1Git v1.8.2 Release Notes 2======================== 3 4Backward compatibility notes 5---------------------------- 6 7In the upcoming major release (tentatively called 1.8.2), we will 8change the behavior of the "git push" command. 9 10When "git push [$there]" does not say what to push, we have used the 11traditional "matching" semantics so far (all your branches were sent 12to the remote as long as there already are branches of the same name 13over there). We will use the "simple" semantics that pushes the 14current branch to the branch with the same name, only when the current 15branch is set to integrate with that remote branch. There is a user 16preference configuration variable "push.default" to change this. 17 18"git push $there tag v1.2.3" used to allow replacing a tag v1.2.3 19that already exists in the repository $there, if the rewritten tag 20you are pushing points at a commit that is a decendant of a commit 21that the old tag v1.2.3 points at. This was found to be error prone 22and starting with this release, any attempt to update an existing 23ref under refs/tags/ hierarchy will fail, without "--force". 24 25 26Updates since v1.8.1 27-------------------- 28 29UI, Workflows & Features 30 31 * Initial ports to QNX and z/OS UNIX System Services have started. 32 33 * Output from the tests is coloured using "green is okay, yellow is 34 questionable, red is bad and blue is informative" scheme. 35 36 * Mention of "GIT/Git/git" in the documentation have been updated to 37 be more uniform and consistent. The name of the system and the 38 concept it embodies is "Git"; the command the users type is "git". 39 All-caps "GIT" was merely a way to imitate "Git" typeset in small 40 caps in our ASCII text only documentation and to be avoided. 41 42 * In bare repositories, "git shortlog" and other commands now read 43 mailmap files from the tip of the history, to help running these 44 tools in server settings. 45 46 * Color specifiers, e.g. "%C(blue)Hello%C(reset)", used in the 47 "--format=" option of "git log" and friends can be disabled when 48 the output is not sent to a terminal by prefixing them with 49 "auto,", e.g. "%C(auto,blue)Hello%C(auto,reset)". 50 51 * Scripts can ask Git that wildcard patterns in pathspecs they give do 52 not have any significance, i.e. take them as literal strings. 53 54 * The patterns in .gitignore and .gitattributes files can have **/, 55 as a pattern that matches 0 or more levels of subdirectory. 56 E.g. "foo/**/bar" matches "bar" in "foo" itself or in a 57 subdirectory of "foo". 58 59 * When giving arguments without "--" disambiguation, object names 60 that come earlier on the command line must not be interpretable as 61 pathspecs and pathspecs that come later on the command line must 62 not be interpretable as object names. This disambiguation rule has 63 been tweaked so that ":/" (no other string before or after) is 64 always interpreted as a pathspec; "git cmd -- :/" is no longer 65 needed, you can just say "git cmd :/". 66 67 * Various "hint" lines Git gives when it asks the user to edit 68 messages in the editor are commented out with '#' by default. The 69 core.commentchar configuration variable can be used to customize 70 this '#' to a different character. 71 72 * "git add -u" and "git add -A" without pathspec issues warning to 73 make users aware that they are only operating on paths inside the 74 subdirectory they are in. Use ":/" (everything from the top) or 75 "." (everything from the $cwd) to disambiguate. 76 77 * "git blame" (and "git diff") learned the "--no-follow" option. 78 79 * "git branch" now rejects some nonsense combinations of command line 80 arguments (e.g. giving more than one branch name to rename) with 81 more case-specific error messages. 82 83 * "git check-ignore" command to help debugging .gitignore files has 84 been added. 85 86 * "git cherry-pick" can be used to replay a root commit to an unborn 87 branch. 88 89 * "git commit" can be told to use --cleanup=whitespace by setting the 90 configuration variable commit.cleanup to 'whitespace'. 91 92 * "git fetch --mirror" and fetch that uses other forms of refspec 93 with wildcard used to attempt to update a symbolic ref that match 94 the wildcard on the receiving end, which made little sense (the 95 real ref that is pointed at by the symbolic ref would be updated 96 anyway). Symbolic refs no longer are affected by such a fetch. 97 98 * "git format-patch" now detects more cases in which a whole branch 99 is being exported, and uses the description for the branch, when 100 asked to write a cover letter for the series. 