1Git v1.8.2 Release Notes
2========================
34
Backward compatibility notes
5----------------------------
67
In the next major release Git 2.0 (not *this* one), we will change the
8behavior of the "git push" command.
910
When "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.
1718
"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".
2425
When "git add -u" and "git add -A", that does not specify what paths
26to add on the command line, is run from inside a subdirectory, the
27scope of the operation has always been limited to the subirectory.
28Many users found this counter-intuitive, given that "git commit -a"
29and other commands operate on the entire tree regardless of where you
30are. In this release, these commands give warning in such a case and
31encourage the user to say "git add -u/-A ." instead when restricting
32the scope to the current directory. At Git 2.0 (not *this* one), we
33plan to change these commands without pathspec to operate on the
34entire tree, and training your fingers to type "." will protect you
35against the future change.
3637
38
Updates since v1.8.1
39--------------------
4041
UI, Workflows & Features
4243
* Initial ports to QNX and z/OS UNIX System Services have started.
4445
* Output from the tests is coloured using "green is okay, yellow is
46questionable, red is bad and blue is informative" scheme.
4748
* Mention of "GIT/Git/git" in the documentation have been updated to
49be more uniform and consistent. The name of the system and the
50concept it embodies is "Git"; the command the users type is "git".
51All-caps "GIT" was merely a way to imitate "Git" typeset in small
52caps in our ASCII text only documentation and to be avoided.
5354
* The completion script (in contrib/completion) used to let the
55default completer to suggest pathnames, which gave too many
56irrelevant choices (e.g. "git add" would not want to add an
57unmodified path). It learnt to use a more git-aware logic to
58enumerate only relevant ones.
5960
* In bare repositories, "git shortlog" and other commands now read
61mailmap files from the tip of the history, to help running these
62tools in server settings.
6364
* Color specifiers, e.g. "%C(blue)Hello%C(reset)", used in the
65"--format=" option of "git log" and friends can be disabled when
66the output is not sent to a terminal by prefixing them with
67"auto,", e.g. "%C(auto,blue)Hello%C(auto,reset)".
6869
* Scripts can ask Git that wildcard patterns in pathspecs they give do
70not have any significance, i.e. take them as literal strings.
7172
* The patterns in .gitignore and .gitattributes files can have **/,
73as a pattern that matches 0 or more levels of subdirectory.
74E.g. "foo/**/bar" matches "bar" in "foo" itself or in a
75subdirectory of "foo".
7677
* When giving arguments without "--" disambiguation, object names
78that come earlier on the command line must not be interpretable as
79pathspecs and pathspecs that come later on the command line must
80not be interpretable as object names. This disambiguation rule has
81been tweaked so that ":/" (no other string before or after) is
82always interpreted as a pathspec; "git cmd -- :/" is no longer
83needed, you can just say "git cmd :/".
8485
* Various "hint" lines Git gives when it asks the user to edit
86messages in the editor are commented out with '#' by default. The
87core.commentchar configuration variable can be used to customize
88this '#' to a different character.
8990
* "git add -u" and "git add -A" without pathspec issues warning to
91make users aware that they are only operating on paths inside the
92subdirectory they are in. Use ":/" (everything from the top) or
93"." (everything from the $cwd) to disambiguate.
9495
* "git blame" (and "git diff") learned the "--no-follow" option.
9697
* "git branch" now rejects some nonsense combinations of command line
98arguments (e.g. giving more than one branch name to rename) with
99more case-specific error messages.
100101
* "git check-ignore" command to help debugging .gitignore files has
102been added.
103104
* "git cherry-pick" can be used to replay a root commit to an unborn
105branch.
106107
* "git commit" can be told to use --cleanup=whitespace by setting the
108configuration variable commit.cleanup to 'whitespace'.
109110
* "git diff" and other Porcelain commands can be told to use a
111non-standard algorithm by setting diff.algorithm configuration
112variable.
113114
* "git fetch --mirror" and fetch that uses other forms of refspec
115with wildcard used to attempt to update a symbolic ref that match
116the wildcard on the receiving end, which made little sense (the
117real ref that is pointed at by the symbolic ref would be updated
118anyway). Symbolic refs no longer are affected by such a fetch.
119120
* "git format-patch" now detects more cases in which a whole branch
121is being exported, and uses the description for the branch, when
122asked to write a cover letter for the series.
123124
* "git format-patch" learned "-v $count" option, and prepends a
125string "v$count-" to the names of its output files, and also
126automatically sets the subject prefix to "PATCH v$count". This
127allows patches from rerolled series to be stored under different
128names and makes it easier to reuse cover letter messsages.
