 The patch-id generation logic has been updated to ignore merges to 273 avoid the wastage. 274 275 * The http transport (with curl-multi option, which is the default 276 these days) failed to remove curl-easy handle from a curlm session, 277 which led to unnecessary API failures. 278 279 * There were numerous corner cases in which the configuration files 280 are read and used or not read at all depending on the directory a 281 Git command was run, leading to inconsistent behaviour. The code 282 to set-up repository access at the beginning of a Git process has 283 been updated to fix them. 284 (merge 4d0efa1 jk/setup-sequence-update later to maint). 285 286 * "git diff -W" output needs to extend the context backward to 287 include the header line of the current function and also forward to 288 include the body of the entire current function up to the header 289 line of the next one. This process may have to merge to adjacent 290 hunks, but the code forgot to do so in some cases. 291 292 * Performance tests done via "t/perf" did not use the same set of 293 build configuration if the user relied on autoconf generated 294 configuration. 295 296 * "git format-patch --base=..." feature that was recently added 297 showed the base commit information after "-- " e-mail signature 298 line, which turned out to be inconvenient. The base information 299 has been moved above the signature line. 300 301 * More i18n. 302 303 * Even when "git pull --rebase=preserve" (and the underlying "git 304 rebase --preserve") can complete without creating any new commit 305 (i.e. fast-forwards), it still insisted on having a usable ident 306 information (read: user.email is set correctly), which was less 307 than nice. As the underlying commands used inside "git rebase" 308 would fail with a more meaningful error message and advice text 309 when the bogus ident matters, this extra check was removed. 310 311 * "git gc --aggressive" used to limit the delta-chain length to 250, 312 which is way too deep for gaining additional space savings and is 313 detrimental for runtime performance. The limit has been reduced to 314 50. 315 316 * Documentation for individual configuration variables to control use 317 of color (like `color.grep`) said that their default value is 318 'false', instead of saying their default is taken from `color.ui`. 319 When we updated the default value for color.ui from 'false' to 320 'auto' quite a while ago, all of them broke. This has been 321 corrected. 322 323 * The pretty-format specifier "%C(auto)" used by the "log" family of 324 commands to enable coloring of the output is taught to also issue a 325 color-reset sequence to the output. 326 (merge 82b83da8d3 rs/c-auto-resets-attributes later to maint). 327 328 * A shell script example in check-ref-format documentation has been 329 fixed. 330 331 * "git checkout <word>" does not follow the usual disambiguation 332 rules when the <word> can be both a rev and a path, to allow 333 checking out a branch 'foo' in a project that happens to have a 334 file 'foo' in the working tree without having to disambiguate. 335 This was poorly documented and the check was incorrect when the 336 command was run from a subdirectory. 337 338 * Some codepaths in "git diff" used regexec(3) on a buffer that was 339 mmap(2)ed, which may not have a terminating NUL, leading to a read 340 beyond the end of the mapped region. This was fixed by introducing 341 a regexec_buf() helper that takes a <ptr,len> pair with REG_STARTEND 342 extension. 343 (merge 842a516cb0 js/regexec-buf later to maint). 344 345 * The procedure to build Git on Mac OS X for Travis CI hardcoded the 346 internal directory structure we assumed HomeBrew uses, which was a 347 no-no. The procedure has been updated to ask HomeBrew things we 348 need to know to fix this. 349 350 * When "git rebase -i" is given a broken instruction, it told the 351 user to fix it with "--edit-todo", but didn't say what the step 352 after that was (i.e. "--continue"). 353 354 * Documentation around tools to import from CVS was fairly outdated. 355 356 * "git clone --recurse-submodules" lost the progress eye-candy in 357 recent update, which has been corrected. 358 359 * A low-level function verify_packfile() was meant to show errors 360 that were detected without dying itself, but under some conditions 361 it didn't and died instead, which has been fixed. 