101 102 * "git format-patch" learned "-v $count" option, and prepends a 103 string "v$count-" to the names of its output files, and also 104 automatically sets the subject prefix to "PATCH v$count". This 105 allows patches from rerolled series to be stored under different 106 names and makes it easier to reuse cover letter messsages. 107 108 * "git log" and friends can be told with --use-mailmap option to 109 rewrite the names and email addresses of people using the mailmap 110 mechanism. 111 112 * "git mergetool" and "git difftool" learned to list the available 113 tool backends in a more consistent manner. 114 115 * "git mergetool" is aware of TortoiseGitMerge now and uses it over 116 TortoiseMerge when available. 117 118 * "git push" now requires "-f" to update a tag, even if it is a 119 fast-forward, as tags are meant to be fixed points. 120 121 * Error messages from "git push" when it stops to prevent remote refs 122 from getting overwritten by mistake have been improved to explain 123 various situations separately. 124 125 * "git push" will stop without doing anything if the new "pre-push" 126 hook exists and exits with a failure. 127 128 * When "git rebase" fails to generate patches to be applied (e.g. due 129 to oom), it failed to detect the failure and instead behaved as if 130 there were nothing to do. A workaround to use a temporary file has 131 been applied, but we probably would want to revisit this later, as 132 it hurts the common case of not failing at all. 133 134 * Input and preconditions to "git reset" has been loosened where 135 appropriate. "git reset $fromtree Makefile" requires $fromtree to 136 be any tree (it used to require it to be a commit), for example. 137 "git reset" (without options or parameters) used to error out when 138 you do not have any commits in your history, but it now gives you 139 an empty index (to match non-existent commit you are not even on). 140 141 * "git submodule" started learning a new mode to integrate with the 142 tip of the remote branch (as opposed to integrating with the commit 143 recorded in the superproject's gitlink). 144 145 146Foreign Interface 147 148 * "git fast-export" has been updated for its use in the context of 149 the remote helper interface. 150 151 * A new remote helper to interact with bzr has been added to contrib/. 152 153 * "git p4" got various bugfixes around its branch handling. It is 154 also made usable with Python 2.4/2.5. In addition, its various 155 portability issues for Cygwin have been addressed. 156 157 * The remote helper to interact with Hg in contrib/ has seen a few 158 fixes. 159 160 161Performance, Internal Implementation, etc. 162 163 * "git fsck" has been taught to be pickier about entries in tree 164 objects that should not be there, e.g. ".", ".git", and "..". 165 166 * Matching paths with common forms of pathspecs that contain wildcard 167 characters has been optimized further. 168 169 * We stopped paying attention to $GIT_CONFIG environment that points 170 at a single configuration file from any command other than "git config" 171 quite a while ago, but "git clone" internally set, exported, and 172 then unexported the variable during its operation unnecessarily. 173 174 * "git reset" internals has been reworked and should be faster in 175 general. We tried to be careful not to break any behaviour but 176 there could be corner cases, especially when running the command 177 from a conflicted state, that we may have missed. 178 179 * The implementation of "imap-send" has been updated to reuse xml 180 quoting code from http-push codepath, and lost a lot of unused 181 code. 182 183 * There is a simple-minded checker for the test scripts in t/ 184 directory to catch most common mistakes (it is not enabled by 185 default). 186 187 * You can build with USE_WILDMATCH=YesPlease to use a replacement 188 implementation of pattern matching logic used for pathname-like 189 things, e.g. refnames and paths in the repository. This new 190 implementation is not expected change the existing behaviour of Git 191 in this release, except for "git for-each-ref" where you can now 192 say "refs/**/master" and match with both refs/heads/master and 193 refs/remotes/origin/master. We plan to use this new implementation 194 in wider places (e.g. "git ls-files '**/Makefile' may find Makefile 195 at the top-level, and "git log '**/t*.sh'" may find commits that 196 touch a shell script whose name begins with "t" at any level) in 197 future versions of Git, but we are not there yet. By building with 198 USE_WILDMATCH, using the resulting Git daily and reporting when you 199 find breakages, you can help us get closer to that goal. 200 201 * Some reimplementations of Git do not write all the stat info back 202 to the index due to their implementation limitations (e.g. jgit). 