1Git 2.6 Release Notes 2===================== 3 4Updates since v2.5 5------------------ 6 7UI, Workflows & Features 8 9 * An asterisk as a substring (as opposed to the entirety) of a path 10 component for both side of a refspec, e.g. 11 "refs/heads/o*:refs/remotes/heads/i*", is now allowed. 12 13 * New userdiff pattern definition for fountain screenwriting markup 14 format has been added. 15 16 * "git log" and friends learned a new "--date=format:..." option to 17 format timestamps using system's strftime(3). 18 19 * "git fast-import" learned to respond to the get-mark command via 20 its cat-blob-fd interface. 21 22 * "git rebase -i" learned "drop commit-object-name subject" command 23 as another way to skip replaying of a commit. 24 25 * A new configuration variable can enable "--follow" automatically 26 when "git log" is run with one pathspec argument. 27 28 * "git status" learned to show a more detailed information regarding 29 the "rebase -i" session in progress. 30 31 * "git cat-file" learned "--batch-all-objects" option to enumerate all 32 available objects in the repository more quickly than "rev-list 33 --all --objects" (the output includes unreachable objects, though). 34 35 * "git fsck" learned to ignore errors on a set of known-to-be-bad 36 objects, and also allows the warning levels of various kinds of 37 non-critical breakages to be tweaked. 38 39 * "git rebase -i"'s list of todo is made configurable. 40 41 * "git send-email" now performs alias-expansion on names that are 42 given via --cccmd, etc. 43 44 * An environment variable GIT_REPLACE_REF_BASE tells Git to look into 45 refs hierarchy other than refs/replace/ for the object replacement 46 data. 47 48 * Allow untracked cache (experimental) to be used when sparse 49 checkout (experimental) is also in use. 50 51 * "git pull --rebase" has been taught to pay attention to 52 rebase.autostash configuration. 53 54 * The command-line completion script (in contrib/) has been updated. 55 56 * A negative !ref entry in multi-value transfer.hideRefs 57 configuration can be used to say "don't hide this one". 58 59 * After "git am" without "-3" stops, running "git am -" pays attention 60 to "-3" only for the patch that caused the original invocation 61 to stop. 62 63 * When linked worktree is used, simultaneous "notes merge" instances 64 for the same ref in refs/notes/* are prevented from stomping on 65 each other. 66 67 68Performance, Internal Implementation, Development Support etc. 69 70 * In preparation for allowing different "backends" to store the refs 71 in a way different from the traditional "one ref per file in 72 $GIT_DIR or in a $GIT_DIR/packed-refs file" filesystem storage, 73 direct filesystem access to ref-like things like CHERRY_PICK_HEAD 74 from scripts and programs has been reduced. 75 76 * Computation of untracked status indicator by bash prompt 77 script (in contrib/) has been optimized. 78 79 * Memory use reduction when commit-slab facility is used to annotate 80 sparsely (which is not recommended in the first place). 81 82 * Clean up refs API and make "git clone" less intimate with the 83 implementation detail. 84 85 * "git pull" was reimplemented in C. 86 87 * The packet tracing machinery allows to capture an incoming pack 88 data to a file for debugging. 89 90 * Move machinery to parse human-readable scaled numbers like 1k, 4M, 91 and 2G as an option parameter's value from pack-objects to 92 parse-options API, to make it available to other codepaths. 93 94 * "git verify-tag" and "git verify-commit" have been taught to share 95 more code, and then learned to optionally show the verification 96 message from the underlying GPG implementation. 97 98 * Various enhancements around "git am" reading patches generated by 99 foreign SCM have been made. 100 101 * Ref listing by "git branch -l" and "git tag -l" commands has 102 started to be rebuilt, based on the for-each-ref machinery. 103 104 * The code to perform multi-tree merges has been taught to repopulate 105 the cache-tree upon a successful merge into the index, so that 106 subsequent "diff-index --cached" (hence "status") and "write-tree" 107 (hence "commit") will go faster. 108 109 The same logic in "git checkout" may now be removed, but that is a 110 separate issue. 111 112 * Tests that assume how reflogs are represented on the filesystem too 113 much have been corrected. 114 115 * "git am" has been rewritten in "C". 116 117 * git_path() and mkpath() are handy helper functions but it is easy 118 to misuse, as the callers need to be careful to keep the number of 119 active results below 4. Their uses have been reduced. 120 121 * The "lockfile" API has been rebuilt on top of a new "tempfile" API. 122 123 * To prepare for allowing a different "ref" backend to be plugged in 124 to the system, update_ref()/delete_ref() have been taught about 125 ref-like things like MERGE_HEAD that are per-worktree (they will 126 always be written to the filesystem inside $GIT_DIR). 127 128 129Also contains various documentation updates and code clean-ups. 130 131 132Fixes since v2.5 133---------------- 134 135Unless otherwise noted, all the fixes since v2.5 in the maintenance 136track are contained in this release (see the maintenance releases' 137notes for details). 138 139 * "git subtree" (in contrib/) depended on "git log" output to be 140 stable, which was a no-no. Apply a workaround to force a 141 particular date format. 142 (merge e7aac44 da/subtree-date-confusion later to maint). 143 144 * An attempt to delete a ref by pushing into a repository whose HEAD 145 symbolic reference points at an unborn branch that cannot be 146 created due to ref D/F conflict (e.g. refs/heads/a/b exists, HEAD 147 points at refs/heads/a) failed. 