1Git 2.10 Release Notes 2====================== 3 4Backward compatibility notes 5---------------------------- 6 7Updates since v2.9 8------------------ 9 10UI, Workflows & Features 11 12 * "git pull --rebase --verify-signature" learned to warn the user 13 that "--verify-signature" is a no-op when rebasing. 14 15 * An upstream project can make a recommendation to shallowly clone 16 some submodules in the .gitmodules file it ships. 17 18 * "git worktree add" learned that '-' can be used as a short-hand for 19 "@{-1}", the previous branch. 20 21 * Update the funcname definition to support css files. 22 23 * The completion script (in contrib/) learned to complete "git 24 status" options. 25 26 * Messages that are generated by auto gc during "git push" on the 27 receiving end are now passed back to the sending end in such a way 28 that they are shown with "remote: " prefix to avoid confusing the 29 users. 30 31 * "git add -i/-p" learned to honor diff.compactionHeuristic 32 experimental knob, so that the user can work on the same hunk split 33 as "git diff" output. 34 35 * "upload-pack" allows a custom "git pack-objects" replacement when 36 responding to "fetch/clone" via the uploadpack.packObjectsHook. 37 (merge b738396 jk/upload-pack-hook later to maint). 38 39 * Teach format-patch and mailsplit (hence "am") how a line that 40 happens to begin with "From " in the e-mail message is quoted with 41 ">", so that these lines can be restored to their original shape. 42 (merge d9925d1 ew/mboxrd-format-am later to maint). 43 44 * "git repack" learned the "--keep-unreachable" option, which sends 45 loose unreachable objects to a pack instead of leaving them loose. 46 This helps heuristics based on the number of loose objects 47 (e.g. "gc --auto"). 48 (merge e26a8c4 jk/repack-keep-unreachable later to maint). 49 50 * "log --graph --format=" learned that "%>|(N)" specifies the width 51 relative to the terminal's left edge, not relative to the area to 52 draw text that is to the right of the ancestry-graph section. It 53 also now accepts negative N that means the column limit is relative 54 to the right border. 55 56 * A careless invocation of "git send-email directory/" after editing 57 0001-change.patch with an editor often ends up sending both 58 0001-change.patch and its backup file, 0001-change.patch~, causing 59 embarrassment and a minor confusion. Detect such an input and 60 offer to skip the backup files when sending the patches out. 61 (merge 531220b jc/send-email-skip-backup later to maint). 62 63 * "git submodule update" that drives many "git clone" could 64 eventually hit flaky servers/network conditions on one of the 65 submodules; the command learned to retry the attempt. 66 67 * The output coloring scheme learned two new attributes, italic and 68 strike, in addition to existing bold, reverse, etc. 69 70 * "git log" learns log.showSignature configuration variable, and a 71 command line option "--no-show-signature" to countermand it. 72 (merge fce04c3 mj/log-show-signature-conf later to maint). 73 74 * More markings of messages for i18n, with updates to various tests 75 to pass GETTEXT_POISON tests. 76 77 * "git archive" learned to handle files that are larger than 8GB and 78 commits far in the future than expressible by the traditional US-TAR 79 format. 80 (merge 560b0e8 jk/big-and-future-archive-tar later to maint). 81 82 83 * A new configuration variable core.sshCommand has been added to 84 specify what value for GIT_SSH_COMMAND to use per repository. 85 86 * "git worktree prune" protected worktrees that are marked as 87 "locked" by creating a file in a known location. "git worktree" 88 command learned a dedicated command pair to create and remove such 89 a file, so that the users do not have to do this with editor. 90 91 * A handful of "git svn" updates. 92 93 * "git push" learned to accept and pass extra options to the 94 receiving end so that hooks can read and react to them. 95 96 * "git status" learned to suggest "merge --abort" during a conflicted 97 merge, just like it already suggests "rebase --abort" during a 98 conflicted rebase. 99 100 * "git jump" script (in contrib/) has been updated a bit. 101 (merge a91e692 jk/git-jump later to maint). 102 103 * "git push" and "git clone" learned to give better progress meters 104 to the end user who is waiting on the terminal. 