4 Backward compatibility notes
5 ----------------------------
7 When "git push [$there]" does not say what to push, we have used the
8 traditional "matching" semantics so far (all your branches were sent
9 to the remote as long as there already are branches of the same name
10 over there). In Git 2.0, the default is now the "simple" semantics,
13 - only the current branch to the branch with the same name, and only
14 when the current branch is set to integrate with that remote
15 branch, if you are pushing to the same remote as you fetch from; or
17 - only the current branch to the branch with the same name, if you
18 are pushing to a remote that is not where you usually fetch from.
20 You can use the configuration variable "push.default" to change
21 this. If you are an old-timer who wants to keep using the
22 "matching" semantics, you can set the variable to "matching", for
23 example. Read the documentation for other possibilities.
25 When "git add -u" and "git add -A" are run inside a subdirectory
26 without specifying which paths to add on the command line, they
27 operate on the entire tree for consistency with "git commit -a" and
28 other commands (these commands used to operate only on the current
29 subdirectory). Say "git add -u ." or "git add -A ." if you want to
30 limit the operation to the current directory.
32 "git add <path>" is the same as "git add -A <path>" now, so that
33 "git add dir/" will notice paths you removed from the directory and
34 record the removal. In older versions of Git, "git add <path>" used
35 to ignore removals. You can say "git add --ignore-removal <path>" to
36 add only added or modified paths in <path>, if you really want to.
38 The "-q" option to "git diff-files", which does *NOT* mean "quiet",
39 has been removed (it told Git to ignore deletion, which you can do
40 with "git diff-files --diff-filter=d").
42 "git request-pull" lost a few "heuristics" that often led to mistakes.
44 The default prefix for "git svn" has changed in Git 2.0. For a long
45 time, "git svn" created its remote-tracking branches directly under
46 refs/remotes, but it now places them under refs/remotes/origin/ unless
47 it is told otherwise with its --prefix option.
50 Updates since v1.9 series
51 -------------------------
53 UI, Workflows & Features
55 * The "multi-mail" post-receive hook (in contrib/) has been updated
56 to a more recent version from the upstream.
58 * The "remote-hg/bzr" remote-helper interfaces (in contrib/) are
59 now maintained separately as a third-party plug-in.
61 * "git gc --aggressive" learned "--depth" option and
62 "gc.aggressiveDepth" configuration variable to allow use of a less
63 insane depth than the built-in default value of 250.
65 * "git log" learned the "--show-linear-break" option to show where a
66 single strand-of-pearls is broken in its output.
68 * The "rev-parse --parseopt" mechanism used by scripted Porcelains to
69 parse command line options and to give help text learned to take
70 the argv-help (the placeholder string for an option parameter,
71 e.g. "key-id" in "--gpg-sign=<key-id>").
73 * The pattern to find where the function begins in C/C++ used in
74 "diff" and "grep -p" has been updated to help C++ source better.
76 * "git rebase" learned to interpret a lone "-" as "@{-1}", the
77 branch that we were previously on.
79 * "git commit --cleanup=<mode>" learned a new mode, scissors.
81 * "git tag --list" output can be sorted using "version sort" with
82 "--sort=version:refname".
84 * Discard the accumulated "heuristics" to guess from which branch the
85 result wants to be pulled from and make sure what the end user
86 specified is not second-guessed by "git request-pull", to avoid
87 mistakes. When you pushed out your 'master' branch to your public
88 repository as 'for-linus', use the new "master:for-linus" syntax to
89 denote the branch to be pulled.
91 * "git grep" learned to behave in a way similar to native grep when
92 "-h" (no header) and "-c" (count) options are given.
94 * "git push" via transport-helper interface has been updated to
95 allow forced ref updates in a way similar to the natively
98 * The "simple" mode is the default for "git push".
100 * "git add -u" and "git add -A", when run without any pathspec, is a
101 tree-wide operation even when run inside a subdirectory of a
104 * "git add <path>" is the same as "git add -A <path>" now.
106 * "core.statinfo" configuration variable, which is a
107 never-advertised synonym to "core.checkstat", has been removed.
109 * The "-q" option to "git diff-files", which does *NOT* mean
110 "quiet", has been removed (it told Git to ignore deletion, which
111 you can do with "git diff-files --diff-filter=d").
113 * Server operators can loosen the "tips of refs only" restriction for
114 the remote archive service with the uploadarchive.allowUnreachable
115 configuration option.
117 * The progress indicators from various time-consuming commands have
118 been marked for i18n/l10n.
120 * "git notes -C <blob>" diagnoses an attempt to use an object that
121 is not a blob as an error.
123 * "git config" learned to read from the standard input when "-" is
124 given as the value to its "--file" parameter (attempting an
125 operation to update the configuration in the standard input of
128 * Trailing whitespaces in .gitignore files, unless they are quoted
129 for fnmatch(3), e.g. "path\ ", are warned and ignored. Strictly
130 speaking, this is a backward incompatible change, but very unlikely
131 to bite any sane user and adjusting should be obvious and easy.