203 A configuration option can tell Git to ignore changes to most of 204 the stat fields and only pay attention to mtime and size, which 205 these implementations can reliably update. This can be used to 206 avoid excessive revalidation of contents. 207 208 209Also contains minor documentation updates and code clean-ups. 210 211 212Fixes since v1.8.1 213------------------ 214 215Unless otherwise noted, all the fixes since v1.8.1 in the maintenance 216track are contained in this release (see release notes to them for 217details). 218 219 * An element on GIT_CEILING_DIRECTORIES list that does not name the 220 real path to a directory (i.e. a symbolic link) could have caused 221 the GIT_DIR discovery logic to escape the ceiling. 222 223 * When attempting to read the XDG-style $HOME/.config/git/config and 224 finding that $HOME/.config/git is a file, we gave a wrong error 225 message, instead of treating the case as "a custom config file does 226 not exist there" and moving on. 227 228 * The behaviour visible to the end users was confusing, when they 229 attempt to kill a process spawned in the editor that was in turn 230 launched by Git with SIGINT (or SIGQUIT), as Git would catch that 231 signal and die. We ignore these signals now. 232 (merge 0398fc34 pf/editor-ignore-sigint later to maint). 233 234 * A child process that was killed by a signal (e.g. SIGINT) was 235 reported in an inconsistent way depending on how the process was 236 spawned by us, with or without a shell in between. 237 238 * After failing to create a temporary file using mkstemp(), failing 239 pathname was not reported correctly on some platforms. 240 241 * We used to stuff "user@" and then append what we read from 242 /etc/mailname to come up with a default e-mail ident, but a bug 243 lost the "user@" part. 244 245 * The attribute mechanism didn't allow limiting attributes to be 246 applied to only a single directory itself with "path/" like the 247 exclude mechanism does. The initial implementation of this that 248 was merged to 'maint' and 1.8.1.2 was with a severe performance 249 degradations and needs to merge a fix-up topic. 250 251 * "git am" did not parse datestamp correctly from Hg generated patch, 252 when it is run in a locale outside C (or en). 253 254 * "git apply" misbehaved when fixing whitespace breakages by removing 255 excess trailing blank lines. 256 257 * A tar archive created by "git archive" recorded a directory in a 258 way that made NetBSD's implementation of "tar" sometimes unhappy. 259 260 * "git archive" did not record uncompressed size in the header when 261 streaming a zip archive, which confused some implementations of unzip. 262 263 * "git archive" did not parse configuration values in tar.* namespace 264 correctly. 265 (merge b3873c3 jk/config-parsing-cleanup later to maint). 266 267 * Attempt to "branch --edit-description" an existing branch, while 268 being on a detached HEAD, errored out. 269 270 * "git clean" showed what it was going to do, but sometimes end up 271 finding that it was not allowed to do so, which resulted in a 272 confusing output (e.g. after saying that it will remove an 273 untracked directory, it found an embedded git repository there 274 which it is not allowed to remove). It now performs the actions 275 and then reports the outcome more faithfully. 276 (merge f538a91 zk/clean-report-failure later to maint). 277 278 * When "git clone --separate-git-dir=$over_there" is interrupted, it 279 failed to remove the real location of the $GIT_DIR it created. 280 This was most visible when interrupting a submodule update. 281 282 * "git fetch --depth" was broken in at least three ways. The 283 resulting history was deeper than specified by one commit, it was 284 unclear how to wipe the shallowness of the repository with the 285 command, and documentation was misleading. 286 (merge cfb70e1 nd/fetch-depth-is-broken later to maint). 287 288 * "git log --all -p" that walked refs/notes/textconv/ ref can later 289 try to use the textconv data incorrectly after it gets freed. 290 (merge be5c9fb jk/read-commit-buffer-data-after-free later to maint). 