Context Navigation

← Previous Change
Ticket History
Next Change →

Changes between Initial Version and Version 2 of Ticket #6759

Timestamp:: 09/30/2015 12:29:49 PM (9 years ago)
Author:: Fernando de Oliveira
Comment:

Legend:

: Unmodified
: Added
: Removed
: Modified

Ticket #6759
- Property Summary git-2.6 (wait for minor releases or before BLFS-7.8 release - do not update too frequently) → git-2.6.0
- Property Milestone hold → 7.9

Ticket #6759 – Description

-              initial
+              v2
+git is releasing about once a week.  That's really too often for the book.  This is a placeholder for the 2.5 minor release and should stay open while subsequent point releases are made.
+Another condition for update, this time only point release is acceptable: just before BLFS-7.8 release.
+[https://www.kernel.org/pub/software/scm/git/git-2.6.0.tar.xz]
+[https://www.kernel.org/pub/software/scm/git/git-2.6.0.tar.sign]
+[https://raw.githubusercontent.com/git/git/master/Documentation/RelNotes/2.6.0.txt]
+{{{
+Git 2.6 Release Notes
+=====================
+Updates since v2.5
+------------------
+UI, Workflows & Features
+ * An asterisk as a substring (as opposed to the entirety) of a path
+   component for both side of a refspec, e.g.
+   "refs/heads/o*:refs/remotes/heads/i*", is now allowed.
+ * New userdiff pattern definition for fountain screenwriting markup
+   format has been added.
+ * "git log" and friends learned a new "--date=format:..." option to
+   format timestamps using system's strftime(3).
+ * "git fast-import" learned to respond to the get-mark command via
+   its cat-blob-fd interface.
+ * "git rebase -i" learned "drop commit-object-name subject" command
+   as another way to skip replaying of a commit.
+ * A new configuration variable can enable "--follow" automatically
+   when "git log" is run with one pathspec argument.
+ * "git status" learned to show a more detailed information regarding
+   the "rebase -i" session in progress.
+ * "git cat-file" learned "--batch-all-objects" option to enumerate all
+   available objects in the repository more quickly than "rev-list
+   --all --objects" (the output includes unreachable objects, though).
+ * "git fsck" learned to ignore errors on a set of known-to-be-bad
+   objects, and also allows the warning levels of various kinds of
+   non-critical breakages to be tweaked.
+ * "git rebase -i"'s list of todo is made configurable.
+ * "git send-email" now performs alias-expansion on names that are
+   given via --cccmd, etc.
+ * An environment variable GIT_REPLACE_REF_BASE tells Git to look into
+   refs hierarchy other than refs/replace/ for the object replacement
+   data.
+ * Allow untracked cache (experimental) to be used when sparse
+   checkout (experimental) is also in use.
+ * "git pull --rebase" has been taught to pay attention to
+   rebase.autostash configuration.
+ * The command-line completion script (in contrib/) has been updated.
+ * A negative !ref entry in multi-value transfer.hideRefs
+   configuration can be used to say "don't hide this one".
+ * After "git am" without "-3" stops, running "git am -3" pays attention
+   to "-3" only for the patch that caused the original invocation
+   to stop.
+ * When linked worktree is used, simultaneous "notes merge" instances
+   for the same ref in refs/notes/* are prevented from stomping on
+   each other.
+ * "git send-email" learned a new option --smtp-auth to limit the SMTP
+   AUTH mechanisms to be used to a subset of what the system library
+   supports.
+ * A new configuration variable http.sslVersion can be used to specify
+   what specific version of SSL/TLS to use to make a connection.
+ * "git notes merge" can be told with "--strategy=<how>" option how to
+   automatically handle conflicts; this can now be configured by
+   setting notes.mergeStrategy configuration variable.
+ * "git log --cc" did not show any patch, even though most of the time
+   the user meant "git log --cc -p -m" to see patch output for commits
+   with a single parent, and combined diff for merge commits.  The
+   command is taught to DWIM "--cc" (without "--raw" and other forms
+   of output specification) to "--cc -p -m".
+ * "git config --list" output was hard to parse when values consist of
+   multiple lines.  "--name-only" option is added to help this.
+ * A handful of usability & cosmetic fixes to gitk and l10n updates.
