Junio C Hamano | b051c59 | 2016-05-26 13:21:00 -0700 | [diff] [blame] | 1 | Git v2.8.4 Release Notes |
| 2 | ======================== |
| 3 | |
| 4 | Fixes since v2.8.3 |
| 5 | ------------------ |
| 6 | |
| 7 | * Documentation for "git merge --verify-signatures" has been updated |
| 8 | to clarify that the signature of only the commit at the tip is |
| 9 | verified. Also the phrasing used for signature and key validity is |
| 10 | adjusted to align with that used by OpenPGP. |
| 11 | |
| 12 | * On Windows, .git and optionally any files whose name starts with a |
| 13 | dot are now marked as hidden, with a core.hideDotFiles knob to |
| 14 | customize this behaviour. |
| 15 | |
| 16 | * Portability enhancement for "rebase -i" to help platforms whose |
| 17 | shell does not like "for i in <empty>" (which is not POSIX-kosher). |
| 18 | |
| 19 | * "git fsck" learned to catch NUL byte in a commit object as |
| 20 | potential error and warn. |
| 21 | |
| 22 | * CI test was taught to build documentation pages. |
| 23 | |
| 24 | * Many 'linkgit:<git documentation page>' references were broken, |
| 25 | which are all fixed with this. |
| 26 | |
Junio C Hamano | 4b0891f | 2016-05-31 14:11:38 -0700 | [diff] [blame] | 27 | * "git describe --contains" often made a hard-to-justify choice of |
| 28 | tag to give name to a given commit, because it tried to come up |
| 29 | with a name with smallest number of hops from a tag, causing an old |
| 30 | commit whose close descendant that is recently tagged were not |
| 31 | described with respect to an old tag but with a newer tag. It did |
| 32 | not help that its computation of "hop" count was further tweaked to |
| 33 | penalize being on a side branch of a merge. The logic has been |
| 34 | updated to favor using the tag with the oldest tagger date, which |
| 35 | is a lot easier to explain to the end users: "We describe a commit |
| 36 | in terms of the (chronologically) oldest tag that contains the |
| 37 | commit." |
| 38 | |
| 39 | * Running tests with '-x' option to trace the individual command |
| 40 | executions is a useful way to debug test scripts, but some tests |
| 41 | that capture the standard error stream and check what the command |
| 42 | said can be broken with the trace output mixed in. When running |
| 43 | our tests under "bash", however, we can redirect the trace output |
| 44 | to another file descriptor to keep the standard error of programs |
| 45 | being tested intact. |
| 46 | |
| 47 | * "http.cookieFile" configuration variable clearly wants a pathname, |
| 48 | but we forgot to treat it as such by e.g. applying tilde expansion. |
| 49 | |
| 50 | * When de-initialising all submodules, "git submodule deinit" gave a |
| 51 | faulty recommendation to use "git submodule deinit .", which would |
| 52 | result in a strange error message in a pathological corner case. |
| 53 | This has been corrected to suggest "submodule deinit --all" instead. |
| 54 | |
Junio C Hamano | 0b65a8d | 2016-06-06 14:29:32 -0700 | [diff] [blame] | 55 | * Many commands normalize command line arguments from NFD to NFC |
| 56 | variant of UTF-8 on OSX, but commands in the "diff" family did |
| 57 | not, causing "git diff $path" to complain that no such path is |
| 58 | known to Git. They have been taught to do the normalization. |
| 59 | |
| 60 | * A couple of bugs around core.autocrlf have been fixed. |
| 61 | |
| 62 | * "git difftool" learned to handle unmerged paths correctly in |
| 63 | dir-diff mode. |
| 64 | |
| 65 | * The "are we talking with TTY, doing an interactive session?" |
| 66 | detection has been updated to work better for "Git for Windows". |
| 67 | |
| 68 | |
Junio C Hamano | b051c59 | 2016-05-26 13:21:00 -0700 | [diff] [blame] | 69 | Also contains other minor documentation updates and code clean-ups. |