105 106 * An entry "git log --decorate" for the tip of the current branch is 107 shown as "HEAD -> name" (where "name" is the name of the branch); 108 the arrow is now painted in the same color as "HEAD", not in the 109 color for commits. 110 111 * "git format-patch" learned format.from configuration variable to 112 specify the default settings for its "--from" option. 113 114 * "git am -3" calls "git merge-recursive" when it needs to fall back 115 to a three-way merge; this call has been turned into an internal 116 subroutine call instead of spawning a separate subprocess. 117 118 * The command line completion scripts (in contrib/) now knows about 119 "git branch --delete/--move [--remote]". 120 (merge 2703c22 vs/completion-branch-fully-spelled-d-m-r later to maint). 121 122 * "git rev-parse --git-path hooks/<hook>" learned to take 123 core.hooksPath configuration variable (introduced during 2.9 cycle) 124 into account. 125 (merge 9445b49 ab/hooks later to maint). 126 127 * "git log --show-signature" and other commands that display the 128 verification status of PGP signature now shows the longer key-id, 129 as 32-bit key-id is so last century. 130 131 132Performance, Internal Implementation, Development Support etc. 133 134 * "git fast-import" learned the same performance trick to avoid 135 creating too small a packfile as "git fetch" and "git push" have, 136 using *.unpackLimit configuration. 137 138 * When "git daemon" is run without --[init-]timeout specified, a 139 connection from a client that silently goes offline can hang around 140 for a long time, wasting resources. The socket-level KEEPALIVE has 141 been enabled to allow the OS to notice such failed connections. 142 143 * "git upload-pack" command has been updated to use the parse-options 144 API. 145 146 * The "git apply" standalone program is being libified; the first 147 step to move many state variables into a structure that can be 148 explicitly (re)initialized to make the machinery callable more 149 than once has been merged. 150 151 * HTTP transport gained an option to produce more detailed debugging 152 trace. 153 (merge 73e57aa ep/http-curl-trace later to maint). 154 155 * Instead of taking advantage of the fact that a struct string_list 156 that is allocated with all NULs happens to be the INIT_NODUP kind, 157 the users of string_list structures are taught to initialize them 158 explicitly as such, to document their behaviour better. 159 (merge 2721ce2 jk/string-list-static-init later to maint). 160 161 * HTTPd tests learned to show the server error log to help diagnosing 162 a failing tests. 163 (merge 44f243d nd/test-lib-httpd-show-error-log-in-verbose later to maint). 164 165 * The ownership rule for the piece of memory that hold references to 166 be fetched in "git fetch" was screwy, which has been cleaned up. 167 168 * "git bisect" makes an internal call to "git diff-tree" when 169 bisection finds the culprit, but this call did not initialize the 170 data structure to pass to the diff-tree API correctly. 171 172 * Further preparatory clean-up for "worktree" feature continues. 173 (merge 0409e0b nd/worktree-cleanup-post-head-protection later to maint). 174 175 * Formats of the various data (and how to validate them) where we use 176 GPG signature have been documented. 177 178 * A new run-command API function pipe_command() is introduced to 179 sanely feed data to the standard input while capturing data from 180 the standard output and the standard error of an external process, 181 which is cumbersome to hand-roll correctly without deadlocking. 182 183 * The codepath to sign data in a prepared buffer with GPG has been 184 updated to use this API to read from the status-fd to check for 185 errors (instead of relying on GPG's exit status). 186 (merge efee955 jk/gpg-interface-cleanup later to maint). 187 188 * Allow t/perf framework to use the features from the most recent 189 version of Git even when testing an older installed version. 190 191 * The commands in the "log/diff" family have had an FILE* pointer in the 192 data structure they pass around for a long time, but some codepaths 193 used to always write to the standard output. As a preparatory step 194 to make "git format-patch" available to the internal callers, these 195 codepaths have been updated to consistently write into that FILE* 196 instead. 197 198 * Conversion from unsigned char sha1[20] to struct object_id 199 continues. 200 201 * Improve the look of the way "git fetch" reports what happened to 202 each ref that was fetched. 