362 363 * When "git fetch" tries to find where the history of the repository 364 it runs in has diverged from what the other side has, it has a 365 mechanism to avoid digging too deep into irrelevant side branches. 366 This however did not work well over the "smart-http" transport due 367 to a design bug, which has been fixed. 368 (merge 06b3d386e0 jt/fetch-pack-in-vain-count-with-stateless later to maint). 369 370 * In the codepath that comes up with the hostname to be used in an 371 e-mail when the user didn't tell us, we looked at ai_canonname 372 field in struct addrinfo without making sure it is not NULL first. 373 374 * "git worktree", even though it used the default_abbrev setting that 375 ought to be affected by core.abbrev configuration variable, ignored 376 the variable setting. The command has been taught to read the 377 default set of configuration variables to correct this. 378 379 * "git init" tried to record core.worktree in the repository's 380 'config' file when GIT_WORK_TREE environment variable was set and 381 it was different from where GIT_DIR appears as ".git" at its top, 382 but the logic was faulty when .git is a "gitdir:" file that points 383 at the real place, causing trouble in working trees that are 384 managed by "git worktree". This has been corrected. 385 386 * Codepaths that read from an on-disk loose object were too loose in 387 validating what they are reading is a proper object file and 388 sometimes read past the data they read from the disk, which has 389 been corrected. H/t to Gustavo Grieco for reporting. 390 391 * The original command line syntax for "git merge", which was "git 392 merge <msg> HEAD <parent>...", has been deprecated for quite some 393 time, and "git gui" was the last in-tree user of the syntax. This 394 is finally fixed, so that we can move forward with the deprecation. 395 396 * An author name, that spelled a backslash-quoted double quote in the 397 human readable part "My \"double quoted\" name", was not unquoted 398 correctly while applying a patch from a piece of e-mail. 399 400 * Doc update to clarify what "log -3 --reverse" does. 401 402 * Almost everybody uses DEFAULT_ABBREV to refer to the default 403 setting for the abbreviation, but "git blame" peeked into 404 underlying variable bypassing the macro for no good reason. 405 406 * The "graph" API used in "git log --graph" miscounted the number of 407 output columns consumed so far when drawing a padding line, which 408 has been fixed; this did not affect any existing code as nobody 409 tried to write anything after the padding on such a line, though. 410 411 * The code that parses the format parameter of for-each-ref command 412 has seen a micro-optimization. 413 414 * When we started cURL to talk to imap server when a new enough 415 version of cURL library is available, we forgot to explicitly add 416 imap(s):// before the destination. To some folks, that didn't work 417 and the library tried to make HTTP(s) requests instead. 418 (merge d2d07ab861 ak/curl-imap-send-explicit-scheme later to maint). 419 420 * The ./configure script generated from configure.ac was taught how 421 to detect support of SSL by libcurl better. 422 (merge 924b7eb1c9 dp/autoconf-curl-ssl later to maint). 423 424 * The command-line completion script (in contrib/) learned to 425 complete "git cmd ^mas<HT>" to complete the negative end of 426 reference to "git cmd ^master". 427 (merge 49416ad22a cp/completion-negative-refs later to maint). 428 429 * The existing "git fetch --depth=<n>" option was hard to use 430 correctly when making the history of an existing shallow clone 431 deeper. A new option, "--deepen=<n>", has been added to make this 432 easier to use. "git clone" also learned "--shallow-since=<date>" 433 and "--shallow-exclude=<tag>" options to make it easier to specify 434 "I am interested only in the recent N months worth of history" and 435 "Give me only the history since that version". 436 (merge cccf74e2da nd/shallow-deepen later to maint). 437 438 * It is a common mistake to say "git blame --reverse OLD path", 439 expecting that the command line is dwimmed as if asking how lines 440 in path in an old revision OLD have survived up to the current 441 commit. 442 (merge e1d09701a4 jc/blame-reverse later to maint). 