129130
* "git log" and friends can be told with --use-mailmap option to
131rewrite the names and email addresses of people using the mailmap
132mechanism.
133134
* "git log --cc --graph" now shows the combined diff output with the
135ancestry graph.
136137
* "git log --grep=<pattern>" honors i18n.logoutputencoding to look
138for the pattern after fixing the log message to the specified
139encoding.
140141
* "git mergetool" and "git difftool" learned to list the available
142tool backends in a more consistent manner.
143144
* "git mergetool" is aware of TortoiseGitMerge now and uses it over
145TortoiseMerge when available.
146147
* "git push" now requires "-f" to update a tag, even if it is a
148fast-forward, as tags are meant to be fixed points.
149150
* Error messages from "git push" when it stops to prevent remote refs
151from getting overwritten by mistake have been improved to explain
152various situations separately.
153154
* "git push" will stop without doing anything if the new "pre-push"
155hook exists and exits with a failure.
156157
* When "git rebase" fails to generate patches to be applied (e.g. due
158to oom), it failed to detect the failure and instead behaved as if
159there were nothing to do. A workaround to use a temporary file has
160been applied, but we probably would want to revisit this later, as
161it hurts the common case of not failing at all.
162163
* Input and preconditions to "git reset" has been loosened where
164appropriate. "git reset $fromtree Makefile" requires $fromtree to
165be any tree (it used to require it to be a commit), for example.
166"git reset" (without options or parameters) used to error out when
167you do not have any commits in your history, but it now gives you
168an empty index (to match non-existent commit you are not even on).
169170
* "git status" says what branch is being bisected or rebased when
171able, not just "bisecting" or "rebasing".
172173
* "git submodule" started learning a new mode to integrate with the
174tip of the remote branch (as opposed to integrating with the commit
175recorded in the superproject's gitlink).
176177
* "git upload-pack" which implements the service "ls-remote" and
178"fetch" talk to can be told to hide ref hierarchies the server
179side internally uses (and that clients have no business learning
180about) with transfer.hiderefs configuration.
181182
183
Foreign Interface
184185
* "git fast-export" has been updated for its use in the context of
186the remote helper interface.
187188
* A new remote helper to interact with bzr has been added to contrib/.
189190
* "git p4" got various bugfixes around its branch handling. It is
191also made usable with Python 2.4/2.5. In addition, its various
192portability issues for Cygwin have been addressed.
193194
* The remote helper to interact with Hg in contrib/ has seen a few
195fixes.
196197
198
Performance, Internal Implementation, etc.
199200
* "git fsck" has been taught to be pickier about entries in tree
201objects that should not be there, e.g. ".", ".git", and "..".
202203
* Matching paths with common forms of pathspecs that contain wildcard
204characters has been optimized further.
205206
* We stopped paying attention to $GIT_CONFIG environment that points
207at a single configuration file from any command other than "git config"
208quite a while ago, but "git clone" internally set, exported, and
209then unexported the variable during its operation unnecessarily.
210211
* "git reset" internals has been reworked and should be faster in
212general. We tried to be careful not to break any behaviour but
213there could be corner cases, especially when running the command
214from a conflicted state, that we may have missed.
215216
* The implementation of "imap-send" has been updated to reuse xml
217quoting code from http-push codepath, and lost a lot of unused
218code.
219220
* There is a simple-minded checker for the test scripts in t/
221directory to catch most common mistakes (it is not enabled by
222default).
223224
* You can build with USE_WILDMATCH=YesPlease to use a replacement
225implementation of pattern matching logic used for pathname-like
226things, e.g. refnames and paths in the repository. This new
227implementation is not expected change the existing behaviour of Git
228in this release, except for "git for-each-ref" where you can now
229say "refs/**/master" and match with both refs/heads/master and
230refs/remotes/origin/master. We plan to use this new implementation
231in wider places (e.g. "git ls-files '**/Makefile' may find Makefile
232at the top-level, and "git log '**/t*.sh'" may find commits that
233touch a shell script whose name begins with "t" at any level) in
234future versions of Git, but we are not there yet. By building with
235USE_WILDMATCH, using the resulting Git daily and reporting when you
236find breakages, you can help us get closer to that goal.
237238
* Some reimplementations of Git do not write all the stat info back
239to the index due to their implementation limitations (e.g. jgit).