203 204 * The .c/.h sources are marked as such in our .gitattributes file so 205 that "git diff -W" and friends would work better. 206 207 * Code clean-up to avoid using a variable string that compilers may 208 feel untrustable as printf-style format given to write_file() 209 helper function. 210 211 * "git p4" used a location outside $GIT_DIR/refs/ to place its 212 temporary branches, which has been moved to refs/git-p4-tmp/. 213 214 * Existing autoconf generated test for the need to link with pthread 215 library did not check all the functions from pthread libraries; 216 recent FreeBSD has some functions in libc but not others, and we 217 mistakenly thought linking with libc is enough when it is not. 218 219 * When "git fsck" reports a broken link (e.g. a tree object contains 220 a blob that does not exist), both containing object and the object 221 that is referred to were reported with their 40-hex object names. 222 The command learned the "--name-objects" option to show the path to 223 the containing object from existing refs (e.g. "HEAD~24^2:file.txt"). 224 225 * Allow http daemon tests in Travis CI tests. 226 227 * Makefile assumed that -lrt is always available on platforms that 228 want to use clock_gettime() and CLOCK_MONOTONIC, which is not a 229 case for recent Mac OS X. The necessary symbols are often found in 230 libc on many modern systems and having -lrt on the command line, as 231 long as the library exists, had no effect, but when the platform 232 removes librt.a that is a different matter--having -lrt will break 233 the linkage. 234 235 This change could be seen as a regression for those who do need to 236 specify -lrt, as they now specifically ask for NEEDS_LIBRT when 237 building. Hopefully they are in the minority these days. 238 239 * Further preparatory work on the refs API before the pluggable 240 backend series can land. 241 242 * Error handling in the codepaths that updates refs has been 243 improved. 244 245 * The API to iterate over all the refs (i.e. for_each_ref(), etc.) 246 has been revamped. 247 248 * The handling of the "text=auto" attribute has been corrected. 249 $ echo "* text=auto eol=crlf" >.gitattributes 250 used to have the same effect as 251 $ echo "* text eol=crlf" >.gitattributes 252 i.e. declaring all files are text (ignoring "auto"). The 253 combination has been fixed to be equivalent to doing 254 $ git config core.autocrlf true 255 256 * Documentation has been updated to show better example usage 257 of the updated "text=auto" attribute. 258 259 * A few tests that specifically target "git rebase -i" have been 260 added. 261 262 * Dumb http transport on the client side has been optimized. 263 (merge ecba195 ew/http-walker later to maint). 264 265 * Users of the parse_options_concat() API function need to allocate 266 extra slots in advance and fill them with OPT_END() when they want 267 to decide the set of supported options dynamically, which makes the 268 code error-prone and hard to read. This has been corrected by tweaking 269 the API to allocate and return a new copy of "struct option" array. 270 271 * "git fetch" exchanges batched have/ack messages between the sender 272 and the receiver, initially doubling every time and then falling 273 back to enlarge the window size linearly. The "smart http" 274 transport, being an half-duplex protocol, outgrows the preset limit 275 too quickly and becomes inefficient when interacting with a large 276 repository. The internal mechanism learned to grow the window size 277 more aggressively when working with the "smart http" transport. 278 279 * Tests for "git svn" have been taught to reuse the lib-httpd test 280 infrastructure when testing the subversion integration that 281 interacts with subversion repositories served over the http:// 282 protocol. 283 (merge a8a5d25 ew/git-svn-http-tests later to maint). 284 285 * "git pack-objects" has a few options that tell it not to pack 286 objects found in certain packfiles, which require it to scan .idx 287 files of all available packs. The codepaths involved in these 288 operations have been optimized for a common case of not having any 289 non-local pack and/or any .kept pack. 290 291 * The t3700 test about "add --chmod=-x" have been made a bit more 292 robust and generally cleaned up. 293 (merge 766cdc4 ib/t3700-add-chmod-x-updates later to maint). 294 295 * The build procedure learned PAGER_ENV knob that lists what default 296 environment variable settings to export for popular pagers. This 297 mechanism is used to tweak the default settings to MORE on FreeBSD. 