1Git 1.7.12.1 Release Notes 2========================== 3 4Fixes since v1.7.12 5------------------- 6 7 * "git apply -p0" did not parse pathnames on "diff --git" line 8 correctly. This caused patches that had pathnames in no other 9 places to be mistakenly rejected (most notably, binary patch that 10 does not rename nor change mode). Textual patches, renames or mode 11 changes have preimage and postimage pathnames in different places 12 in a form that can be parsed unambiguously and did not suffer from 13 this problem. 14 15 * "git cherry-pick A C B" used to replay changes in A and then B and 16 then C if these three commits had committer timestamps in that 17 order, which is not what the user who said "A C B" naturally 18 expects. 19 20 * "git commit --amend" let the user edit the log message and then 21 died when the human-readable committer name was given 22 insufficiently by getpwent(3). 23 24 * Some capabilities were asked by fetch-pack even when upload-pack 25 did not advertise that they are available. fetch-pack has been 26 fixed not to do so. 27 28 * "git diff" had a confusion between taking data from a path in the 29 working tree and taking data from an object that happens to have 30 name 0{40} recorded in a tree. 31 32 * "git for-each-ref" did not correctly support more than one --sort 33 option. 34 35 * "git log .." errored out saying it is both rev range and a path 36 when there is no disambiguating "--" is on the command line. 37 Update the command line parser to interpret ".." as a path in such 38 a case. 39 40 * The "--topo-order", "--date-order" (and the lack of either means 41 the default order) options to "rev-list" and "log" family of 42 commands were poorly described in the documentation. 43 44 * "git prune" without "-v" used to warn about leftover temporary 45 files (which is an indication of an earlier aborted operation). 46 47 * Pushing to smart HTTP server with recent Git fails without having 48 the username in the URL to force authentication, if the server is 49 configured to allow GET anonymously, while requiring authentication 50 for POST. 51 52 * The reflog entries left by "git rebase" and "git rebase -i" were 53 inconsistent (the interactive one gave an abbreviated object name). 54 55 * When "git push" triggered the automatic gc on the receiving end, a 56 message from "git prune" that said it was removing cruft leaked to 57 the standard output, breaking the communication protocol. 58 59 * "git show --quiet" ought to be a synonym for "git show -s", but 60 wasn't. 61 62 * "git show --format='%ci'" did not give timestamp correctly for 63 commits created without human readable name on "committer" line. 64 65 * "git send-email" did not unquote encoded words that appear on the 66 header correctly, and lost "_" from strings. 67 68 * The interactive prompt "git send-email" gives was error prone. It 69 asked "What e-mail address do you want to use?" with the address it 70 guessed (correctly) the user would want to use in its prompt, 71 tempting the user to say "y". But the response was taken as "No, 72 please use 'y' as the e-mail address instead", which is most 73 certainly not what the user meant. 74 75 * "gitweb" when used with PATH_INFO failed to notice directories with 76 SP (and other characters that need URL-style quoting) in them. 77 78 * When the user gives an argument that can be taken as both a 79 revision name and a pathname without disambiguating with "--", we 80 used to give a help message "Use '--' to separate". The message 81 has been clarified to show where that '--' goes on the command 82 line. 83 84 * When the user exports a non-default IFS without HT, scripts that 85 rely on being able to parse "ls-files -s | while read a b c..." 86 started to fail. Protect them from such a misconfiguration. 87 88 * The attribute system may be asked for a path that itself or its 89 leading directories no longer exists in the working tree, and it is 90 fine if we cannot open .gitattribute file in such a case. Failure 91 to open per-directory .gitattributes with error status other than 92 ENOENT and ENOTDIR should be diagnosed, but it wasn't. 93 94 * After "gitk" showed the contents of a tag, neither "Reread 95 references" nor "Reload" did not update what is shown as the 96 contents of it, when the user overwrote the tag with "git tag -f". 97 98 * "ciabot" script (in contrib/) has been updated with extensive 99 documentation. 100 101 * "git-jump" script (in contrib/) did not work well when 102 diff.noprefix or diff.mnemonicprefix is in effect. 103 104 * Older parts of the documentation described as if having a regular 105 file in .git/refs/ hierarchy were the only way to have branches and 106 tags, which is not true for quite some time. 107 108 * A utility shell function test_seq has been added as a replacement 109 for the 'seq' utility found on some platforms. 110 111 * Compatibility wrapper to learn the maximum number of file 112 descriptors we can open around sysconf(_SC_OPEN_MAX) and 113 getrlimit(RLIMIT_NO_FILE) has been introduced for portability. 114 115 * We used curl_easy_strerror() without checking version of cURL, 116 breaking the build for versions before curl 7.12.0. 117 118 * Code to work around MacOS X UTF-8 gotcha has been cleaned up. 119 120 * Fallback 'getpass' implementation made unportable use of stdio API. 121 122 * The "--rebase" option to "git pull" can be abbreviated to "-r", 123 but we didn't document it. 124 125 * It was generally understood that "--long-option"s to many of our 126 subcommands can be abbreviated to the unique prefix, but it was not 127 easy to find it described for new readers of the documentation set. 128 129 * The synopsis said "checkout [-B branch]" to make it clear the 130 branch name is a parameter to the option, but the heading for the 131 option description was "-B::", not "-B branch::", making the 132 documentation misleading. 133 134Also contains numerous documentation updates.