1Git v2.19.2 Release Notes 2========================= 3 4Fixes since v2.19.1 5------------------- 6 7 * "git interpret-trailers" and its underlying machinery had a buggy 8 code that attempted to ignore patch text after commit log message, 9 which triggered in various codepaths that will always get the log 10 message alone and never get such an input. 11 12 * "git rebase -i" did not clear the state files correctly when a run 13 of "squash/fixup" is aborted and then the user manually amended the 14 commit instead, which has been corrected. 15 16 * When fsmonitor is in use, after operation on submodules updates 17 .gitmodules, we lost track of the fact that we did so and relied on 18 stale fsmonitor data. 19 20 * Fix for a long-standing bug that leaves the index file corrupt when 21 it shrinks during a partial commit. 22 23 * Further fix for O_APPEND emulation on Windows 24 25 * A corner case bugfix in "git rerere" code. 26 27 * "git add ':(attr:foo)'" is not supported and is supposed to be 28 rejected while the command line arguments are parsed, but we fail 29 to reject such a command line upfront. 30 31 * "git rebase" etc. in Git 2.19 fails to abort when given an empty 32 commit log message as result of editing, which has been corrected. 33 34 * The code to backfill objects in lazily cloned repository did not 35 work correctly, which has been corrected. 36 37 * Update error messages given by "git remote" and make them consistent. 38 39 * "git update-ref" learned to make both "--no-deref" and "--stdin" 40 work at the same time. 41 42 * Recently added "range-diff" had a corner-case bug to cause it 43 segfault, which has been corrected. 44 45 * The recently introduced commit-graph auxiliary data is incompatible 46 with mechanisms such as replace & grafts that "breaks" immutable 47 nature of the object reference relationship. Disable optimizations 48 based on its use (and updating existing commit-graph) when these 49 incompatible features are in use in the repository. 50 51 * The mailmap file update. 52 53 * The code in "git status" sometimes hit an assertion failure. This 54 was caused by a structure that was reused without cleaning the data 55 used for the first run, which has been corrected. 56 57 * A corner-case bugfix. 58 59 * A partial clone that is configured to lazily fetch missing objects 60 will on-demand issue a "git fetch" request to the originating 61 repository to fill not-yet-obtained objects. The request has been 62 optimized for requesting a tree object (and not the leaf blob 63 objects contained in it) by telling the originating repository that 64 no blobs are needed. 65 66 * The codepath to support the experimental split-index mode had 67 remaining "racily clean" issues fixed. 68 69 * "git log --graph" showing an octopus merge sometimes miscounted the 70 number of display columns it is consuming to show the merge and its 71 parent commits, which has been corrected. 72 73 * The implementation of run_command() API on the UNIX platforms had a 74 bug that caused a command not on $PATH to be found in the current 75 directory. 76 77 * A mutex used in "git pack-objects" were not correctly initialized 78 and this caused "git repack" to dump core on Windows. 79 80 * Under certain circumstances, "git diff D:/a/b/c D:/a/b/d" on 81 Windows would strip initial parts from the paths because they 82 were not recognized as absolute, which has been corrected. 83 84 * The receive.denyCurrentBranch=updateInstead codepath kicked in even 85 when the push should have been rejected due to other reasons, such 86 as it does not fast-forward or the update-hook rejects it, which 87 has been corrected. 88 89 * "git repack" in a shallow clone did not correctly update the 90 shallow points in the repository, leading to a repository that 91 does not pass fsck. 92 93 * Operations on promisor objects make sense in the context of only a 94 small subset of the commands that internally use the revisions 95 machinery, but the "--exclude-promisor-objects" option were taken 96 and led to nonsense results by commands like "log", to which it 97 didn't make much sense. This has been corrected. 98 99 * The "container" mode of TravisCI is going away. Our .travis.yml 100 file is getting prepared for the transition. 101 102 * Our test scripts can now take the '-V' option as a synonym for the 103 '--verbose-log' option. 104 105 * A regression in Git 2.12 era made "git fsck" fall into an infinite 106 loop while processing truncated loose objects. 107 108Also contains various documentation updates and code clean-ups.