298 (merge 995bc22 ew/build-time-pager-tweaks later to maint). 299 300 * The http-backend (the server-side component of smart-http 301 transport) used to trickle the HTTP header one at a time. Now 302 these write(2)s are batched. 303 (merge b36045c ew/http-backend-batch-headers later to maint). 304 305 * When "git rebase" tries to compare set of changes on the updated 306 upstream and our own branch, it computes patch-id for all of these 307 changes and attempts to find matches. This has been optimized by 308 lazily computing the full patch-id (which is expensive) to be 309 compared only for changes that touch the same set of paths. 310 (merge ba67504 kw/patch-ids-optim later to maint). 311 312 * A handful of tests that were broken under gettext-poison build have 313 been fixed. 314 315 * The recent i18n patch we added during this cycle did a bit too much 316 refactoring of the messages to avoid word-legos; the repetition has 317 been reduced to help translators. 318 319 320Also contains various documentation updates and code clean-ups. 321 322 323Fixes since v2.9 324---------------- 325 326Unless otherwise noted, all the fixes since v2.8 in the maintenance 327track are contained in this release (see the maintenance releases' 328notes for details). 329 330 * The commands in `git log` family take %C(auto) in a custom format 331 string. This unconditionally turned the color on, ignoring 332 --no-color or with --color=auto when the output is not connected to 333 a tty; this was corrected to make the format truly behave as 334 "auto". 335 336 * "git rev-list --count" whose walk-length is limited with "-n" 337 option did not work well with the counting optimized to look at the 338 bitmap index. 339 340 * "git show -W" (extend hunks to cover the entire function, delimited 341 by lines that match the "funcname" pattern) used to show the entire 342 file when a change added an entire function at the end of the file, 343 which has been fixed. 344 345 * The documentation set has been updated so that literal commands, 346 configuration variables and environment variables are consistently 347 typeset in fixed-width font and bold in manpages. 348 349 * "git svn propset" subcommand that was added in 2.3 days is 350 documented now. 351 352 * The documentation tries to consistently spell "GPG"; when 353 referring to the specific program name, "gpg" is used. 354 355 * "git reflog" stopped upon seeing an entry that denotes a branch 356 creation event (aka "unborn"), which made it appear as if the 357 reflog was truncated. 358 359 * The git-prompt scriptlet (in contrib/) was not friendly with those 360 who uses "set -u", which has been fixed. 361 362 * compat/regex code did not cleanly compile. 363 364 * A codepath that used alloca(3) to place an unbounded amount of data 365 on the stack has been updated to avoid doing so. 366 367 * "git update-index --add --chmod=+x file" may be usable as an escape 368 hatch, but not a friendly thing to force for people who do need to 369 use it regularly. "git add --chmod=+x file" can be used instead. 370 371 * Build improvements for gnome-keyring (in contrib/) 372 373 * "git status" used to say "working directory" when it meant "working 374 tree". 375 376 * Comments about misbehaving FreeBSD shells have been clarified with 377 the version number (9.x and before are broken, newer ones are OK). 378 379 * "git cherry-pick A" worked on an unborn branch, but "git 380 cherry-pick A..B" didn't. 381 382 * Fix an unintended regression in v2.9 that breaks "clone --depth" 383 that recurses down to submodules by forcing the submodules to also 384 be cloned shallowly, which many server instances that host upstream 385 of the submodules are not prepared for. 386 387 * Fix unnecessarily waste in the idiomatic use of ': ${VAR=default}' 388 to set the default value, without enclosing it in double quotes. 389 390 * Some platform-specific code had non-ANSI strict declarations of C 391 functions that do not take any parameters, which has been 392 corrected. 393 394 * The internal code used to show local timezone offset is not 395 prepared to handle timestamps beyond year 2100, and gave a 396 bogus offset value to the caller. Use a more benign looking 397 +0000 instead and let "git log" going in such a case, instead 398 of aborting. 