240A configuration option can tell Git to ignore changes to most of
241the stat fields and only pay attention to mtime and size, which
242these implementations can reliably update. This can be used to
243avoid excessive revalidation of contents.
244245
* Some platforms ship with old version of expat where xmlparse.h
246needs to be included instead of expat.h; the build procedure has
247been taught about this.
248249
* "make clean" on platforms that cannot compute header dependencies
250on the fly did not work with implementations of "rm" that do not
251like an empty argument list.
252253
Also contains minor documentation updates and code clean-ups.
254255
256
Fixes since v1.8.1
257------------------
258259
Unless otherwise noted, all the fixes since v1.8.1 in the maintenance
260track are contained in this release (see release notes to them for
261details).
262263
* An element on GIT_CEILING_DIRECTORIES list that does not name the
264real path to a directory (i.e. a symbolic link) could have caused
265the GIT_DIR discovery logic to escape the ceiling.
266267
* When attempting to read the XDG-style $HOME/.config/git/config and
268finding that $HOME/.config/git is a file, we gave a wrong error
269message, instead of treating the case as "a custom config file does
270not exist there" and moving on.
271272
* The behaviour visible to the end users was confusing, when they
273attempt to kill a process spawned in the editor that was in turn
274launched by Git with SIGINT (or SIGQUIT), as Git would catch that
275signal and die. We ignore these signals now.
276(merge 0398fc34 pf/editor-ignore-sigint later to maint).
277278
* A child process that was killed by a signal (e.g. SIGINT) was
279reported in an inconsistent way depending on how the process was
280spawned by us, with or without a shell in between.
281282
* After failing to create a temporary file using mkstemp(), failing
283pathname was not reported correctly on some platforms.
284285
* We used to stuff "user@" and then append what we read from
286/etc/mailname to come up with a default e-mail ident, but a bug
287lost the "user@" part.
288289
* The attribute mechanism didn't allow limiting attributes to be
290applied to only a single directory itself with "path/" like the
291exclude mechanism does. The initial implementation of this that
292was merged to 'maint' and 1.8.1.2 was with a severe performance
293degradations and needs to merge a fix-up topic.
294295
* The smart HTTP clients forgot to verify the content-type that comes
296back from the server side to make sure that the request is being
297handled properly.
298(merge 3443db5 sp/smart-http-content-type-check later to maint).
299300
* "git am" did not parse datestamp correctly from Hg generated patch,
301when it is run in a locale outside C (or en).
302303
* "git apply" misbehaved when fixing whitespace breakages by removing
304excess trailing blank lines.
305306
* "git apply --summary" has been taught to make sure the similarity
307value shown in its output is sensible, even when the input had a
308bogus value.
309(merge afcb6ac jk/apply-similaritly-parsing later to maint).
310311
* A tar archive created by "git archive" recorded a directory in a
312way that made NetBSD's implementation of "tar" sometimes unhappy.
313314
* "git archive" did not record uncompressed size in the header when
315streaming a zip archive, which confused some implementations of unzip.
316317
* "git archive" did not parse configuration values in tar.* namespace
318correctly.
319(merge b3873c3 jk/config-parsing-cleanup later to maint).
320321
* Attempt to "branch --edit-description" an existing branch, while
322being on a detached HEAD, errored out.
323324
* "git clean" showed what it was going to do, but sometimes end up
325finding that it was not allowed to do so, which resulted in a
326confusing output (e.g. after saying that it will remove an
327untracked directory, it found an embedded git repository there
328which it is not allowed to remove). It now performs the actions
329and then reports the outcome more faithfully.
330(merge f538a91 zk/clean-report-failure later to maint).
331332
* When "git clone --separate-git-dir=$over_there" is interrupted, it
333failed to remove the real location of the $GIT_DIR it created.
334This was most visible when interrupting a submodule update.
335336
* "git cvsimport" mishandled timestamps at DST boundary.
337(merge 48c9162 bw/get-tz-offset-perl later to maint).
338339
* We used to have an arbitrary 32 limit for combined diff input,
340resulting in incorrect number of leading colons shown when showing
341the "--raw --cc" output.
342(merge edbc00e jc/combine-diff-many-parents later to maint).
343344
* "git fetch --depth" was broken in at least three ways. The
345resulting history was deeper than specified by one commit, it was
346unclear how to wipe the shallowness of the repository with the
347command, and documentation was misleading.
348(merge cfb70e1 nd/fetch-depth-is-broken later to maint).
349350
* "git log --all -p" that walked refs/notes/textconv/ ref can later
351try to use the textconv data incorrectly after it gets freed.