+ * A completely empty e-mail address <> is now allowed in the authors
+   file used by git-svn, to match the way it accepts the output from
+   authors-prog.
+Performance, Internal Implementation, Development Support etc.
+ * In preparation for allowing different "backends" to store the refs
+   in a way different from the traditional "one ref per file in
+   $GIT_DIR or in a $GIT_DIR/packed-refs file" filesystem storage,
+   direct filesystem access to ref-like things like CHERRY_PICK_HEAD
+   from scripts and programs has been reduced.
+ * Computation of untracked status indicator by bash prompt
+   script (in contrib/) has been optimized.
+ * Memory use reduction when commit-slab facility is used to annotate
+   sparsely (which is not recommended in the first place).
+ * Clean up refs API and make "git clone" less intimate with the
+   implementation detail.
+ * "git pull" was reimplemented in C.
+ * The packet tracing machinery allows to capture an incoming pack
+   data to a file for debugging.
+ * Move machinery to parse human-readable scaled numbers like 1k, 4M,
+   and 2G as an option parameter's value from pack-objects to
+   parse-options API, to make it available to other codepaths.
+ * "git verify-tag" and "git verify-commit" have been taught to share
+   more code, and then learned to optionally show the verification
+   message from the underlying GPG implementation.
+ * Various enhancements around "git am" reading patches generated by
+   foreign SCM have been made.
+ * Ref listing by "git branch -l" and "git tag -l" commands has
+   started to be rebuilt, based on the for-each-ref machinery.
+ * The code to perform multi-tree merges has been taught to repopulate
+   the cache-tree upon a successful merge into the index, so that
+   subsequent "diff-index --cached" (hence "status") and "write-tree"
+   (hence "commit") will go faster.
+   The same logic in "git checkout" may now be removed, but that is a
+   separate issue.
+ * Tests that assume how reflogs are represented on the filesystem too
+   much have been corrected.
+ * "git am" has been rewritten in "C".
+ * git_path() and mkpath() are handy helper functions but it is easy
+   to misuse, as the callers need to be careful to keep the number of
+   active results below 4.  Their uses have been reduced.
+ * The "lockfile" API has been rebuilt on top of a new "tempfile" API.
+ * To prepare for allowing a different "ref" backend to be plugged in
+   to the system, update_ref()/delete_ref() have been taught about
+   ref-like things like MERGE_HEAD that are per-worktree (they will
+   always be written to the filesystem inside $GIT_DIR).
+ * The gitmodules API that is accessed from the C code learned to
+   cache stuff lazily.
+Also contains various documentation updates and code clean-ups.
+Fixes since v2.5
+----------------
+Unless otherwise noted, all the fixes since v2.5 in the maintenance
+track are contained in this release (see the maintenance releases'
+notes for details).
+ * "git subtree" (in contrib/) depended on "git log" output to be
+   stable, which was a no-no.  Apply a workaround to force a
+   particular date format.
+   (merge e7aac44 da/subtree-date-confusion later to maint).
+ * An attempt to delete a ref by pushing into a repository whose HEAD
+   symbolic reference points at an unborn branch that cannot be
+   created due to ref D/F conflict (e.g. refs/heads/a/b exists, HEAD
+   points at refs/heads/a) failed.
+   (merge b112b14 jx/do-not-crash-receive-pack-wo-head later to maint).
+ * The low-level "git send-pack" did not honor 'user.signingkey'
+   configuration variable when sending a signed-push.
+   (merge d830d39 db/send-pack-user-signingkey later to maint).
+ * "sparse checkout" misbehaved for a path that is excluded from the
+   checkout when switching between branches that differ at the path.
+   (merge 7d78241 as/sparse-checkout-removal later to maint).
+ * An experimental "untracked cache" feature used uname(2) in a
+   slightly unportable way.
+   (merge 100e433 cb/uname-in-untracked later to maint).
+ * A "rebase" replays changes of the local branch on top of something
+   else, as such they are placed in stage #3 and referred to as
+   "theirs", while the changes in the new base, typically a foreign
+   work, are placed in stage #2 and referred to as "ours".  Clarify
+   the "checkout --ours/--theirs".
+   (merge f303016 se/doc-checkout-ours-theirs later to maint).
