9 * Building on older MacOS X systems automatically sets
10 the necessary NO_APPLE_COMMON_CRYPTO build-time option.
13 UI, Workflows & Features
15 * "git config --edit --global" starts from a skeletal per-user
16 configuration file contents, instead of a total blank, when the
17 user does not already have any. This immediately reduces the
18 need for a later "Have you forgotten setting core.user?" and we
19 can add more to the template as we gain more experience.
21 * "git stash list -p" used to be almost always a no-op because each
22 stash entry is represented as a merge commit. It learned to show
23 the difference between the base commit version and the working tree
24 version, which is in line with what "git show" gives.
26 * Sometimes users want to report a bug they experience on their
27 repository, but they are not at liberty to share the contents of
28 the repository. "fast-export" was taught an "--anonymize" option
29 to replace blob contents, names of people and paths and log
30 messages with bland and simple strings to help them.
32 * "log --date=iso" uses a slight variant of ISO 8601 format that is
33 made more human readable. A new "--date=iso-strict" option gives
34 datetime output that is more strictly conformant.
36 * A broken reimplementation of Git could write an invalid index that
37 records both stage #0 and higher stage entries for the same path.
38 We now notice and reject such an index, as there is no sensible
39 fallback (we do not know if the broken tool wanted to resolve and
40 forgot to remove higher stage entries, or if it wanted to unresolve
41 and forgot to remove the stage#0 entry).
43 * The "pre-receive" and "post-receive" hooks are no longer required
44 to consume their input fully (not following this requirement used
45 to result in intermittent errors in "git push").
47 * The pretty-format specifier "%d", which expanded to " (tagname)"
48 for a tagged commit, gained a cousin "%D" that just gives the
49 "tagname" without frills.
51 Performance, Internal Implementation, etc.
53 * The API to manipulate the "refs" is currently undergoing a revamp
54 to make it more transactional, with the eventual goal to allow
55 all-or-none atomic updates and migrating the storage to something
56 other than the traditional filesystem based one (e.g. databases).
58 * We no longer attempt to keep track of individual dependencies to
59 the header files in the build procedure, relying on automated
60 dependency generation support from modern compilers.
62 * In tests, we have been using NOT_{MINGW,CYGWIN} test prerequisites
63 long before negated prerequisites e.g. !MINGW were invented.
64 The former has been converted to the latter to avoid confusion.
66 * Looking up remotes configuration in a repository with very many
67 remotes defined has been optimized.
69 * There are cases where you lock and open to write a file, close it
70 to show the updated contents to external processes, and then have
71 to update the file again while still holding the lock, but the
72 lockfile API lacked support for such an access pattern.
74 * The API to allocate the structure to keep track of commit
75 decoration has been updated to make it less cumbersome to use.
77 * An in-core caching layer to let us avoid reading the same
78 configuration files number of times has been added. A few commands
79 have been converted to use this subsystem.
81 * Various code paths have been cleaned up and simplified by using
82 "strbuf", "starts_with()", and "skip_prefix()" APIs more.
84 * A few codepaths that died when large blobs that would not fit in
85 core are involved in their operation have been taught to punt
86 instead, by e.g. marking too large a blob as not to be diffed.
88 * A few more code paths in "commit" and "checkout" have been taught
89 to repopulate the cache-tree in the index, to help speed up later
90 "write-tree" (used in "commit") and "diff-index --cached" (used in
93 * A common programming mistake to assign the same short option name
94 to two separate options is detected by parse_options() API to help
97 * The code path to write out the packed-refs file has been optimized,
98 which especially matters in a repository with a large number of
101 * The check to see if a ref $F can be created by making sure no
102 existing ref has $F/ as its prefix has been optimized, which
103 especially matters in a repository with a large number of existing
106 * "git fsck" was taught to check contents of tag objects a bit more.
108 * "git hash-object" was taught a "--literally" option to help
112 Also contains various documentation updates and code clean-ups.
118 Unless otherwise noted, all the fixes since v2.1 in the maintenance
119 track are contained in this release (see the maintenance releases'
122 * "git log --pretty/format=" with an empty format string did not
123 mean the more obvious "No output whatsoever" but "Use default
124 format", which was counterintuitive.
126 * Implementations of "tar" that do not understand an extended pax
127 header would extract the contents of it in a regular file; make
128 sure the permission bits of this file follows the same tar.umask
129 configuration setting.
131 * "git -c section.var command" and "git -c section.var= command"
132 should pass the configuration differently (the former should be a
133 boolean true, the latter should be an empty string).
135 * Applying a patch not generated by Git in a subdirectory used to
136 check the whitespace breakage using the attributes for incorrect
137 paths. Also whitespace checks were performed even for paths
138 excluded via "git apply --exclude=<path>" mechanism.
140 * "git bundle create" with date-range specification were meant to
141 exclude tags outside the range, but it didn't.
143 * "git add x" where x that used to be a directory has become a
144 symbolic link to a directory misbehaved.
146 * The prompt script checked $GIT_DIR/ref/stash file to see if there
147 is a stash, which was a no-no.
149 * Pack-protocol documentation had a minor typo.
151 * "git checkout -m" did not switch to another branch while carrying
152 the local changes forward when a path was deleted from the index.
154 * With sufficiently long refnames, "git fast-import" could have
155 overflown an on-stack buffer.
157 * After "pack-refs --prune" packed refs at the top-level, it failed
160 * Progress output from "git gc --auto" was visible in "git fetch -q".
162 * We used to pass -1000 to poll(2), expecting it to also mean "no
163 timeout", which should be spelled as -1.
165 * "git rebase" documentation was unclear that it is required to
166 specify on what <upstream> the rebase is to be done when telling it
167 to first check out <branch>.
168 (merge 95c6826 so/rebase-doc later to maint).
170 * "git push" over HTTP transport had an artificial limit on number of
171 refs that can be pushed imposed by the command line length.
172 (merge 26be19b jk/send-pack-many-refspecs later to maint).
174 * When receiving an invalid pack stream that records the same object
175 twice, multiple threads got confused due to a race.
176 (merge ab791dd jk/index-pack-threading-races later to maint).
178 * An attempt to remove the entire tree in the "git fast-import" input
179 stream caused it to misbehave.
180 (merge 2668d69 mb/fast-import-delete-root later to maint).
182 * Reachability check (used in "git prune" and friends) did not add a
183 detached HEAD as a starting point to traverse objects still in use.
184 (merge c40fdd0 mk/reachable-protect-detached-head later to maint).
186 * "git config --add section.var val" used to lose existing
187 section.var whose value was an empty string.
188 (merge c1063be ta/config-add-to-empty-or-true-fix later to maint).
190 * "git fsck" failed to report that it found corrupt objects via its
191 exit status in some cases.
192 (merge 30d1038 jk/fsck-exit-code-fix later to maint).
194 * Use of "--verbose" option used to break "git branch --merged".
195 (merge 12994dd jk/maint-branch-verbose-merged later to maint).
197 * Some MUAs mangled a line in a message that begins with "From " to
198 ">From " when writing to a mailbox file and feeding such an input
199 to "git am" used to lose such a line.
200 (merge 85de86a jk/mbox-from-line later to maint).
202 * "rev-parse --verify --quiet $name" is meant to quietly exit with a
203 non-zero status when $name is not a valid object name, but still
204 gave error messages in some cases.