| 1 | Git v2.11.1 Release Notes |
| 2 | ========================= |
| 3 | |
| 4 | Fixes since v2.11 |
| 5 | ----------------- |
| 6 | |
| 7 | * The default Travis-CI configuration specifies newer P4 and GitLFS. |
| 8 | |
| 9 | * The character width table has been updated to match Unicode 9.0 |
| 10 | |
| 11 | * Update the isatty() emulation for Windows by updating the previous |
| 12 | hack that depended on internals of (older) MSVC runtime. |
| 13 | |
| 14 | * "git rev-parse --symbolic" failed with a more recent notation like |
| 15 | "HEAD^-1" and "HEAD^!". |
| 16 | |
| 17 | * An empty directory in a working tree that can simply be nuked used |
| 18 | to interfere while merging or cherry-picking a change to create a |
| 19 | submodule directory there, which has been fixed.. |
| 20 | |
| 21 | * The code in "git push" to compute if any commit being pushed in the |
| 22 | superproject binds a commit in a submodule that hasn't been pushed |
| 23 | out was overly inefficient, making it unusable even for a small |
| 24 | project that does not have any submodule but have a reasonable |
| 25 | number of refs. |
| 26 | |
| 27 | * "git push --dry-run --recurse-submodule=on-demand" wasn't |
| 28 | "--dry-run" in the submodules. |
| 29 | |
| 30 | * The output from "git worktree list" was made in readdir() order, |
| 31 | and was unstable. |
| 32 | |
| 33 | * mergetool.<tool>.trustExitCode configuration variable did not apply |
| 34 | to built-in tools, but now it does. |
| 35 | |
| 36 | * "git p4" LFS support was broken when LFS stores an empty blob. |
| 37 | |
| 38 | * Fix a corner case in merge-recursive regression that crept in |
| 39 | during 2.10 development cycle. |
| 40 | |
| 41 | * Update the error messages from the dumb-http client when it fails |
| 42 | to obtain loose objects; we used to give sensible error message |
| 43 | only upon 404 but we now forbid unexpected redirects that needs to |
| 44 | be reported with something sensible. |
| 45 | |
| 46 | * When diff.renames configuration is on (and with Git 2.9 and later, |
| 47 | it is enabled by default, which made it worse), "git stash" |
| 48 | misbehaved if a file is removed and another file with a very |
| 49 | similar content is added. |
| 50 | |
| 51 | * "git diff --no-index" did not take "--no-abbrev" option. |
| 52 | |
| 53 | * "git difftool --dir-diff" had a minor regression when started from |
| 54 | a subdirectory, which has been fixed. |
| 55 | |
| 56 | * "git commit --allow-empty --only" (no pathspec) with dirty index |
| 57 | ought to be an acceptable way to create a new commit that does not |
| 58 | change any paths, but it was forbidden, perhaps because nobody |
| 59 | needed it so far. |
| 60 | |
| 61 | * A pathname that begins with "//" or "\\" on Windows is special but |
| 62 | path normalization logic was unaware of it. |
| 63 | |
| 64 | * "git pull --rebase", when there is no new commits on our side since |
| 65 | we forked from the upstream, should be able to fast-forward without |
| 66 | invoking "git rebase", but it didn't. |
| 67 | |
| 68 | * The way to specify hotkeys to "xxdiff" that is used by "git |
| 69 | mergetool" has been modernized to match recent versions of xxdiff. |
| 70 | |
| 71 | * Unlike "git am --abort", "git cherry-pick --abort" moved HEAD back |
| 72 | to where cherry-pick started while picking multiple changes, when |
| 73 | the cherry-pick stopped to ask for help from the user, and the user |
| 74 | did "git reset --hard" to a different commit in order to re-attempt |
| 75 | the operation. |
| 76 | |
| 77 | * Code cleanup in shallow boundary computation. |
| 78 | |
| 79 | * A recent update to receive-pack to make it easier to drop garbage |
| 80 | objects made it clear that GIT_ALTERNATE_OBJECT_DIRECTORIES cannot |
| 81 | have a pathname with a colon in it (no surprise!), and this in turn |
| 82 | made it impossible to push into a repository at such a path. This |
| 83 | has been fixed by introducing a quoting mechanism used when |
| 84 | appending such a path to the colon-separated list. |