133 * Many commands that create commits, e.g. "pull", "rebase",
134 learned to take the --gpg-sign option on the command line.
136 * "git commit" can be told to always GPG sign the resulting commit
137 by setting "commit.gpgsign" configuration variable to true (the
138 command line option --no-gpg-sign should override it).
140 * "git pull" can be told to only accept fast-forward by setting the
141 new "pull.ff" configuration.
143 * "git reset" learned the "-N" option, which does not reset the index
144 fully for paths the index knows about but the tree-ish the command
145 resets to does not (these paths are kept as intend-to-add entries).
148 Performance, Internal Implementation, etc.
150 * The compilation options to port to AIX and to MSVC have been
153 * We started using wildmatch() in place of fnmatch(3) a few releases
154 ago; complete the process and stop using fnmatch(3).
156 * Uses of curl's "multi" interface and "easy" interface do not mix
157 well when we attempt to reuse outgoing connections. Teach the RPC
158 over http code, used in the smart HTTP transport, not to use the
161 * The bitmap-index feature from JGit has been ported, which should
162 significantly improve performance when serving objects from a
163 repository that uses it.
165 * The way "git log --cc" shows a combined diff against multiple
166 parents has been optimized.
168 * The prefixcmp() and suffixcmp() functions are gone. Use
169 starts_with() and ends_with(), and also consider if skip_prefix()
170 suits your needs better when using the former.
173 Also contains various documentation updates and code clean-ups. Many
174 of them came from flurry of activities as GSoC candidate microproject
178 Fixes since v1.9 series
179 -----------------------
181 Unless otherwise noted, all the fixes since v1.9 in the maintenance
182 track are contained in this release (see the maintenance releases'
185 * "git p4" was broken in 1.9 release to deal with changes in binary
187 (merge 749b668 cl/p4-use-diff-tree later to maint).
189 * The shell prompt script (in contrib/), when using the PROMPT_COMMAND
190 interface, used an unsafe construct when showing the branch name in
192 (merge 8976500 rh/prompt-pcmode-avoid-eval-on-refname later to maint).
194 * The remote-helper interface to fast-import/fast-export via the
195 transport-helper has been tightened to avoid leaving the import
196 marks file from a failed/crashed run, as such a file that is out of
197 sync with the reality confuses a later invocation of itself.
199 * "git rebase" used a POSIX shell construct FreeBSD /bin/sh does not
201 (merge 8cd6596 km/avoid-non-function-return-in-rebase later to maint).
203 * zsh prompt (in contrib/) leaked unnecessary error messages.
205 * bash completion (in contrib/) did not complete the refs and remotes
206 correctly given "git pu<TAB>" when "pu" is aliased to "push".
208 * Some more Unicode codepoints defined in Unicode 6.3 as having zero
209 width have been taught to our display column counting logic.
210 (merge d813ab9 tb/unicode-6.3-zero-width later to maint).
212 * Some tests used shell constructs that did not work well on FreeBSD
213 (merge ff7a1c6 km/avoid-bs-in-shell-glob later to maint).
214 (merge 00764ca km/avoid-cp-a later to maint).
216 * "git update-ref --stdin" did not fail a request to create a ref
217 when the ref already existed.
218 (merge b9d56b5 mh/update-ref-batch-create-fix later to maint).
220 * "git diff --no-index -Mq a b" fell into an infinite loop.
221 (merge ad1c3fb jc/fix-diff-no-index-diff-opt-parse later to maint).
223 * "git fetch --prune", when the right-hand-side of multiple fetch
224 refspecs overlap (e.g. storing "refs/heads/*" to
225 "refs/remotes/origin/*", while storing "refs/frotz/*" to
226 "refs/remotes/origin/fr/*"), aggressively thought that lack of
227 "refs/heads/fr/otz" on the origin site meant we should remove
228 "refs/remotes/origin/fr/otz" from us, without checking their
229 "refs/frotz/otz" first.
231 Note that such a configuration is inherently unsafe (think what
232 should happen when "refs/heads/fr/otz" does appear on the origin
233 site), but that is not a reason not to be extra careful.
234 (merge e6f6371 cn/fetch-prune-overlapping-destination later to maint).
236 * "git status --porcelain --branch" showed its output with labels
237 "ahead/behind/gone" translated to the user's locale.
238 (merge 7a76c28 mm/status-porcelain-format-i18n-fix later to maint).
240 * A stray environment variable $prefix could have leaked into and
241 affected the behaviour of the "subtree" script (in contrib/).
243 * When it is not necessary to edit a commit log message (e.g. "git
244 commit -m" is given a message without specifying "-e"), we used to
245 disable the spawning of the editor by overriding GIT_EDITOR, but
246 this means all the uses of the editor, other than to edit the
247 commit log message, are also affected.
248 (merge b549be0 bp/commit-p-editor later to maint).
250 * "git mv" that moves a submodule forgot to adjust the array that
251 uses to keep track of which submodules were to be moved to update
253 (merge fb8a4e8 jk/mv-submodules-fix later to maint).
255 * Length limit for the pathname used when removing a path in a deep
256 subdirectory has been removed to avoid buffer overflows.