399 400 * One among four invocations of readlink(1) in our test suite has 401 been rewritten so that the test can run on systems without the 402 command (others are in valgrind test framework and t9802). 403 404 * t/perf needs /usr/bin/time with GNU extension; the invocation of it 405 is updated to "gtime" on Darwin. 406 407 * A bug, which caused "git p4" while running under verbose mode to 408 report paths that are omitted due to branch prefix incorrectly, has 409 been fixed; the command said "Ignoring file outside of prefix" for 410 paths that are _inside_. 411 412 * The top level documentation "git help git" still pointed at the 413 documentation set hosted at now-defunct google-code repository. 414 Update it to point to https://git.github.io/htmldocs/git.html 415 instead. 416 417 * A helper function that takes the contents of a commit object and 418 finds its subject line did not ignore leading blank lines, as is 419 commonly done by other codepaths. Make it ignore leading blank 420 lines to match. 421 422 * For a long time, we carried an in-code comment that said our 423 colored output would work only when we use fprintf/fputs on 424 Windows, which no longer is the case for the past few years. 425 426 * "gc.autoPackLimit" when set to 1 should not trigger a repacking 427 when there is only one pack, but the code counted poorly and did 428 so. 429 430 * Add a test to specify the desired behaviour that currently is not 431 available in "git rebase -Xsubtree=...". 432 433 * More mark-up updates to typeset strings that are expected to 434 literally typed by the end user in fixed-width font. 435 436 * "git commit --amend --allow-empty-message -S" for a commit without 437 any message body could have misidentified where the header of the 438 commit object ends. 439 440 * "git rebase -i --autostash" did not restore the auto-stashed change 441 when the operation was aborted. 442 443 * Git does not know what the contents in the index should be for a 444 path added with "git add -N" yet, so "git grep --cached" should not 445 show hits (or show lack of hits, with -L) in such a path, but that 446 logic does not apply to "git grep", i.e. searching in the working 447 tree files. But we did so by mistake, which has been corrected. 448 449 * "git blame -M" missed a single line that was moved within the file. 450 451 * Fix recently introduced codepaths that are involved in parallel 452 submodule operations, which gave up on reading too early, and 453 could have wasted CPU while attempting to write under a corner 454 case condition. 455 456 * "git grep -i" has been taught to fold case in non-ascii locales 457 correctly. 458 459 * A test that unconditionally used "mktemp" learned that the command 460 is not necessarily available everywhere. 461 462 * There are certain house-keeping tasks that need to be performed at 463 the very beginning of any Git program, and programs that are not 464 built-in commands had to do them exactly the same way as "git" 465 potty does. It was easy to make mistakes in one-off standalone 466 programs (like test helpers). A common "main()" function that 467 calls cmd_main() of individual program has been introduced to 468 make it harder to make mistakes. 469 (merge de61ceb jk/common-main later to maint). 470 471 * The test framework learned a new helper test_match_signal to 472 check an exit code from getting killed by an expected signal. 473 474 * General code clean-up around a helper function to write a 475 single-liner to a file. 476 (merge 7eb6e10 jk/write-file later to maint). 477 478 * One part of "git am" had an oddball helper function that called 479 stuff from outside "his" as opposed to calling what we have "ours", 480 which was not gender-neutral and also inconsistent with the rest of 481 the system where outside stuff is usuall called "theirs" in 482 contrast to "ours". 483 484 * "git blame file" allowed the lineage of lines in the uncommitted, 485 unadded contents of "file" to be inspected, but it refused when 486 "file" did not appear in the current commit. When "file" was 487 created by renaming an existing file (but the change has not been 488 committed), this restriction was unnecessarily tight. 489 490 * "git add -N dir/file && git write-tree" produced an incorrect tree 491 when there are other paths in the same directory that sorts after 492 "file". 493 494 * "git fetch http://user:pass@host/repo..." scrubbed the userinfo 495 part, but "git push" didn't. 496 497 * "git merge" with renormalization did not work well with 498 merge-recursive, due to "safer crlf" conversion kicking in when it 499 shouldn't. 500 (merge 1335d76 jc/renormalize-merge-kill-safer-crlf later to maint). 