291 292 * We forgot to close the file descriptor reading from "gpg" output, 293 killing "git log --show-signature" on a long history. 294 295 * The way "git svn" asked for password using SSH_ASKPASS and 296 GIT_ASKPASS was not in line with the rest of the system. 297 298 * The --graph code fell into infinite loop when asked to do what the 299 code did not expect. 300 301 * http transport was wrong to ask for the username when the 302 authentication is done by certificate identity. 303 304 * "git pack-refs" that ran in parallel to another process that 305 created new refs had a nasty race. 306 307 * Rebasing the history of superproject with change in the submodule 308 has been broken since v1.7.12. 309 310 * After "git add -N" and then writing a tree object out of the 311 index, the cache-tree data structure got corrupted. 312 313 * "git clone" used to allow --bare and --separate-git-dir=$there 314 options at the same time, which was nonsensical. 315 (merge 95b63f1 nd/clone-no-separate-git-dir-with-bare later to maint). 316 317 * "git rebase --preserve-merges" lost empty merges in recent versions 318 of Git. 319 320 * "git merge --no-edit" computed who were involved in the work done 321 on the side branch, even though that information is to be discarded 322 without getting seen in the editor. 323 324 * "git merge" started calling prepare-commit-msg hook like "git 325 commit" does some time ago, but forgot to pay attention to the exit 326 status of the hook. 327 328 * A failure to push due to non-ff while on an unborn branch 329 dereferenced a NULL pointer when showing an error message. 330 331 * When users spell "cc:" in lowercase in the fake "header" in the 332 trailer part, "git send-email" failed to pick up the addresses from 333 there. As e-mail headers field names are case insensitive, this 334 script should follow suit and treat "cc:" and "Cc:" the same way. 335 336 * Output from "git status --ignored" showed an unexpected interaction 337 with "--untracked". 338 339 * "gitweb", when sorting by age to show repositories with new 340 activities first, used to sort repositories with absolutely 341 nothing in it early, which was not very useful. 342 343 * "gitweb"'s code to sanitize control characters before passing it to 344 "highlight" filter lost known-to-be-safe control characters by 345 mistake. 346 347 * "gitweb" pages served over HTTPS, when configured to show picon or 348 gravatar, referred to these external resources to be fetched via 349 HTTP, resulting in mixed contents warning in browsers. 350 (merge 5748558 ab/gitweb-use-same-scheme later to maint). 351 352 * When a line to be wrapped has a solid run of non space characters 353 whose length exactly is the wrap width, "git shortlog -w" failed 354 to add a newline after such a line. 355 356 * Command line completion leaked an unnecessary error message while 357 looking for possible matches with paths in <tree-ish>. 358 359 * Command line completion for "tcsh" emitted an unwanted space 360 after completing a single directory name. 361 362 * Command line completion code was inadvertently made incompatible with 363 older versions of bash by using a newer array notation. 364 365 * Some shells do not behave correctly when IFS is unset; work it 366 around by explicitly setting it to the default value. 367 368 * Some scripted programs written in Python did not get updated when 369 PYTHON_PATH changed. 370 (cherry-pick 96a4647fca54031974cd6ad1 later to maint). 371 372 * When autoconf is used, any build on a different commit always ran 373 "config.status --recheck" even when unnecessary. 374 375 * A fix was added to the build procedure to work around buggy 376 versions of ccache broke the auto-generation of dependencies, which 377 unfortunately is still relevant because some people use ancient 378 distros. 379 380 * We have been carrying a translated and long-unmaintained copy of an 381 old version of the tutorial; removed. 382 383 * t0050 had tests expecting failures from a bug that was fixed some 384 time ago. 385 386 * t4014, t9502 and t0200 tests had various portability issues that 387 broke on OpenBSD. 388 389 * t9020 and t3600 tests had various portability issues. 390 391 * t9200 runs "cvs init" on a directory that already exists, but a 392 platform can configure this fail for the current user (e.g. you 393 need to be in the cvsadmin group on NetBSD 6.0). 394 395 * t9020 and t9810 had a few non-portable shell script construct. 396 397 * Scripts to test bash completion was inherently flaky as it was 398 affected by whatever random things the user may have on $PATH.