| blob | 72db8d211fbf94ae94f8f372237b4b1de153b932 |
1 Git v2.3 Release Notes
2 ======================
4 Updates since v2.2
5 ------------------
7 Ports
9 * Recent gcc toolchain on Cygwin started throwing compilation warning,
10 which has been squelched.
13 UI, Workflows & Features
15 * It was cumbersome to use "GIT_SSH" mechanism when the user wanted
16 to pass an extra set of arguments to the underlying ssh. A new
17 environment variable GIT_SSH_COMMAND can be used for this.
19 * A request to store an empty note via "git notes" meant to remove
20 note from the object but with --allow-empty we will store a
21 (surprise!) note that is empty.
23 * "git interpret-trailers" learned to properly handle the
24 "Conflicts:" block at the end.
26 * "git am" learned "--message-id" option to copy the message ID of
27 the incoming e-mail to the log message of resulting commit.
29 * "git clone --reference=<over there>" learned the "--dissociate"
30 option to go with it; it borrows objects from the reference object
31 store while cloning only to reduce network traffic and then
32 dissociates the resulting clone from the reference by performing
33 local copies of borrowed objects.
35 * "git send-email" learned "--transfer-encoding" option to force a
36 non-fault Content-Transfer-Encoding header (e.g. base64).
38 * "git send-email" normally identifies itself via X-Mailer: header in
39 the message it sends out. A new command line flag --no-xmailer
40 allows the user to squelch the header.
42 * "git push" into a repository with a working tree normally refuses
43 to modify the branch that is checked out. The command learned to
44 optionally do an equivalent of "git reset --hard" only when there
45 is no change to the working tree and the index instead, which would
46 be useful to "deploy" by pushing into a repository.
48 * "git new-workdir" (in contrib/) can be used to populate an empty
49 and existing directory now.
51 * Credential helpers are asked in turn until one of them give
52 positive response, which is cumbersome to turn off when you need to
53 run Git in an automated setting. The credential helper interface
54 learned to allow a helper to say "stop, don't ask other helpers."
55 Also GIT_TERMINAL_PROMPT environment can be set to false to disable
56 our built-in prompt mechanism for passwords.
58 * "git branch -d" (delete) and "git branch -m" (move) learned to
59 honor "-f" (force) flag; unlike many other subcommands, the way to
60 force these have been with separate "-D/-M" options, which was
61 inconsistent.
63 * "diff-highlight" filter (in contrib/) allows its color output to be
64 customized via configuration variables.
66 * "git imap-send" learned to take "-v" (verbose) and "-q" (quiet)
67 command line options.
69 * "git imap-send" now can be built to use cURL library to talk to
70 IMAP servers (if the library is recent enough, of course).
71 This allows you to use authenticate method other than CRAM-MD5,
72 among other things.
75 Performance, Internal Implementation, Development Support etc.
77 * Earlier we made "rev-list --object-edge" more aggressively list the
78 objects at the edge commits, in order to reduce number of objects　
79 fetched into a shallow repository, but the change affected cases
80 other than "fetching into a shallow repository" and made it
81 unusably slow (e.g. fetching into a normal repository should not
82 have to suffer the overhead from extra processing). Limit it to a
83 more specific case by introducing --objects-edge-aggressive, a new
84 option to rev-list.
86 * Squelched useless compiler warnings on Mac OS X regarding the
87 crypto API.
89 * The procedure to generate unicode table has been simplified.
91 * Some filesystems assign filemodes in a strange way, fooling then
92 automatic "filemode trustability" check done during a new
93 repository creation. The initialization codepath has been hardened
94 against this issue.
96 * The codepath in "git remote update --prune" to drop many refs has
97 been optimized.
99 * The API into get_merge_bases*() family of functions was easy to
100 misuse, which has been corrected to make it harder to do so.
102 * Long overdue departure from the assumption that S_IFMT is shared by
103 everybody made in 2005, which was necessary to port to z/OS.
105 * "git push" and "git fetch" did not communicate an overlong refname
106 correctly. Now it uses 64kB sideband to accommodate longer ones.
108 * Recent GPG changes the keyring format and drops support for RFC1991
109 formatted signatures, breaking our existing tests.
111 * "git-prompt" (in contrib/) used a variable from the global scope,
112 possibly contaminating end-user's namespace.
115 Also contains various documentation updates and code clean-ups.
118 Fixes since v2.2
119 ----------------
121 Unless otherwise noted, all the fixes since v2.2 in the maintenance
122 track are contained in this release (see the maintenance releases'
123 notes for details).
125 * The logic in "git bisect bad HEAD" etc. to avoid forcing the test
126 of the common ancestor of bad and good commits was broken.
127 (merge 07913d5 cc/bisect-rev-parsing later to maint).
129 * "git checkout-index --temp=$target $path" did not work correctly
130 for paths outside the current subdirectory in the project.
131 (merge 74c4de5 es/checkout-index-temp later to maint).
133 * The report from "git checkout" on a branch that builds on another
134 local branch by setting its branch.*.merge to branch name (not a
135 full refname) incorrectly said that the upstream is gone.
136 (merge 05e7368 jc/checkout-local-track-report later to maint).
138 * With The git-prompt support (in contrib/), using the exit status of
139 the last command in the prompt, e.g. PS1='$(__git_ps1) $? ', did
140 not work well, because the helper function stomped on the exit
141 status.
142 (merge eb443e3 tf/prompt-preserve-exit-status later to maint).