501 502 * The use of strbuf in "git rm" to build filename to remove was a bit 503 suboptimal, which has been fixed. 504 505 * An age old bug that caused "git diff --ignore-space-at-eol" 506 misbehave has been fixed. 507 508 * "git notes merge" had a code to see if a path exists (and fails if 509 it does) and then open the path for writing (when it doesn't). 510 Replace it with open with O_EXCL. 511 512 * "git pack-objects" and "git index-pack" mostly operate with off_t 513 when talking about the offset of objects in a packfile, but there 514 were a handful of places that used "unsigned long" to hold that 515 value, leading to an unintended truncation. 516 517 * Recent update to "git daemon" tries to enable the socket-level 518 KEEPALIVE, but when it is spawned via inetd, the standard input 519 file descriptor may not necessarily be connected to a socket. 520 Suppress an ENOTSOCK error from setsockopt(). 521 522 * Recent FreeBSD stopped making perl available at /usr/bin/perl; 523 switch the default the built-in path to /usr/local/bin/perl on not 524 too ancient FreeBSD releases. 525 526 * "git commit --help" said "--no-verify" is only about skipping the 527 pre-commit hook, and failed to say that it also skipped the 528 commit-msg hook. 529 530 * "git merge" in Git v2.9 was taught to forbid merging an unrelated 531 lines of history by default, but that is exactly the kind of thing 532 the "--rejoin" mode of "git subtree" (in contrib/) wants to do. 533 "git subtree" has been taught to use the "--allow-unrelated-histories" 534 option to override the default. 535 536 * The build procedure for "git persistent-https" helper (in contrib/) 537 has been updated so that it can be built with more recent versions 538 of Go. 539 540 * There is an optimization used in "git diff $treeA $treeB" to borrow 541 an already checked-out copy in the working tree when it is known to 542 be the same as the blob being compared, expecting that open/mmap of 543 such a file is faster than reading it from the object store, which 544 involves inflating and applying delta. This however kicked in even 545 when the checked-out copy needs to go through the convert-to-git 546 conversion (including the clean filter), which defeats the whole 547 point of the optimization. The optimization has been disabled when 548 the conversion is necessary. 549 550 * "git -c grep.patternType=extended log --basic-regexp" misbehaved 551 because the internal API to access the grep machinery was not 552 designed well. 553 554 * Windows port was failing some tests in t4130, due to the lack of 555 inum in the returned values by its lstat(2) emulation. 556 557 * The reflog output format is documented better, and a new format 558 --date=unix to report the seconds-since-epoch (without timezone) 559 has been added. 560 (merge 442f6fd jk/reflog-date later to maint). 561 562 * "git difftool <paths>..." started in a subdirectory failed to 563 interpret the paths relative to that directory, which has been 564 fixed. 565 566 * The characters in the label shown for tags/refs for commits in 567 "gitweb" output are now properly escaped for proper HTML output. 568 569 * FreeBSD can lie when asked mtime of a directory, which made the 570 untracked cache code to fall back to a slow-path, which in turn 571 caused tests in t7063 to fail because it wanted to verify the 572 behaviour of the fast-path. 573 574 * Squelch compiler warnings for nedmalloc (in compat/) library. 575 576 * A small memory leak in the command line parsing of "git blame" 577 has been plugged. 578 579 * The API documentation for hashmap was unclear if hashmap_entry 580 can be safely discarded without any other consideration. State 581 that it is safe to do so. 582 583 * Not-so-recent rewrite of "git am" that started making internal 584 calls into the commit machinery had an unintended regression, in 585 that no matter how many seconds it took to apply many patches, the 586 resulting committer timestamp for the resulting commits were all 587 the same. 588 589 * "git push --force-with-lease" already had enough logic to allow 590 ensuring that such a push results in creation of a ref (i.e. the 591 receiving end did not have another push from sideways that would be 592 discarded by our force-pushing), but didn't expose this possibility 593 to the users. It does so now. 594 (merge 9eed4f3 jk/push-force-with-lease-creation later to maint). 