| 1 | Git v1.8.4.1 Release Notes |
| 2 | ========================== |
| 3 | |
| 4 | Fixes since v1.8.4 |
| 5 | ------------------ |
| 6 | |
| 7 | * Some old versions of bash do not grok some constructs like |
| 8 | 'printf -v varname' which the prompt and completion code started |
| 9 | to use recently. The completion and prompt scripts have been |
| 10 | adjusted to work better with these old versions of bash. |
| 11 | |
| 12 | * In FreeBSD's and NetBSD's "sh", a return in a dot script in a |
| 13 | function returns from the function, not only in the dot script, |
| 14 | breaking "git rebase" on these platforms (regression introduced |
| 15 | in 1.8.4-rc1). |
| 16 | |
| 17 | * "git rebase -i" and other scripted commands were feeding a |
| 18 | random, data dependent error message to 'echo' and expecting it |
| 19 | to come out literally. |
| 20 | |
| 21 | * Setting the "submodule.<name>.path" variable to the empty |
| 22 | "true" caused the configuration parser to segfault. |
| 23 | |
| 24 | * Output from "git log --full-diff -- <pathspec>" looked strange |
| 25 | because comparison was done with the previous ancestor that |
| 26 | touched the specified <pathspec>, causing the patches for paths |
| 27 | outside the pathspec to show more than the single commit has |
| 28 | changed. |
| 29 | |
| 30 | * The auto-tag-following code in "git fetch" tries to reuse the |
| 31 | same transport twice when the serving end does not cooperate and |
| 32 | does not give tags that point to commits that are asked for as |
| 33 | part of the primary transfer. Unfortunately, Git-aware transport |
| 34 | helper interface is not designed to be used more than once, hence |
| 35 | this did not work over smart-http transfer. Fixed. |
| 36 | |
| 37 | * Send a large request to read(2)/write(2) as a smaller but still |
| 38 | reasonably large chunks, which would improve the latency when the |
| 39 | operation needs to be killed and incidentally works around broken |
| 40 | 64-bit systems that cannot take a 2GB write or read in one go. |
| 41 | |
| 42 | * A ".mailmap" file that ends with an incomplete line, when read |
| 43 | from a blob, was not handled properly. |
| 44 | |
| 45 | * The recent "short-cut clone connectivity check" topic broke a |
| 46 | shallow repository when a fetch operation tries to auto-follow |
| 47 | tags. |
| 48 | |
| 49 | * When send-email comes up with an error message to die with upon |
| 50 | failure to start an SSL session, it tried to read the error |
| 51 | string from a wrong place. |
| 52 | |
| 53 | * A call to xread() was used without a loop to cope with short |
| 54 | read in the codepath to stream large blobs to a pack. |
| 55 | |
| 56 | * On platforms with fgetc() and friends defined as macros, the |
| 57 | configuration parser did not compile. |
| 58 | |
| 59 | * New versions of MediaWiki introduced a new API for returning |
| 60 | more than 500 results in response to a query, which would cause |
| 61 | the MediaWiki remote helper to go into an infinite loop. |
| 62 | |
| 63 | * Subversion's serf access method (the only one available in |
| 64 | Subversion 1.8) for http and https URLs in skelta mode tells its |
| 65 | caller to open multiple files at a time, which made "git svn |
| 66 | fetch" complain that "Temp file with moniker 'svn_delta' already |
| 67 | in use" instead of fetching. |
| 68 | |
| 69 | |
| 70 | Also contains a handful of trivial code clean-ups, documentation |
| 71 | updates, updates to the test suite, etc. |