| 85 | |
| 86 | * The function usage_msg_opt() has been updated to say "fatal:" |
| 87 | before the custom message programs give, when they want to die |
| 88 | with a message about wrong command line options followed by the |
| 89 | standard usage string. |
| 90 | |
| 91 | * "git index-pack --stdin" needs an access to an existing repository, |
| 92 | but "git index-pack file.pack" to generate an .idx file that |
| 93 | corresponds to a packfile does not. |
| 94 | |
| 95 | * Fix for NDEBUG builds. |
| 96 | |
| 97 | * A lazy "git push" without refspec did not internally use a fully |
| 98 | specified refspec to perform 'current', 'simple', or 'upstream' |
| 99 | push, causing unnecessary "ambiguous ref" errors. |
| 100 | |
| 101 | * "git p4" misbehaved when swapping a directory and a symbolic link. |
| 102 | |
| 103 | * Even though an fix was attempted in Git 2.9.3 days, but running |
| 104 | "git difftool --dir-diff" from a subdirectory never worked. This |
| 105 | has been fixed. |
| 106 | |
| 107 | * "git p4" that tracks multiple p4 paths imported a single changelist |
| 108 | that touches files in these multiple paths as one commit, followed |
| 109 | by many empty commits. This has been fixed. |
| 110 | |
| 111 | * A potential but unlikely buffer overflow in Windows port has been |
| 112 | fixed. |
| 113 | |
| 114 | * When the http server gives an incomplete response to a smart-http |
| 115 | rpc call, it could lead to client waiting for a full response that |
| 116 | will never come. Teach the client side to notice this condition |
| 117 | and abort the transfer. |
| 118 | |
| 119 | * Some platforms no longer understand "latin-1" that is still seen in |
| 120 | the wild in e-mail headers; replace them with "iso-8859-1" that is |
| 121 | more widely known when conversion fails from/to it. |
| 122 | |
| 123 | * Update the procedure to generate "tags" for developer support. |
| 124 | |
| 125 | * Update the definition of the MacOSX test environment used by |
| 126 | TravisCI. |
| 127 | |
| 128 | * A few git-svn updates. |
| 129 | |
| 130 | * Compression setting for producing packfiles were spread across |
| 131 | three codepaths, one of which did not honor any configuration. |
| 132 | Unify these so that all of them honor core.compression and |
| 133 | pack.compression variables the same way. |
| 134 | |
| 135 | * "git fast-import" sometimes mishandled while rebalancing notes |
| 136 | tree, which has been fixed. |
| 137 | |
| 138 | * Recent update to the default abbreviation length that auto-scales |
| 139 | lacked documentation update, which has been corrected. |
| 140 | |
| 141 | * Leakage of lockfiles in the config subsystem has been fixed. |
| 142 | |
| 143 | * It is natural that "git gc --auto" may not attempt to pack |
| 144 | everything into a single pack, and there is no point in warning |
| 145 | when the user has configured the system to use the pack bitmap, |
| 146 | leading to disabling further "gc". |
| 147 | |
| 148 | * "git archive" did not read the standard configuration files, and |
| 149 | failed to notice a file that is marked as binary via the userdiff |
| 150 | driver configuration. |
| 151 | |
| 152 | * "git blame --porcelain" misidentified the "previous" <commit, path> |
| 153 | pair (aka "source") when contents came from two or more files. |
| 154 | |
| 155 | * "git rebase -i" with a recent update started showing an incorrect |
| 156 | count when squashing more than 10 commits. |
| 157 | |
| 158 | * "git <cmd> @{push}" on a detached HEAD used to segfault; it has |
| 159 | been corrected to error out with a message. |
| 160 | |
| 161 | * Tighten a test to avoid mistaking an extended ERE regexp engine as |
| 162 | a PRE regexp engine. |
| 163 | |
| 164 | * Typing ^C to pager, which usually does not kill it, killed Git and |
| 165 | took the pager down as a collateral damage in certain process-tree |
| 166 | structure. This has been fixed. |
| 167 | |
| 168 | Also contains various documentation updates and code clean-ups. |