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 * In bare repositories, "git shortlog" and other commands now read 37 mailmap files from the tip of the history, to help running these 38 tools in server settings. 39 40 * Color specifiers, e.g. "%C(blue)Hello%C(reset)", used in the 41 "--format=" option of "git log" and friends can be disabled when 42 the output is not sent to a terminal by prefixing them with 43 "auto,", e.g. "%C(auto,blue)Hello%C(auto,reset)". 44 45 * Scripts can ask Git that wildcard patterns in pathspecs they give do 46 not have any significance, i.e. take them as literal strings. 47 48 * The patterns in .gitignore and .gitattributes files can have **/, 49 as a pattern that matches 0 or more levels of subdirectory. 50 E.g. "foo/**/bar" matches "bar" in "foo" itself or in a 51 subdirectory of "foo". 52 53 * When giving arguments without "--" disambiguation, object names 54 that come earlier on the command line must not be interpretable as 55 pathspecs and pathspecs that come later on the command line must 56 not be interpretable as object names. This disambiguation rule has 57 been tweaked so that ":/" (no other string before or after) is 58 always interpreted as a pathspec; "git cmd -- :/" is no longer 59 needed, you can just say "git cmd :/". 60 61 * Various "hint" lines Git gives when it asks the user to edit 62 messages in the editor are commented out with '#' by default. The 63 core.commentchar configuration variable can be used to customize 64 this '#' to a different character. 65 66 * "git add -u" and "git add -A" without pathspec issues warning to 67 make users aware that they are only operating on paths inside the 68 subdirectory they are in. Use ":/" (everything from the top) or 69 "." (everything from the $cwd) to disambiguate. 70 71 * "git blame" (and "git diff") learned the "--no-follow" option. 72 73 * "git check-ignore" command to help debugging .gitignore files has 74 been added. 75 76 * "git cherry-pick" can be used to replay a root commit to an unborn 77 branch. 78 79 * "git commit" can be told to use --cleanup=whitespace by setting the 80 configuration variable commit.cleanup to 'whitespace'. 81 82 * "git fetch --mirror" and fetch that uses other forms of refspec 83 with wildcard used to attempt to update a symbolic ref that match 84 the wildcard on the receiving end, which made little sense (the 85 real ref that is pointed at by the symbolic ref would be updated 86 anyway). Symbolic refs no longer are affected by such a fetch. 87 88 * "git format-patch" now detects more cases in which a whole branch 89 is being exported, and uses the description for the branch, when 90 asked to write a cover letter for the series. 91 92 * "git format-patch" learned "-v $count" option, and prepends a 93 string "v$count-" to the names of its output files, and also 94 automatically sets the subject prefix to "PATCH v$count". This 95 allows patches from rerolled series to be stored under different 96 names and makes it easier to reuse cover letter messsages. 97 98 * "git log" and friends can be told with --use-mailmap option to 99 rewrite the names and email addresses of people using the mailmap 100 mechanism. 101 102 * "git push" now requires "-f" to update a tag, even if it is a 103 fast-forward, as tags are meant to be fixed points. 104 105 * Error messages from "git push" when it stops to prevent remote refs 106 from getting overwritten by mistake have been improved to explain 107 various situations separately. 108 109 * "git push" will stop without doing anything if the new "pre-push" 110 hook exists and exits with a failure. 111 112 * When "git rebase" fails to generate patches to be applied (e.g. due 113 to oom), it failed to detect the failure and instead behaved as if 114 there were nothing to do. A workaround to use a temporary file has 115 been applied, but we probably would want to revisit this later, as 116 it hurts the common case of not failing at all. 117 118 * Input and preconditions to "git reset" has been loosened where 119 appropriate. "git reset $fromtree Makefile" requires $fromtree to 120 be any tree (it used to require it to be a commit), for example. 121 "git reset" (without options or parameters) used to error out when 122 you do not have any commits in your history, but it now gives you 123 an empty index (to match non-existent commit you are not even on). 124 125 * "git submodule" started learning a new mode to integrate with the 126 tip of the remote branch (as opposed to integrating with the commit 127 recorded in the superproject's gitlink). 128 129 130Foreign Interface 131 132 * "git fast-export" has been updated for its use in the context of 133 the remote helper interface. 134 135 * A new remote helper to interact with bzr has been added to contrib/. 136 137 * "git p4" got various bugfixes around its branch handling. It is 138 also made usable with Python 2.4/2.5. In addition, its various 139 portability issues for Cygwin have been addressed. 140 141 * The remote helper to interact with Hg in contrib/ has seen a few 142 fixes. 