1Git v1.7.11 Release Notes 2========================= 3 4Updates since v1.7.10 5--------------------- 6 7UI, Workflows & Features 8 9 * A new mode for push, "simple", which is a cross between "current" 10 and "upstream", has been introduced. "git push" without any refspec 11 will push the current branch out to the same name at the remote 12 repository only when it is set to track the branch with the same 13 name over there. The plan is to make this mode the new default 14 value when push.default is not configured. 15 16 * A couple of commands learned the "--column" option to produce 17 columnar output. 18 19 * A third-party tool "git subtree" is distributed in contrib/ 20 21 * Error messages given when @{u} is used for a branch without its 22 upstream configured have been clatified. 23 24 * Even with "-q"uiet option, "checkout" used to report setting up 25 tracking. Also "branch" learned the "-q"uiet option to squelch 26 informational message. 27 28 * Your build platform may support hardlinks but you may prefer not to 29 use them, e.g. when installing to DESTDIR to make a tarball and 30 untarring on a filesystem that has poor support for hardlinks. 31 There is a Makefile option NO_INSTALL_HARDLINKS for you. 32 33 * The smart-http backend used to always override GIT_COMMITTER_* 34 variables with REMOTE_USER and REMOTE_ADDR, but these variables are 35 now preserved when set. 36 37 * "git am" learned the "--include" option, which is an opposite of 38 existing the "--exclude" option. 39 40 * When "git am -3" needs to fall back to an application to a 41 synthesized preimage followed by a 3-way merge, the paths that 42 needed such treatment are now reported to the end user, so that the 43 result in them can be eyeballed with extra care. 44 45 * The output from "diff/log --stat" used to always allocate 4 columns 46 to show the number of modified lines, but not anymore. 47 48 * "git difftool" learned the "--dir-diff" option to spawn external 49 diff tools that can compare two directory hierarchies at a time 50 after populating two temporary directories, instead of running an 51 instance of the external tool once per a file pair. 52 53 * The "fmt-merge-msg" command learns to list the primary contributors 54 involved in the side topic you are merging. 55 56 * "git rebase" learned to optionally keep commits that do not 57 introduce any change in the original history. 58 59 * "git push --recurse-submodules" learned to optionally look into the 60 histories of submodules bound to the superproject and push them 61 out. 62 63 * A 'snapshot' request to "gitweb" honors If-Modified-Since: header, 64 based on the commit date. 65 66 * "gitweb" learned to highlight the patch it outputs even more. 67 68Foreign Interface 69 70 * "git svn" used to die with unwanted SIGPIPE when talking with HTTP 71 server that uses keep-alive. 72 73 * "git svn" learned to use platform specific authentication 74 providers, e.g. gnome-keyring, kwallet, etc. 75 76 * "git p4" has been moved out of contrib/ area and has seen more work 77 on importing labels as tags from (and exporting tags as labels to) 78 p4. 79 80Performance and Internal Implementation (please report possible regressions) 81 82 * An experimental "version 4" format of the index file has been 83 introduced to reduce on-disk footprint and I/O overhead. 84 85 * "git archive" learned to produce its output without reading the 86 blob object it writes out in memory in its entirety. 87 88 * The code to compute hash values for lines used by the internal diff 89 engine was optimized on little-endian machines, using the same 90 trick the kernel folks came up with. 91 92 * "git apply" had some memory leaks plugged. 93 94 * Setting up a revision traversal with many starting points was 95 inefficient as these were placed in a date-order priority queue 96 one-by-one. Now they are collected in the queue unordered first, 97 and sorted immediately before getting used. 98 99 * More lower-level commands learned to use the streaming API to read 100 from the object store without keeping everything in core. 101 102 * Because "sh" on the user's PATH may be utterly broken on some 103 systems, run-command API now uses SHELL_PATH, not /bin/sh, when 104 spawning an external command (not applicable to Windows port). 105 106 * The API to iterate over refs/ hierarchy has been tweaked to allow 107 walking only a subset of it more efficiently. 108 109Also contains minor documentation updates and code clean-ups. 110 111 112Fixes since v1.7.10 113------------------- 114 115Unless otherwise noted, all the fixes since v1.7.10 in the maintenance 116releases are contained in this release (see release notes to them for 117details). 118 119 * The DWIM behaviour for "log --pretty=format:%gd -g" was somewhat 120 broken and gave undue precedence to configured log.date, causing 121 "git stash list" to show "stash@{time stamp string}". 122 (merge 55ccf85 jk/maint-reflog-walk-count-vs-time later to maint). 123 124 * Running "git checkout" on an unborn branch used to corrupt HEAD. 125 (merge 8338f77 ef/checkout-empty later to maint). 126 127 * When checking out another commit from an already detached state, we 128 used to report all commits that are not reachable from any of the 129 refs as lossage, but some of them might be reachable from the new 130 HEAD, and there is no need to warn about them. 131 (merge 5d88639 js/checkout-detach-count later to maint). 132 133 * Some time ago, "git clone" lost the progress output for its 134 "checkout" phase; when run without any "--quiet" option, it should 135 give progress to the lengthy operation. 136 (merge 8f63da1 ef/maint-clone-progress-fix later to maint). 137 138 * "git status --porcelain" ignored "--branch" option by mistake. The 139 output for "git status --branch -z" was also incorrect and did not 140 terminate the record for the current branch name with NUL as asked. 141 (merge d4a6bf1 jk/maint-status-porcelain-z-b later to maint). 142 143 * "git diff --stat" used to fully count a binary file with modified 144 execution bits whose contents is unmodified, which was not quite 145 right. 146 147 * "log -z --pretty=tformat:..." did not terminate each record with 148 NUL. The fix is not entirely correct when the output also asks for 149 --patch and/or --stat, though. 150 (merge fafd382 jk/maint-tformat-with-z later to maint).