257 (merge 2f29e0c mh/remove-subtree-long-pathname-fix later to maint).
259 * The test helper lib-terminal always run an actual test_expect_*
260 when included, which screwed up with the use of skil-all that may
261 have to be done later.
262 (merge 7e27173 jk/lib-terminal-lazy later to maint).
264 * "git index-pack" used a wrong variable to name the keep-file in an
265 error message when the file cannot be written or closed.
266 (merge de983a0 nd/index-pack-error-message later to maint).
268 * "rebase -i" produced a broken insn sheet when the title of a commit
269 happened to contain '\n' (or ended with '\c') due to a careless use
271 (merge cb1aefd us/printf-not-echo later to maint).
273 * There were a few instances of 'git-foo' remaining in the
274 documentation that should have been spelled 'git foo'.
275 (merge 3c3e6f5 rr/doc-merge-strategies later to maint).
277 * Serving objects from a shallow repository needs to write a
278 new file to hold the temporary shallow boundaries but it was not
279 cleaned when we exit due to die() or a signal.
280 (merge 7839632 jk/shallow-update-fix later to maint).
282 * When "git stash pop" stops after failing to apply the stash
283 (e.g. due to conflicting changes), the stash is not dropped. State
284 that explicitly in the output to let the users know.
285 (merge 2d4c993 jc/stash-pop-not-popped later to maint).
287 * The labels in "git status" output that describe the nature of
288 conflicts (e.g. "both deleted") were limited to 20 bytes, which was
289 too short for some l10n (e.g. fr).
290 (merge c7cb333 jn/wt-status later to maint).
292 * "git clean -d pathspec" did not use the given pathspec correctly
293 and ended up cleaning too much.
294 (merge 1f2e108 jk/clean-d-pathspec later to maint).
296 * "git difftool" misbehaved when the repository is bound to the
297 working tree with the ".git file" mechanism, where a textual file
298 ".git" tells us where it is.
299 (merge fcfec8b da/difftool-git-files later to maint).
301 * "git push" did not pay attention to branch.*.pushremote if it is
302 defined earlier than remote.pushdefault; the order of these two
303 variables in the configuration file should not matter, but it did
305 (merge 98b406f jk/remote-pushremote-config-reading later to maint).
307 * Codepaths that parse timestamps in commit objects have been
309 (merge f80d1f9 jk/commit-dates-parsing-fix later to maint).
311 * "git diff --external-diff" incorrectly fed the submodule directory
312 in the working tree to the external diff driver when it knew it is
313 the same as one of the versions being compared.
314 (merge aba4727 tr/diff-submodule-no-reuse-worktree later to maint).
316 * "git reset" needs to refresh the index when working in a working
317 tree (it can also be used to match the index to the HEAD in an
318 otherwise bare repository), but it failed to set up the working
319 tree properly, causing GIT_WORK_TREE to be ignored.
320 (merge b7756d4 nd/reset-setup-worktree later to maint).
322 * "git check-attr" when working on a repository with a working tree
323 did not work well when the working tree was specified via the
324 --work-tree (and obviously with --git-dir) option.
325 (merge cdbf623 jc/check-attr-honor-working-tree later to maint).
327 * "merge-recursive" was broken in 1.7.7 era and stopped working in
328 an empty (temporary) working tree, when there are renames
329 involved. This has been corrected.
330 (merge 6e2068a bk/refresh-missing-ok-in-merge-recursive later to maint.)
332 * "git rev-parse" was loose in rejecting command line arguments
333 that do not make sense, e.g. "--default" without the required
334 value for that option.
335 (merge a43219f ds/rev-parse-required-args later to maint.)
337 * include.path variable (or any variable that expects a path that
338 can use ~username expansion) in the configuration file is not a
339 boolean, but the code failed to check it.
340 (merge 67beb60 jk/config-path-include-fix later to maint.)
342 * Commands that take pathspecs on the command line misbehaved when
343 the pathspec is given as an absolute pathname (which is a
344 practice not particularly encouraged) that points at a symbolic
345 link in the working tree.
346 (merge later 655ee9e mw/symlinks to maint.)
348 * "git diff --quiet -- pathspec1 pathspec2" sometimes did not return
349 correct status value.
350 (merge f34b205 nd/diff-quiet-stat-dirty later to maint.)
352 * Attempting to deepen a shallow repository by fetching over smart
353 HTTP transport failed in the protocol exchange, when no-done
354 extension was used. The fetching side waited for the list of
355 shallow boundary commits after the sending end stopped talking to
357 (merge 0232852 nd/http-fetch-shallow-fix later to maint.)
359 * Allow "git cmd path/", when the 'path' is where a submodule is
360 bound to the top-level working tree, to match 'path', despite the
361 extra and unnecessary trailing slash (such a slash is often
362 given by command line completion).
363 (merge 2e70c01 nd/submodule-pathspec-ending-with-slash later to maint.)
365 * Documentation and in-code comments had many instances of mistaken
366 use of "nor", which have been corrected.
367 (merge 235e8d5 jl/nor-or-nand-and later to maint).