| 1 | Git v2.6.3 Release Notes |
| 2 | ======================== |
| 3 | |
| 4 | Fixes since v2.6.2 |
| 5 | ------------------ |
| 6 | |
| 7 | * The error message from "git blame --contents --reverse" incorrectly |
| 8 | talked about "--contents --children". |
| 9 | |
| 10 | * "git merge-file" tried to signal how many conflicts it found, which |
| 11 | obviously would not work well when there are too many of them. |
| 12 | |
| 13 | * The name-hash subsystem that is used to cope with case insensitive |
| 14 | filesystems keeps track of directories and their on-filesystem |
| 15 | cases for all the paths in the index by holding a pointer to a |
| 16 | randomly chosen cache entry that is inside the directory (for its |
| 17 | ce->ce_name component). This pointer was not updated even when the |
| 18 | cache entry was removed from the index, leading to use after free. |
| 19 | This was fixed by recording the path for each directory instead of |
| 20 | borrowing cache entries and restructuring the API somewhat. |
| 21 | |
| 22 | * When the "git am" command was reimplemented in C, "git am -3" had a |
| 23 | small regression where it is aborted in its error handling codepath |
| 24 | when underlying merge-recursive failed in some ways. |
| 25 | |
| 26 | * The synopsis text and the usage string of subcommands that read |
| 27 | list of things from the standard input are often shown as if they |
| 28 | only take input from a file on a filesystem, which was misleading. |
| 29 | |
| 30 | * A couple of commands still showed "[options]" in their usage string |
| 31 | to note where options should come on their command line, but we |
| 32 | spell that "[<options>]" in most places these days. |
| 33 | |
| 34 | * The submodule code has been taught to work better with separate |
| 35 | work trees created via "git worktree add". |
| 36 | |
| 37 | * When "git gc --auto" is backgrounded, its diagnosis message is |
| 38 | lost. It now is saved to a file in $GIT_DIR and is shown next time |
| 39 | the "gc --auto" is run. |
| 40 | |
| 41 | * Work around "git p4" failing when the P4 depot records the contents |
| 42 | in UTF-16 without UTF-16 BOM. |
| 43 | |
| 44 | * Recent update to "rebase -i" that tries to sanity check the edited |
| 45 | insn sheet before it uses it has become too picky on Windows where |
| 46 | CRLF left by the editor is turned into a trailing CR on the line |
| 47 | read via the "read" built-in command. |
| 48 | |
| 49 | * "git clone --dissociate" runs a big "git repack" process at the |
| 50 | end, and it helps to close file descriptors that are open on the |
| 51 | packs and their idx files before doing so on filesystems that |
| 52 | cannot remove a file that is still open. |
| 53 | |
| 54 | * Correct "git p4 --detect-labels" so that it does not fail to create |
| 55 | a tag that points at a commit that is also being imported. |
| 56 | |
| 57 | * The internal stripspace() function has been moved to where it |
| 58 | logically belongs to, i.e. strbuf API, and the command line parser |
| 59 | of "git stripspace" has been updated to use the parse_options API. |
| 60 | |
| 61 | * Prepare for Git on-disk repository representation to undergo |
| 62 | backward incompatible changes by introducing a new repository |
| 63 | format version "1", with an extension mechanism. |
| 64 | |
| 65 | * "git gc" used to barf when a symbolic ref has gone dangling |
| 66 | (e.g. the branch that used to be your upstream's default when you |
| 67 | cloned from it is now gone, and you did "fetch --prune"). |
| 68 | |
| 69 | * The normalize_ceiling_entry() function does not muck with the end |
| 70 | of the path it accepts, and the real world callers do rely on that, |
| 71 | but a test insisted that the function drops a trailing slash. |
| 72 | |
| 73 | * "git gc" is safe to run anytime only because it has the built-in |
| 74 | grace period to protect young objects. In order to run with no |
| 75 | grace period, the user must make sure that the repository is |
| 76 | quiescent. |
| 77 | |
| 78 | * A recent "filter-branch --msg-filter" broke skipping of the commit |
| 79 | object header, which is fixed. |
| 80 | |
| 81 | * "git --literal-pathspecs add -u/-A" without any command line |
| 82 | argument misbehaved ever since Git 2.0. |
| 83 | |
| 84 | * Merging a branch that removes a path and another that changes the |
| 85 | mode bits on the same path should have conflicted at the path, but |
| 86 | it didn't and silently favoured the removal. |
| 87 | |
| 88 | * "git imap-send" did not compile well with older version of cURL library. |
| 89 | |
| 90 | * The linkage order of libraries was wrong in places around libcurl. |
| 91 | |
| 92 | * It was not possible to use a repository-lookalike created by "git |
| 93 | worktree add" as a local source of "git clone". |
| 94 | |
| 95 | * When "git send-email" wanted to talk over Net::SMTP::SSL, |
| 96 | Net::Cmd::datasend() did not like to be fed too many bytes at the |
| 97 | same time and failed to send messages. Send the payload one line |
| 98 | at a time to work around the problem. |
| 99 | |
| 100 | * We peek objects from submodule's object store by linking it to the |
| 101 | list of alternate object databases, but the code to do so forgot to |
| 102 | correctly initialize the list. |
| 103 | |
| 104 | * "git status --branch --short" accessed beyond the constant string |
| 105 | "HEAD", which has been corrected. |
| 106 | |
| 107 | * "git daemon" uses "run_command()" without "finish_command()", so it |
| 108 | needs to release resources itself, which it forgot to do. |
| 109 | |
| 110 | Also contains typofixes, documentation updates and trivial code |
| 111 | clean-ups. |