+ * The "rev-parse --parseopt" mode parsed the option specification
+   and the argument hint in a strange way to allow '=' and other
+   special characters in the option name while forbidding them from
+   the argument hint.  This made it impossible to define an option
+   like "--pair <key>=<value>" with "pair=key=value" specification,
+   which instead would have defined a "--pair=key <value>" option.
+   (merge 2d893df ib/scripted-parse-opt-better-hint-string later to
+   maint).
+ * Often a fast-import stream builds a new commit on top of the
+   previous commit it built, and it often unconditionally emits a
+   "from" command to specify the first parent, which can be omitted in
+   such a case.  This caused fast-import to forget the tree of the
+   previous commit and then re-read it from scratch, which was
+   inefficient.  Optimize for this common case.
+   (merge 0df3245 mh/fast-import-optimize-current-from later to maint).
+ * Running an aliased command from a subdirectory when the .git thing
+   in the working tree is a gitfile pointing elsewhere did not work.
+   (merge d95138e nd/export-worktree later to maint).
+ * "Is this subdirectory a separate repository that should not be
+   touched?" check "git clean" was inefficient.  This was replaced
+   with a more optimized check.
+   (merge fbf2fec ee/clean-remove-dirs later to maint).
+ * The "new-worktree-mode" hack in "checkout" that was added in
+   nd/multiple-work-trees topic has been removed by updating the
+   implementation of new "worktree add".
+   (merge 65f9b75 es/worktree-add-cleanup later to maint).
+ * Remove remaining cruft from  "git checkout --to", which
+   transitioned to "git worktree add".
+   (merge 114ff88 es/worktree-add later to maint).
+ * An off-by-one error made "git remote" to mishandle a remote with a
+   single letter nickname.
+   (merge bc598c3 mh/get-remote-group-fix later to maint).
+ * "git clone $URL", when cloning from a site whose sole purpose is to
+   host a single repository (hence, no path after <scheme>://<site>/),
+   tried to use the site name as the new repository name, but did not
+   remove username or password when <site> part was of the form
+   <user>@<pass>:<host>.  The code is taught to redact these.
+   (merge adef956 ps/guess-repo-name-at-root later to maint).
+ * Running tests with the "-x" option to make them verbose had some
+   unpleasant interactions with other features of the test suite.
+   (merge 9b5fe78 jk/test-with-x later to maint).
+ * t1509 test that requires a dedicated VM environment had some
+   bitrot, which has been corrected.
+   (merge faacc5a ps/t1509-chroot-test-fixup later to maint).
+ * "git pull" in recent releases of Git has a regression in the code
+   that allows custom path to the --upload-pack=<program>.  This has
+   been corrected.
+   Note that this is irrelevant for 'master' with "git pull" rewritten
+   in C.
+   (merge 13e0e28 mm/pull-upload-pack later to maint).
+ * When trying to see that an object does not exist, a state errno
+   leaked from our "first try to open a packfile with O_NOATIME and
+   then if it fails retry without it" logic on a system that refuses
+   O_NOATIME.  This confused us and caused us to die, saying that the
+   packfile is unreadable, when we should have just reported that the
+   object does not exist in that packfile to the caller.
+   (merge dff6f28 cb/open-noatime-clear-errno later to maint).
+ * The codepath to produce error messages had a hard-coded limit to
+   the size of the message, primarily to avoid memory allocation while
+   calling die().
+   (merge f4c3edc jk/long-error-messages later to maint).
+ * strbuf_read() used to have one extra iteration (and an unnecessary
+   strbuf_grow() of 8kB), which was eliminated.
+   (merge 3ebbd00 jh/strbuf-read-use-read-in-full later to maint).
+ * We rewrote one of the build scripts in Perl but this reimplements
+   in Bourne shell.
+   (merge 57cee8a sg/help-group later to maint).
+ * The experimental untracked-cache feature were buggy when paths with
+   a few levels of subdirectories are involved.
+   (merge 73f9145 dt/untracked-subdir later to maint).
+ * "interpret-trailers" helper mistook a single-liner log message that
+   has a colon as the end of existing trailer.
+ * The "interpret-trailers" helper mistook a multi-paragraph title of
+   a commit log message with a colon in it as the end of the trailer