595 596 * The mechanism to limit the pack window memory size, when packing is 597 done using multiple threads (which is the default), is per-thread, 598 but this was not documented clearly. 599 (merge 954176c ms/document-pack-window-memory-is-per-thread later to maint). 600 601 * "import-tars" fast-import script (in contrib/) used to ignore a 602 hardlink target and replaced it with an empty file, which has been 603 corrected to record the same blob as the other file the hardlink is 604 shared with. 605 (merge 04e0869 js/import-tars-hardlinks later to maint). 606 607 * "git mv dir non-existing-dir/" did not work in some environments 608 the same way as existing mainstream platforms. The code now moves 609 "dir" to "non-existing-dir", without relying on rename("A", "B/") 610 that strips the trailing slash of '/'. 611 (merge 189d035 js/mv-dir-to-new-directory later to maint). 612 613 * The "t/" hierarchy is prone to get an unusual pathname; "make test" 614 has been taught to make sure they do not contain paths that cannot 615 be checked out on Windows (and the mechanism can be reusable to 616 catch pathnames that are not portable to other platforms as need 617 arises). 618 (merge c2cafd3 js/test-lint-pathname later to maint). 619 620 * When "git merge-recursive" works on history with many criss-cross 621 merges in "verbose" mode, the names the command assigns to the 622 virtual merge bases could have overwritten each other by unintended 623 reuse of the same piece of memory. 624 (merge 5447a76 rs/pull-signed-tag later to maint). 625 626 * "git checkout --detach <branch>" used to give the same advice 627 message as that is issued when "git checkout <tag>" (or anything 628 that is not a branch name) is given, but asking with "--detach" is 629 an explicit enough sign that the user knows what is going on. The 630 advice message has been squelched in this case. 631 (merge 779b88a sb/checkout-explit-detach-no-advice later to maint). 632 633 * "git difftool" by default ignores the error exit from the backend 634 commands it spawns, because often they signal that they found 635 differences by exiting with a non-zero status code just like "diff" 636 does; the exit status codes 126 and above however are special in 637 that they are used to signal that the command is not executable, 638 does not exist, or killed by a signal. "git difftool" has been 639 taught to notice these exit status codes. 640 (merge 45a4f5d jk/difftool-command-not-found later to maint). 641 642 * On Windows, help.browser configuration variable used to be ignored, 643 which has been corrected. 644 (merge 6db5967 js/no-html-bypass-on-windows later to maint). 645 646 * The "git -c var[=val] cmd" facility to append a configuration 647 variable definition at the end of the search order was described in 648 git(1) manual page, but not in git-config(1), which was more likely 649 place for people to look for when they ask "can I make a one-shot 650 override, and if so how?" 651 (merge ae1f709 dg/document-git-c-in-git-config-doc later to maint). 652 653 * The tempfile (hence its user lockfile) API lets the caller to open 654 a file descriptor to a temporary file, write into it and then 655 finalize it by first closing the filehandle and then either 656 removing or renaming the temporary file. When the process spawns a 657 subprocess after obtaining the file descriptor, and if the 658 subprocess has not exited when the attempt to remove or rename is 659 made, the last step fails on Windows, because the subprocess has 660 the file descriptor still open. Open tempfile with O_CLOEXEC flag 661 to avoid this (on Windows, this is mapped to O_NOINHERIT). 662 (merge 05d1ed6 bw/mingw-avoid-inheriting-fd-to-lockfile later to maint). 663 664 * Correct an age-old calco (is that a typo-like word for calc) 665 in the documentation. 666 (merge 7841c48 ls/packet-line-protocol-doc-fix later to maint). 667 668 * Other minor clean-ups and documentation updates 669 (merge 02a8cfa rs/merge-add-strategies-simplification later to maint). 670 (merge af4941d rs/merge-recursive-string-list-init later to maint). 671 (merge 1eb47f1 rs/use-strbuf-add-unique-abbrev later to maint). 672 (merge ddd0bfa jk/tighten-alloc later to maint). 673 (merge ecf30b2 rs/mailinfo-lib later to maint). 674 (merge 0eb75ce sg/reflog-past-root later to maint). 675 (merge 4369523 hv/doc-commit-reference-style later to maint).