443 444 * http.emptyauth configuration is a way to allow an empty username to 445 pass when attempting to authenticate using mechanisms like 446 Kerberos. We took an unspecified (NULL) username and sent ":" 447 (i.e. no username, no password) to CURLOPT_USERPWD, but did not do 448 the same when the username is explicitly set to an empty string. 449 (merge 5275c3081c dt/http-empty-auth later to maint). 450 451 * "git clone" of a local repository can be done at the filesystem 452 level, but the codepath did not check errors while copying and 453 adjusting the file that lists alternate object stores. 454 (merge 22d3b8de1b jk/clone-copy-alternates-fix later to maint). 455 456 * Documentation for "git commit" was updated to clarify that "commit 457 -p <paths>" adds to the current contents of the index to come up 458 with what to commit. 459 (merge 7431596ab1 nd/commit-p-doc later to maint). 460 461 * A stray symbolic link in $GIT_DIR/refs/ directory could make name 462 resolution loop forever, which has been corrected. 463 (merge e8c42cb9ce jk/ref-symlink-loop later to maint). 464 465 * The "submodule.<name>.path" stored in .gitmodules is never copied 466 to .git/config and such a key in .git/config has no meaning, but 467 the documentation described it and submodule.<name>.url next to 468 each other as if both belong to .git/config. This has been fixed. 469 (merge 72710165c9 sb/submodule-config-doc-drop-path later to maint). 470 471 * In a worktree connected to a repository elsewhere, created via "git 472 worktree", "git checkout" attempts to protect users from confusion 473 by refusing to check out a branch that is already checked out in 474 another worktree. However, this also prevented checking out a 475 branch, which is designated as the primary branch of a bare 476 reopsitory, in a worktree that is connected to the bare 477 repository. The check has been corrected to allow it. 478 (merge 171c646f8c dk/worktree-dup-checkout-with-bare-is-ok later to maint). 479 480 * "git rebase" immediately after "git clone" failed to find the fork 481 point from the upstream. 482 (merge 4f21454b55 jk/merge-base-fork-point-without-reflog later to maint). 483 484 * When fetching from a remote that has many tags that are irrelevant 485 to branches we are following, we used to waste way too many cycles 486 when checking if the object pointed at by a tag (that we are not 487 going to fetch!) exists in our repository too carefully. 488 (merge 5827a03545 jk/fetch-quick-tag-following later to maint). 489 490 * Protect our code from over-eager compilers. 491 (merge 0ac52a38e8 jk/tighten-alloc later to maint). 492 493 * Recent git allows submodule.<name>.branch to use a special token 494 "." instead of the branch name; the documentation has been updated 495 to describe it. 496 (merge 15ef78008a bw/submodule-branch-dot-doc later to maint). 497 498 * A hot-fix for a test added by a recent topic that went to both 499 'master' and 'maint' already. 500 (merge 76e368c378 tg/add-chmod+x-fix later to maint). 501 502 * "git send-email" attempts to pick up valid e-mails from the 503 trailers, but people in real world write non-addresses there, like 504 "Cc: Stable <add@re.ss> # 4.8+", which broke the output depending 505 on the availability and vintage of Mail::Address perl module. 506 (merge dcfafc5214 mm/send-email-cc-cruft-after-address later to maint). 507 508 * The Travis CI configuration we ship ran the tests with --verbose 509 option but this risks non-TAP output that happens to be "ok" to be 510 misinterpreted as TAP signalling a test that passed. This resulted 511 in unnecessary failure. This has been corrected by introducing a 512 new mode to run our tests in the test harness to send the verbose 513 output separately to the log file. 514 (merge 614fe01521 jk/tap-verbose-fix later to maint). 515 516 * Some AsciiDoc formatter mishandles a displayed illustration with 517 tabs in it. Adjust a few of them in merge-base documentation to 518 work around them. 519 (merge 6750f62699 po/fix-doc-merge-base-illustration later to maint). 520 521 * Other minor doc, test and build updates and code cleanups. 522 (merge a94bb68397 rs/cocci later to maint). 523 (merge 641c900b2c js/reset-usage later to maint). 524 (merge 30cfe72d37 rs/pretty-format-color-doc-fix later to maint). 525 (merge d709f1fb9d jc/diff-unique-abbrev-comments later to maint). 526 (merge 13092a916d jc/cocci-xstrdup-or-null later to maint). 527 (merge 86009f32bb pb/test-parse-options-expect later to maint). 528 (merge 749a2279a4 yk/git-tag-remove-mention-of-old-layout-in-doc later to maint).