352(merge be5c9fb jk/read-commit-buffer-data-after-free later to maint).
353354
* We forgot to close the file descriptor reading from "gpg" output,
355killing "git log --show-signature" on a long history.
356357
* The way "git svn" asked for password using SSH_ASKPASS and
358GIT_ASKPASS was not in line with the rest of the system.
359360
* The --graph code fell into infinite loop when asked to do what the
361code did not expect.
362363
* http transport was wrong to ask for the username when the
364authentication is done by certificate identity.
365366
* "git pack-refs" that ran in parallel to another process that
367created new refs had a nasty race.
368369
* Rebasing the history of superproject with change in the submodule
370has been broken since v1.7.12.
371372
* After "git add -N" and then writing a tree object out of the
373index, the cache-tree data structure got corrupted.
374375
* "git clone" used to allow --bare and --separate-git-dir=$there
376options at the same time, which was nonsensical.
377(merge 95b63f1 nd/clone-no-separate-git-dir-with-bare later to maint).
378379
* "git rebase --preserve-merges" lost empty merges in recent versions
380of Git.
381382
* "git merge --no-edit" computed who were involved in the work done
383on the side branch, even though that information is to be discarded
384without getting seen in the editor.
385386
* "git merge" started calling prepare-commit-msg hook like "git
387commit" does some time ago, but forgot to pay attention to the exit
388status of the hook.
389390
* A failure to push due to non-ff while on an unborn branch
391dereferenced a NULL pointer when showing an error message.
392393
* When users spell "cc:" in lowercase in the fake "header" in the
394trailer part, "git send-email" failed to pick up the addresses from
395there. As e-mail headers field names are case insensitive, this
396script should follow suit and treat "cc:" and "Cc:" the same way.
397398
* Output from "git status --ignored" showed an unexpected interaction
399with "--untracked".
400401
* "gitweb", when sorting by age to show repositories with new
402activities first, used to sort repositories with absolutely
403nothing in it early, which was not very useful.
404405
* "gitweb"'s code to sanitize control characters before passing it to
406"highlight" filter lost known-to-be-safe control characters by
407mistake.
408409
* "gitweb" pages served over HTTPS, when configured to show picon or
410gravatar, referred to these external resources to be fetched via
411HTTP, resulting in mixed contents warning in browsers.
412(merge 5748558 ab/gitweb-use-same-scheme later to maint).
413414
* When a line to be wrapped has a solid run of non space characters
415whose length exactly is the wrap width, "git shortlog -w" failed
416to add a newline after such a line.
417418
* Command line completion leaked an unnecessary error message while
419looking for possible matches with paths in <tree-ish>.
420421
* Command line completion for "tcsh" emitted an unwanted space
422after completing a single directory name.
423424
* Command line completion code was inadvertently made incompatible with
425older versions of bash by using a newer array notation.
426427
* "git push" was taught to refuse updating the branch that is
428currently checked out long time ago, but the user manual was left
429stale.
430(merge d9be248 wk/man-deny-current-branch-is-default-these-days later to maint).
431432
* Some shells do not behave correctly when IFS is unset; work it
433around by explicitly setting it to the default value.
434435
* Some scripted programs written in Python did not get updated when
436PYTHON_PATH changed.
437(cherry-pick 96a4647fca54031974cd6ad1 later to maint).
438439
* When autoconf is used, any build on a different commit always ran
440"config.status --recheck" even when unnecessary.
441442
* A fix was added to the build procedure to work around buggy
443versions of ccache broke the auto-generation of dependencies, which
444unfortunately is still relevant because some people use ancient
445distros.
446447
* The autoconf subsystem passed --mandir down to generated
448config.mak.autogen but forgot to do the same for --htmldir.
449(merge fc1c541 ct/autoconf-htmldir later to maint).
450451
* We have been carrying a translated and long-unmaintained copy of an
452old version of the tutorial; removed.
453454
* t0050 had tests expecting failures from a bug that was fixed some
455time ago.
456457
* t4014, t9502 and t0200 tests had various portability issues that
458broke on OpenBSD.
459460
* t9020 and t3600 tests had various portability issues.
461462
* t9200 runs "cvs init" on a directory that already exists, but a
463platform can configure this fail for the current user (e.g. you
464need to be in the cvsadmin group on NetBSD 6.0).
465466
* t9020 and t9810 had a few non-portable shell script construct.
467468
* Scripts to test bash completion was inherently flaky as it was
469affected by whatever random things the user may have on $PATH.