143 144 145Performance, Internal Implementation, etc. 146 147 * "git fsck" has been taught to be pickier about entries in tree 148 objects that should not be there, e.g. ".", ".git", and "..". 149 150 * Matching paths with common forms of pathspecs that contain wildcard 151 characters has been optimized further. 152 153 * We stopped paying attention to $GIT_CONFIG environment that points 154 at a single configuration file from any command other than "git config" 155 quite a while ago, but "git clone" internally set, exported, and 156 then unexported the variable during its operation unnecessarily. 157 158 * "git reset" internals has been reworked and should be faster in 159 general. We tried to be careful not to break any behaviour but 160 there could be corner cases, especially when running the command 161 from a conflicted state, that we may have missed. 162 163 * The implementation of "imap-send" has been updated to reuse xml 164 quoting code from http-push codepath, and lost a lot of unused 165 code. 166 167 * There is a simple-minded checker for the test scripts in t/ 168 directory to catch most common mistakes (it is not enabled by 169 default). 170 171 * You can build with USE_WILDMATCH=YesPlease to use a replacement 172 implementation of pattern matching logic used for pathname-like 173 things, e.g. refnames and paths in the repository. This new 174 implementation is not expected change the existing behaviour of Git 175 in this release, except for "git for-each-ref" where you can now 176 say "refs/**/master" and match with both refs/heads/master and 177 refs/remotes/origin/master. We plan to use this new implementation 178 in wider places (e.g. "git ls-files '**/Makefile' may find Makefile 179 at the top-level, and "git log '**/t*.sh'" may find commits that 180 touch a shell script whose name begins with "t" at any level) in 181 future versions of Git, but we are not there yet. By building with 182 USE_WILDMATCH, using the resulting Git daily and reporting when you 183 find breakages, you can help us get closer to that goal. 184 185 * Some reimplementations of Git do not write all the stat info back 186 to the index due to their implementation limitations (e.g. jgit). 187 A configuration option can tell Git to ignore changes to most of 188 the stat fields and only pay attention to mtime and size, which 189 these implementations can reliably update. This can be used to 190 avoid excessive revalidation of contents. 191 192 193Also contains minor documentation updates and code clean-ups. 194 195 196Fixes since v1.8.1 197------------------ 198 199Unless otherwise noted, all the fixes since v1.8.1 in the maintenance 200track are contained in this release (see release notes to them for 201details). 202 203 * An element on GIT_CEILING_DIRECTORIES list that does not name the 204 real path to a directory (i.e. a symbolic link) could have caused 205 the GIT_DIR discovery logic to escape the ceiling. 206 207 * When attempting to read the XDG-style $HOME/.config/git/config and 208 finding that $HOME/.config/git is a file, we gave a wrong error 209 message, instead of treating the case as "a custom config file does 210 not exist there" and moving on. 211 212 * The behaviour visible to the end users was confusing, when they 213 attempt to kill a process spawned in the editor that was in turn 214 launched by Git with SIGINT (or SIGQUIT), as Git would catch that 215 signal and die. We ignore these signals now. 216 (merge 0398fc34 pf/editor-ignore-sigint later to maint). 217 218 * A child process that was killed by a signal (e.g. SIGINT) was 219 reported in an inconsistent way depending on how the process was 220 spawned by us, with or without a shell in between. 221 222 * After failing to create a temporary file using mkstemp(), failing 223 pathname was not reported correctly on some platforms. 224 225 * We used to stuff "user@" and then append what we read from 226 /etc/mailname to come up with a default e-mail ident, but a bug 227 lost the "user@" part. 228 229 * The attribute mechanism didn't allow limiting attributes to be 230 applied to only a single directory itself with "path/" like the 231 exclude mechanism does. The initial implementation of this that 232 was merged to 'maint' and 1.8.1.2 was with a severe performance 233 degradations and needs to merge a fix-up topic. 234 235 * "git am" did not parse datestamp correctly from Hg generated patch, 236 when it is run in a locale outside C (or en). 237 238 * "git apply" misbehaved when fixing whitespace breakages by removing 239 excess trailing blank lines. 240 241 * A tar archive created by "git archive" recorded a directory in a 242 way that made NetBSD's implementation of "tar" sometimes unhappy. 243 244 * "git archive" did not record uncompressed size in the header when 245 streaming a zip archive, which confused some implementations of unzip. 246 247 * "git archive" did not parse configuration values in tar.* namespace 248 correctly. 249 (merge b3873c3 jk/config-parsing-cleanup later to maint). 