1Git 1.8.1.1 Release Notes 2========================= 3 4Fixes since v1.8.1 5------------------ 6 7 * The attribute mechanism didn't allow limiting attributes to be 8 applied to only a single directory itself with "path/" like the 9 exclude mechanism does. 10 11 * When attempting to read the XDG-style $HOME/.config/git/config and 12 finding that $HOME/.config/git is a file, we gave a wrong error 13 message, instead of treating the case as "a custom config file does 14 not exist there" and moving on. 15 16 * After failing to create a temporary file using mkstemp(), failing 17 pathname was not reported correctly on some platforms. 18 19 * http transport was wrong to ask for the username when the 20 authentication is done by certificate identity. 21 22 * The behaviour visible to the end users was confusing, when they 23 attempt to kill a process spawned in the editor that was in turn 24 launched by Git with SIGINT (or SIGQUIT), as Git would catch that 25 signal and die. We ignore these signals now. 26 27 * A child process that was killed by a signal (e.g. SIGINT) was 28 reported in an inconsistent way depending on how the process was 29 spawned by us, with or without a shell in between. 30 31 * After "git add -N" and then writing a tree object out of the 32 index, the cache-tree data structure got corrupted. 33 34 * "git apply" misbehaved when fixing whitespace breakages by removing 35 excess trailing blank lines in some corner cases. 36 37 * A tar archive created by "git archive" recorded a directory in a 38 way that made NetBSD's implementation of "tar" sometimes unhappy. 39 40 * When "git clone --separate-git-dir=$over_there" is interrupted, it 41 failed to remove the real location of the $GIT_DIR it created. 42 This was most visible when interrupting a submodule update. 43 44 * "git fetch --mirror" and fetch that uses other forms of refspec 45 with wildcard used to attempt to update a symbolic ref that match 46 the wildcard on the receiving end, which made little sense (the 47 real ref that is pointed at by the symbolic ref would be updated 48 anyway). Symbolic refs no longer are affected by such a fetch. 49 50 * The "log --graph" codepath fell into infinite loop in some 51 corner cases. 52 53 * "git merge" started calling prepare-commit-msg hook like "git 54 commit" does some time ago, but forgot to pay attention to the exit 55 status of the hook. 56 57 * "git pack-refs" that ran in parallel to another process that 58 created new refs had a race that can lose new ones. 59 60 * When a line to be wrapped has a solid run of non space characters 61 whose length exactly is the wrap width, "git shortlog -w" failed 62 to add a newline after such a line. 63 64 * The way "git svn" asked for password using SSH_ASKPASS and 65 GIT_ASKPASS was not in line with the rest of the system. 66 67 * "gitweb", when sorting by age to show repositories with new 68 activities first, used to sort repositories with absolutely 69 nothing in it early, which was not very useful. 70 71 * "gitweb", when sorting by age to show repositories with new 72 activities first, used to sort repositories with absolutely 73 nothing in it early, which was not very useful. 74 75 * When autoconf is used, any build on a different commit always ran 76 "config.status --recheck" even when unnecessary. 77 78 * Some scripted programs written in Python did not get updated when 79 PYTHON_PATH changed. 80 81 * We have been carrying a translated and long-unmaintained copy of an 82 old version of the tutorial; removed. 83 84 * Portability issues in many self-test scripts have been addressed. 85 86 87Also contains other minor fixes and documentation updates.