148 (merge b112b14 jx/do-not-crash-receive-pack-wo-head later to maint). 149 150 * The low-level "git send-pack" did not honor 'user.signingkey' 151 configuration variable when sending a signed-push. 152 (merge d830d39 db/send-pack-user-signingkey later to maint). 153 154 * "sparse checkout" misbehaved for a path that is excluded from the 155 checkout when switching between branches that differ at the path. 156 (merge 7d78241 as/sparse-checkout-removal later to maint). 157 158 * An experimental "untracked cache" feature used uname(2) in a 159 slightly unportable way. 160 (merge 100e433 cb/uname-in-untracked later to maint). 161 162 * A "rebase" replays changes of the local branch on top of something 163 else, as such they are placed in stage #3 and referred to as 164 "theirs", while the changes in the new base, typically a foreign 165 work, are placed in stage #2 and referred to as "ours". Clarify 166 the "checkout --ours/--theirs". 167 (merge f303016 se/doc-checkout-ours-theirs later to maint). 168 169 * The "rev-parse --parseopt" mode parsed the option specification 170 and the argument hint in a strange way to allow '=' and other 171 special characters in the option name while forbidding them from 172 the argument hint. This made it impossible to define an option 173 like "--pair <key>=<value>" with "pair=key=value" specification, 174 which instead would have defined a "--pair=key <value>" option. 175 (merge 2d893df ib/scripted-parse-opt-better-hint-string later to maint). 176 177 * Often a fast-import stream builds a new commit on top of the 178 previous commit it built, and it often unconditionally emits a 179 "from" command to specify the first parent, which can be omitted in 180 such a case. This caused fast-import to forget the tree of the 181 previous commit and then re-read it from scratch, which was 182 inefficient. Optimize for this common case. 183 (merge 0df3245 mh/fast-import-optimize-current-from later to maint). 184 185 * Running an aliased command from a subdirectory when the .git thing 186 in the working tree is a gitfile pointing elsewhere did not work. 187 (merge d95138e nd/export-worktree later to maint). 188 189 * "Is this subdirectory a separate repository that should not be 190 touched?" check "git clean" was inefficient. This was replaced 191 with a more optimized check. 192 (merge 38ae878 ee/clean-remove-dirs later to maint). 193 194 * The "new-worktree-mode" hack in "checkout" that was added in 195 nd/multiple-work-trees topic has been removed by updating the 196 implementation of new "worktree add". 197 (merge 65f9b75 es/worktree-add-cleanup later to maint). 198 199 * Remove remaining cruft from "git checkout --to", which 200 transitioned to "git worktree add". 201 (merge 114ff88 es/worktree-add later to maint). 202 203 * An off-by-one error made "git remote" to mishandle a remote with a 204 single letter nickname. 205 (merge bc598c3 mh/get-remote-group-fix later to maint). 206 207 * "git clone $URL", when cloning from a site whose sole purpose is to 208 host a single repository (hence, no path after <scheme>://<site>/), 209 tried to use the site name as the new repository name, but did not 210 remove username or password when <site> part was of the form 211 <user>@<pass>:<host>. The code is taught to redact these. 212 (merge adef956 ps/guess-repo-name-at-root later to maint). 213 214 * Running tests with the "-x" option to make them verbose had some 215 unpleasant interactions with other features of the test suite. 216 (merge 9b5fe78 jk/test-with-x later to maint). 217 218 * t1509 test that requires a dedicated VM environment had some 219 bitrot, which has been corrected. 220 (merge faacc5a ps/t1509-chroot-test-fixup later to maint). 221 222 * "git pull" in recent releases of Git has a regression in the code 223 that allows custom path to the --upload-pack=<program>. This has 224 been corrected. 225 226 Note that this is irrelevant for 'master' with "git pull" rewritten 227 in C. 228 (merge 22d6857 mm/pull-upload-pack later to maint). 229 230 * When trying to see that an object does not exist, a state errno 231 leaked from our "first try to open a packfile with O_NOATIME and 232 then if it fails retry without it" logic on a system that refuses 233 O_NOATIME. This confused us and caused us to die, saying that the 234 packfile is unreadable, when we should have just reported that the 235 object does not exist in that packfile to the caller. 236 (merge dff6f28 cb/open-noatime-clear-errno later to maint). 237 238 * The codepath to produce error messages had a hard-coded limit to 239 the size of the message, primarily to avoid memory allocation while 240 calling die(). 241 (merge f4c3edc jk/long-error-messages later to maint). 242 243 * strbuf_read() used to have one extra iteration (and an unnecessary 244 strbuf_grow() of 8kB), which was eliminated. 245 (merge 3ebbd00 jh/strbuf-read-use-read-in-full later to maint). 246 247 * Code cleanups and documentation updates. 248 (merge 1c601af es/doc-clean-outdated-tools later to maint). 249 (merge 3581304 kn/tag-doc-fix later to maint). 250 (merge 3a59e59 kb/i18n-doc later to maint). 251 (merge 45abdee sb/remove-unused-var-from-builtin-add later to maint). 252 (merge 14691e3 sb/parse-options-codeformat later to maint). 253 (merge 4a6ada3 ad/bisect-cleanup later to maint). 254 (merge da4c5ad ta/docfix-index-format-tech later to maint). 255 (merge ae25fd3 sb/check-return-from-read-ref later to maint). 256 (merge b3325df nd/dwim-wildcards-as-pathspecs later to maint).