1Git 2.22.1 Release Notes 2======================== 3 4Fixes since v2.22 5----------------- 6 7 * A relative pathname given to "git init --template=<path> <repo>" 8 ought to be relative to the directory "git init" gets invoked in, 9 but it instead was made relative to the repository, which has been 10 corrected. 11 12 * "git worktree add" used to fail when another worktree connected to 13 the same repository was corrupt, which has been corrected. 14 15 * The ownership rule for the file descriptor to fast-import remote 16 backend was mixed up, leading to unrelated file descriptor getting 17 closed, which has been fixed. 18 19 * "git update-server-info" used to leave stale packfiles in its 20 output, which has been corrected. 21 22 * The server side support for "git fetch" used to show incorrect 23 value for the HEAD symbolic ref when the namespace feature is in 24 use, which has been corrected. 25 26 * "git am -i --resolved" segfaulted after trying to see a commit as 27 if it were a tree, which has been corrected. 28 29 * "git bundle verify" needs to see if prerequisite objects exist in 30 the receiving repository, but the command did not check if we are 31 in a repository upfront, which has been corrected. 32 33 * "git merge --squash" is designed to update the working tree and the 34 index without creating the commit, and this cannot be countermanded 35 by adding the "--commit" option; the command now refuses to work 36 when both options are given. 37 38 * The data collected by fsmonitor was not properly written back to 39 the on-disk index file, breaking t7519 tests occasionally, which 40 has been corrected. 41 42 * Update to Unicode 12.1 width table. 43 44 * The command line to invoke a "git cat-file" command from inside 45 "git p4" was not properly quoted to protect a caret and running a 46 broken command on Windows, which has been corrected. 47 48 * "git request-pull" learned to warn when the ref we ask them to pull 49 from in the local repository and in the published repository are 50 different. 51 52 * When creating a partial clone, the object filtering criteria is 53 recorded for the origin of the clone, but this incorrectly used a 54 hardcoded name "origin" to name that remote; it has been corrected 55 to honor the "--origin <name>" option. 56 57 * "git fetch" into a lazy clone forgot to fetch base objects that are 58 necessary to complete delta in a thin packfile, which has been 59 corrected. 60 61 * The filter_data used in the list-objects-filter (which manages a 62 lazily sparse clone repository) did not use the dynamic array API 63 correctly---'nr' is supposed to point at one past the last element 64 of the array in use. This has been corrected. 65 66 * The description about slashes in gitignore patterns (used to 67 indicate things like "anchored to this level only" and "only 68 matches directories") has been revamped. 69 70 * The URL decoding code has been updated to avoid going past the end 71 of the string while parsing %-<hex>-<hex> sequence. 72 73 * The list of for-each like macros used by clang-format has been 74 updated. 75 76 * "git push --atomic" that goes over the transport-helper (namely, 77 the smart http transport) failed to prevent refs to be pushed when 78 it can locally tell that one of the ref update will fail without 79 having to consult the other end, which has been corrected. 80 81 * "git clean" silently skipped a path when it cannot lstat() it; now 82 it gives a warning. 83 84 * A codepath that reads from GPG for signed object verification read 85 past the end of allocated buffer, which has been fixed. 86 87 * "git rm" to resolve a conflicted path leaked an internal message 88 "needs merge" before actually removing the path, which was 89 confusing. This has been corrected. 90 91 * The "git clone" documentation refers to command line options in its 92 description in the short form; they have been replaced with long 93 forms to make them more recognisable. 94 95 * The configuration variable rebase.rescheduleFailedExec should be 96 effective only while running an interactive rebase and should not 97 affect anything when running a non-interactive one, which was not 98 the case. This has been corrected. 99 100 * "git submodule foreach" did not protect command line options passed 101 to the command to be run in each submodule correctly, when the 102 "--recursive" option was in use. 103 104 * Use "Erase in Line" CSI sequence that is already used in the editor 105 support to clear cruft in the progress output. 106 107 * The codepath to compute delta islands used to spew progress output 108 without giving the callers any way to squelch it, which has been 109 fixed. 110 111 * The code to parse scaled numbers out of configuration files has 112 been made more robust and also easier to follow. 113 114 * An incorrect list of options was cached after command line 115 completion failed (e.g. trying to complete a command that requires 116 a repository outside one), which has been corrected. 117 118 * "git rebase --abort" used to leave refs/rewritten/ when concluding 119 "git rebase -r", which has been corrected. 120 121 * "git stash show 23" used to work, but no more after getting 122 rewritten in C; this regression has been corrected. 123 124 * "git interpret-trailers" always treated '#' as the comment 125 character, regardless of core.commentChar setting, which has been 126 corrected. 127 128 * Code clean-up to avoid signed integer overlaps during binary search. 129 130 * "git checkout -p" needs to selectively apply a patch in reverse, 131 which did not work well. 132 133 * The commit-graph file is now part of the "files that the runtime 134 may keep open file descriptors on, all of which would need to be 135 closed when done with the object store", and the file descriptor to 136 an existing commit-graph file now is closed before "gc" finalizes a 137 new instance to replace it. 138 139 * Code restructuring during 2.20 period broke fetching tags via 140 "import" based transports. 141 142 * We have been trying out a few language features outside c89; the 143 coding guidelines document did not talk about them and instead had 144 a blanket ban against them. 145 146 * The internal diff machinery can be made to read out of bounds while 147 looking for --funcion-context line in a corner case, which has been 148 corrected. 149 150Also contains various documentation updates, code clean-ups and minor fixups.