250 251 * Attempt to "branch --edit-description" an existing branch, while 252 being on a detached HEAD, errored out. 253 254 * "git clean" showed what it was going to do, but sometimes end up 255 finding that it was not allowed to do so, which resulted in a 256 confusing output (e.g. after saying that it will remove an 257 untracked directory, it found an embedded git repository there 258 which it is not allowed to remove). It now performs the actions 259 and then reports the outcome more faithfully. 260 (merge f538a91 zk/clean-report-failure later to maint). 261 262 * When "git clone --separate-git-dir=$over_there" is interrupted, it 263 failed to remove the real location of the $GIT_DIR it created. 264 This was most visible when interrupting a submodule update. 265 266 * "git fetch --depth" was broken in at least three ways. The 267 resulting history was deeper than specified by one commit, it was 268 unclear how to wipe the shallowness of the repository with the 269 command, and documentation was misleading. 270 (merge cfb70e1 nd/fetch-depth-is-broken later to maint). 271 272 * "git log --all -p" that walked refs/notes/textconv/ ref can later 273 try to use the textconv data incorrectly after it gets freed. 274 (merge be5c9fb jk/read-commit-buffer-data-after-free later to maint). 275 276 * The way "git svn" asked for password using SSH_ASKPASS and 277 GIT_ASKPASS was not in line with the rest of the system. 278 279 * The --graph code fell into infinite loop when asked to do what the 280 code did not expect. 281 282 * http transport was wrong to ask for the username when the 283 authentication is done by certificate identity. 284 285 * "git pack-refs" that ran in parallel to another process that 286 created new refs had a nasty race. 287 288 * After "git add -N" and then writing a tree object out of the 289 index, the cache-tree data structure got corrupted. 290 291 * "git clone" used to allow --bare and --separate-git-dir=$there 292 options at the same time, which was nonsensical. 293 (merge 95b63f1 nd/clone-no-separate-git-dir-with-bare later to maint). 294 295 * "git rebase --preserve-merges" lost empty merges in recent versions 296 of Git. 297 298 * "git merge --no-edit" computed who were involved in the work done 299 on the side branch, even though that information is to be discarded 300 without getting seen in the editor. 301 302 * "git merge" started calling prepare-commit-msg hook like "git 303 commit" does some time ago, but forgot to pay attention to the exit 304 status of the hook. 305 306 * When users spell "cc:" in lowercase in the fake "header" in the 307 trailer part, "git send-email" failed to pick up the addresses from 308 there. As e-mail headers field names are case insensitive, this 309 script should follow suit and treat "cc:" and "Cc:" the same way. 310 311 * Output from "git status --ignored" showed an unexpected interaction 312 with "--untracked". 313 314 * "gitweb", when sorting by age to show repositories with new 315 activities first, used to sort repositories with absolutely 316 nothing in it early, which was not very useful. 317 318 * "gitweb"'s code to sanitize control characters before passing it to 319 "highlight" filter lost known-to-be-safe control characters by 320 mistake. 321 322 * When a line to be wrapped has a solid run of non space characters 323 whose length exactly is the wrap width, "git shortlog -w" failed 324 to add a newline after such a line. 325 326 * Command line completion leaked an unnecessary error message while 327 looking for possible matches with paths in <tree-ish>. 328 329 * Command line completion for "tcsh" emitted an unwanted space 330 after completing a single directory name. 331 332 * Command line completion code was inadvertently made incompatible with 333 older versions of bash by using a newer array notation. 334 335 * Some shells do not behave correctly when IFS is unset; work it 336 around by explicitly setting it to the default value. 337 338 * Some scripted programs written in Python did not get updated when 339 PYTHON_PATH changed. 340 (cherry-pick 96a4647fca54031974cd6ad1 later to maint). 341 342 * When autoconf is used, any build on a different commit always ran 343 "config.status --recheck" even when unnecessary. 344 345 * We have been carrying a translated and long-unmaintained copy of an 346 old version of the tutorial; removed. 347 348 * t0050 had tests expecting failures from a bug that was fixed some 349 time ago. 350 351 * t4014, t9502 and t0200 tests had various portability issues that 352 broke on OpenBSD. 353 354 * t9020 and t3600 tests had various portability issues. 355 356 * t9200 runs "cvs init" on a directory that already exists, but a 357 platform can configure this fail for the current user (e.g. you 358 need to be in the cvsadmin group on NetBSD 6.0). 359 360 * t9020 and t9810 had a few non-portable shell script construct. 361 362 * Scripts to test bash completion was inherently flaky as it was 363 affected by whatever random things the user may have on $PATH. 364 (merge 5047822 jc/do-not-let-random-file-interfere-with-completion-tests later to maint).