145 * Recent update to "git commit" broke amending an existing commit
146 with bogus author/committer lines without a valid e-mail address.
147 (merge c83a509 jk/commit-date-approxidate later to maint).
149 * The lockfile API used to get confused which file to clean up when
150 the process moved the $cwd after creating a lockfile.
151 (merge fa137f6 nd/lockfile-absolute later to maint).
153 * Traditionally we tried to avoid interpreting date strings given by
154 the user as future dates, e.g. GIT_COMMITTER_DATE=2014-12-10 when
155 used early November 2014 was taken as "October 12, 2014" because it
156 is likely that a date in the future, December 10, is a mistake.
157 This heuristics has been loosened to allow people to express future
158 dates (most notably, --until=<date> may want to be far in the
159 future) and we no longer tiebreak by future-ness of the date when
161 (1) ISO-like format is used, and
162 (2) the string can make sense interpreted as both y-m-d and y-d-m.
164 Git may still have to use the heuristics to tiebreak between dd/mm/yy
165 and mm/dd/yy, though.
166 (merge d372395 jk/approxidate-avoid-y-d-m-over-future-dates later to maint).
168 * Git did not correctly read an overlong refname from a packed refs
169 file.
170 (merge ea41783 jk/read-packed-refs-without-path-max later to maint).
172 * "git apply" was described in the documentation to take --ignore-date
173 option, which it does not.
174 (merge 0cef4e7 rw/apply-does-not-take-ignore-date later to maint).
176 * "git add -i" did not notice when the interactive command input
177 stream went away and kept asking the same question.
178 (merge a8bec7a jk/add-i-read-error later to maint).
180 * "git send-email" did not handle RFC 2047 encoded headers quite
181 right.
182 (merge ab47e2a rd/send-email-2047-fix later to maint).
184 * New tag object format validation added in 2.2 showed garbage after
185 a tagname it reported in its error message.
186 (merge a1e920a js/fsck-tag-validation later to maint).
188 * The code that reads the reflog from the newer to the older entries
189 did not handle an entry that crosses a boundary of block it uses to
190 read them correctly.
191 (merge 69216bf jk/for-each-reflog-ent-reverse later to maint).
193 * "git diff -B -M" after making a new copy B out of an existing file
194 A and then editing A extensively ought to report that B was created
195 by copying A and A was modified, which is what "git diff -C"
196 reports, but it instead said A was renamed to B and A was edited
197 heavily in place. This was not just incoherent but also failed to
198 apply with "git apply". The report has been corrected to match what
199 "git diff -C" produces for this case.
200 (merge 6936b58 jc/diff-b-m later to maint).
202 * In files we pre-populate for the user to edit with commented hints,
203 a line of hint that is indented with a tab used to show as '#' (or
204 any comment char), ' ' (space), and then the hint text that began
205 with the tab, which some editors flag as an indentation error (tab
206 following space). We now omit the space after the comment char in
207 such a case.
208 (merge d55aeb7 jc/strbuf-add-lines-avoid-sp-ht-sequence later to maint).
210 * "git ls-tree" does not support path selection based on negative
211 pathspecs, but did not error out when negative pathspecs are given.
212 (merge f1f6224 nd/ls-tree-pathspec later to maint).
214 * The function sometimes returned a non-freeable memory and some
215 other times returned a piece of memory that must be freed, leading
216 to inevitable leaks.
217 (merge 59362e5 jc/exec-cmd-system-path-leak-fix later to maint).
219 * The code to abbreviate an object name to its short unique prefix
220 has been optimized when no abbreviation was requested.
221 (merge 61e704e mh/find-uniq-abbrev later to maint).
223 * "git add --ignore-errors ..." did not ignore an error to
224 give a file that did not exist.
225 (merge 1d31e5a mg/add-ignore-errors later to maint).
227 * "git checkout $treeish $path", when $path in the index and the
228 working tree already matched what is in $treeish at the $path,
229 still overwrote the $path unnecessarily.
230 (merge c5326bd jk/checkout-from-tree later to maint).
232 * "git config --get-color" did not parse its command line arguments
233 carefully.
234 (merge cb35722 jk/colors-fix later to maint).
236 * open() emulated on Windows platforms did not give EISDIR upon
237 an attempt to open a directory for writing.
238 (merge ba6fad0 js/windows-open-eisdir-error later to maint).
240 * A few code paths used abs() when they should have used labs() on
241 long integers.
242 (merge 83915ba rs/maint-config-use-labs later to maint).
243 (merge 31a8aa1 rs/receive-pack-use-labs later to maint).
245 * "gitweb" used to depend on a behaviour recent CGI.pm deprecated.
246 (merge 13dbf46 jk/gitweb-with-newer-cgi-multi-param later to maint).
248 * "git init" (hence "git clone") initialized the per-repository
249 configuration file .git/config with x-bit by mistake.
250 (merge 1f32ecf mh/config-flip-xbit-back-after-checking later to maint).
252 * Git 2.0 was supposed to make the "simple" mode for the default of
253 "git push", but it didn't.
254 (merge 00a6fa0 jk/push-simple later to maint).
256 * "Everyday" document had a broken link.
257 (merge 366c8d4 po/everyday-doc later to maint).
259 * A few test fixes.
260 (merge 880ef58 jk/no-perl-tests later to maint).
262 * The build procedure did not bother fixing perl and python scripts
263 when NO_PERL and NO_PYTHON build-time configuration changed.
264 (merge ca2051d jk/rebuild-perl-scripts-with-no-perl-seting-change later to maint).