1 Git v2.13.1 Release Notes
2 =========================
7 * The Web interface to gmane news archive is long gone, even though
8 the articles are still accessible via NTTP. Replace the links with
9 ones to public-inbox.org. Because their message identification is
10 based on the actual message-id, it is likely that it will be easier
11 to migrate away from it if/when necessary.
13 * Update tests to pass under GETTEXT_POISON (a mechanism to ensure
14 that output strings that should not be translated are not
15 translated by mistake), and tell TravisCI to run them.
17 * Setting "log.decorate=false" in the configuration file did not take
18 effect in v2.13, which has been corrected.
20 * An earlier update to test 7400 needed to be skipped on CYGWIN.
22 * Git sometimes gives an advice in a rhetorical question that does
23 not require an answer, which can confuse new users and non native
24 speakers. Attempt to rephrase them.
26 * "git read-tree -m" (no tree-ish) gave a nonsense suggestion "use
27 --empty if you want to clear the index". With "-m", such a request
28 will still fail anyway, as you'd need to name at least one tree-ish
31 * The codepath in "git am" that is used when running "git rebase"
32 leaked memory held for the log message of the commits being rebased.
34 * "pack-objects" can stream a slice of an existing packfile out when
35 the pack bitmap can tell that the reachable objects are all needed
36 in the output, without inspecting individual objects. This
37 strategy however would not work well when "--local" and other
38 options are in use, and need to be disabled.
40 * Clarify documentation for include.path and includeIf.<condition>.path
41 configuration variables.
43 * Tag objects, which are not reachable from any ref, that point at
44 missing objects were mishandled by "git gc" and friends (they
45 should silently be ignored instead)
47 * A few http:// links that are redirected to https:// in the
48 documentation have been updated to https:// links.
50 * Make sure our tests would pass when the sources are checked out
51 with "platform native" line ending convention by default on
52 Windows. Some "text" files out tests use and the test scripts
53 themselves that are meant to be run with /bin/sh, ought to be
54 checked out with eol=LF even on Windows.
56 * Fix memory leaks pointed out by Coverity (and people).
58 * The receive-pack program now makes sure that the push certificate
59 records the same set of push options used for pushing.
61 * "git cherry-pick" and other uses of the sequencer machinery
62 mishandled a trailer block whose last line is an incomplete line.
63 This has been fixed so that an additional sign-off etc. are added
64 after completing the existing incomplete line.
66 * The shell completion script (in contrib/) learned "git stash" has
67 a new "push" subcommand.
69 * Travis CI gained a task to format the documentation with both
70 AsciiDoc and AsciiDoctor.
72 * Update the C style recommendation for notes for translators, as
73 recent versions of gettext tools can work with our style of
76 * "git clone --config var=val" is a way to populate the
77 per-repository configuration file of the new repository, but it did
78 not work well when val is an empty string. This has been fixed.
80 * A few codepaths in "checkout" and "am" working on an unborn branch
81 tried to access an uninitialized piece of memory.
83 * "git for-each-ref --format=..." with %(HEAD) in the format used to
84 resolve the HEAD symref as many times as it had processed refs,
85 which was wasteful, and "git branch" shared the same problem.
87 * "git interpret-trailers", when used as GIT_EDITOR for "git commit
88 -v", looked for and appended to a trailer block at the very end,
89 i.e. at the end of the "diff" output. The command has been
90 corrected to pay attention to the cut-mark line "commit -v" adds to
91 the buffer---the real trailer block should appear just before it.
93 * A test allowed both "git push" and "git receive-pack" on the other
94 end write their traces into the same file. This is OK on platforms
95 that allows atomically appending to a file opened with O_APPEND,
96 but on other platforms led to a mangled output, causing
97 intermittent test failures. This has been fixed by disabling
98 traces from "receive-pack" in the test.
100 * "foo\bar\baz" in "git fetch foo\bar\baz", even though there is no
101 slashes in it, cannot be a nickname for a remote on Windows, as
102 that is likely to be a pathname on a local filesystem.
104 * The "collision detecting" SHA-1 implementation shipped with 2.13
105 was quite broken on some big-endian platforms and/or platforms that
106 do not like unaligned fetches. Update to the upstream code which
107 has already fixed these issues.
109 * "git am -h" triggered a BUG().
111 * The interaction of "url.*.insteadOf" and custom URL scheme's
112 whitelisting is now documented better.
114 Also contains various documentation updates and code clean-ups.