+   block.
+   (merge 5c99995 cc/trailers-corner-case-fix later to maint).
+ * "git describe" without argument defaulted to describe the HEAD
+   commit, but "git describe --contains" didn't.  Arguably, in a
+   repository used for active development, such defaulting would not
+   be very useful as the tip of branch is typically not tagged, but it
+   is better to be consistent.
+   (merge 2bd0706 sg/describe-contains later to maint).
+ * The client side codepaths in "git push" have been cleaned up
+   and the user can request to perform an optional "signed push",
+   i.e. sign only when the other end accepts signed push.
+   (merge 68c757f db/push-sign-if-asked later to maint).
+ * Because the configuration system does not allow "alias.0foo" and
+   "pager.0foo" as the configuration key, the user cannot use '0foo'
+   as a custom command name anyway, but "git 0foo" tried to look these
+   keys up and emitted useless warnings before saying '0foo is not a
+   git command'.  These warning messages have been squelched.
+   (merge 9e9de18 jk/fix-alias-pager-config-key-warnings later to
+   maint).
+ * "git rev-list" does not take "--notes" option, but did not complain
+   when one is given.
+   (merge 2aea7a5 jk/rev-list-has-no-notes later to maint).
+ * When re-priming the cache-tree opportunistically while committing
+   the in-core index as-is, we mistakenly invalidated the in-core
+   index too aggressively, causing the experimental split-index code
+   to unnecessarily rewrite the on-disk index file(s).
+   (merge 475a344
+   dt/commit-preserve-base-index-upon-opportunistic-cache-tree-update
+   later to maint).
+ * "git archive" did not use zip64 extension when creating an archive
+   with more than 64k entries, which nobody should need, right ;-)?
+   (merge 88329ca rs/archive-zip-many later to maint).
+ * The code in "multiple-worktree" support that attempted to recover
+   from an inconsistent state updated an incorrect file.
+   (merge 82fde87 nd/fixup-linked-gitdir later to maint).
+ * On case insensitive systems, "git p4" did not work well with client
+   specs.
+ * "git init empty && git -C empty log" said "bad default revision 'HEAD'",
+   which was found to be a bit confusing to new users.
+   (merge ce11360 jk/log-missing-default-HEAD later to maint).
+ * Recent versions of scripted "git am" has a performance regression in
+   "git am --skip" codepath, which no longer exists in the built-in
+   version on the 'master' front.  Fix the regression in the last
+   scripted version that appear in 2.5.x maintenance track and older.
+   (merge b9d6689 js/maint-am-skip-performance-regression later to
+   maint).
+ * The branch descriptions that are set with "git branch
+   --edit-description" option were used in many places but they weren't
+   clearly documented.
+   (merge 561d2b7 po/doc-branch-desc later to maint).
+ * Code cleanups and documentation updates.
+   (merge 1c601af es/doc-clean-outdated-tools later to maint).
+   (merge 3581304 kn/tag-doc-fix later to maint).
+   (merge 3a59e59 kb/i18n-doc later to maint).
+   (merge 45abdee sb/remove-unused-var-from-builtin-add later to maint).
+   (merge 14691e3 sb/parse-options-codeformat later to maint).
+   (merge 4a6ada3 ad/bisect-cleanup later to maint).
+   (merge da4c5ad ta/docfix-index-format-tech later to maint).
+   (merge ae25fd3 sb/check-return-from-read-ref later to maint).
+   (merge b3325df nd/dwim-wildcards-as-pathspecs later to maint).
+   (merge 7aa9b9b sg/wt-status-header-inclusion later to maint).
+   (merge f04c690 as/docfix-reflog-expire-unreachable later to maint).
+   (merge 1269847 sg/t3020-typofix later to maint).
+   (merge 8b54c23 jc/calloc-pathspec later to maint).
+   (merge a6926b8 po/po-readme later to maint).
+   (merge 54d160e ss/fix-config-fd-leak later to maint).
+   (merge b80fa84 ah/submodule-typofix-in-error later to maint).
+   (merge 99885bc ah/reflog-typofix-in-error later to maint).
+   (merge 9476c2c ah/read-tree-usage-string later to maint).
+   (merge b8c1d27 ah/pack-objects-usage-strings later to maint).
+   (merge 486e1e1 br/svn-doc-include-paths-config later to maint).
+   (merge 1733ed3 ee/clean-test-fixes later to maint).
+   (merge 5fcadc3 gb/apply-comment-typofix later to maint).
+   (merge b894d3e mp/t7060-diff-index-test later to maint).
+   (merge d238710 as/config-doc-markup-fix later to maint).
+}}}