tvl-depot/third_party/git/Documentation/RelNotes/2.9.0.txt
2020-05-22 17:46:45 +01:00

512 lines
21 KiB
Text

Git 2.9 Release Notes
=====================
Backward compatibility notes
----------------------------
The end-user facing Porcelain level commands in the "git diff" and
"git log" family by default enable the rename detection; you can still
use "diff.renames" configuration variable to disable this.
Merging two branches that have no common ancestor with "git merge" is
by default forbidden now to prevent creating such an unusual merge by
mistake.
The output formats of "git log" that indents the commit log message by
4 spaces now expands HT in the log message by default. You can use
the "--no-expand-tabs" option to disable this.
"git commit-tree" plumbing command required the user to always sign
its result when the user sets the commit.gpgsign configuration
variable, which was an ancient mistake, which this release corrects.
A script that drives commit-tree, if it relies on this mistake, now
needs to read commit.gpgsign and pass the -S option as necessary.
Updates since v2.8
------------------
UI, Workflows & Features
* Comes with git-multimail 1.3.1 (in contrib/).
* The end-user facing commands like "git diff" and "git log"
now enable the rename detection by default.
* The credential.helper configuration variable is cumulative and
there is no good way to override it from the command line. As
a special case, giving an empty string as its value now serves
as the signal to clear the values specified in various files.
* A new "interactive.diffFilter" configuration can be used to
customize the diff shown in "git add -i" sessions.
* "git p4" now allows P4 author names to be mapped to Git author
names.
* "git rebase -x" can be used without passing "-i" option.
* "git -c credential.<var>=<value> submodule" can now be used to
propagate configuration variables related to credential helper
down to the submodules.
* "git tag" can create an annotated tag without explicitly given an
"-a" (or "-s") option (i.e. when a tag message is given). A new
configuration variable, tag.forceSignAnnotated, can be used to tell
the command to create signed tag in such a situation.
* "git merge" used to allow merging two branches that have no common
base by default, which led to a brand new history of an existing
project created and then get pulled by an unsuspecting maintainer,
which allowed an unnecessary parallel history merged into the
existing project. The command has been taught not to allow this by
default, with an escape hatch "--allow-unrelated-histories" option
to be used in a rare event that merges histories of two projects
that started their lives independently.
* "git pull" has been taught to pass the "--allow-unrelated-histories"
option to underlying "git merge".
* "git apply -v" learned to report paths in the patch that were
skipped via --include/--exclude mechanism or being outside the
current working directory.
* Shell completion (in contrib/) updates.
* The commit object name reported when "rebase -i" stops has been
shortened.
* "git worktree add" can be given "--no-checkout" option to only
create an empty worktree without checking out the files.
* "git mergetools" learned to drive ExamDiff.
* "git pull --rebase" learned "--[no-]autostash" option, so that
the rebase.autostash configuration variable set to true can be
overridden from the command line.
* When "git log" shows the log message indented by 4-spaces, the
remainder of a line after a HT does not align in the way the author
originally intended. The command now expands tabs by default to help
such a case, and allows the users to override it with a new option,
"--no-expand-tabs".
* "git send-email" now uses a more readable timestamps when
formulating a message ID.
* "git rerere" can encounter two or more files with the same conflict
signature that have to be resolved in different ways, but there was
no way to record these separate resolutions.
* "git p4" learned to record P4 jobs in Git commit that imports from
the history in Perforce.
* "git describe --contains" often made a hard-to-justify choice of
tag to name a given commit, because it tried to come up
with a name with smallest number of hops from a tag, causing an old
commit whose close descendant that is recently tagged were not
described with respect to an old tag but with a newer tag. It did
not help that its computation of "hop" count was further tweaked to
penalize being on a side branch of a merge. The logic has been
updated to favor using the tag with the oldest tagger date, which
is a lot easier to explain to the end users: "We describe a commit
in terms of the (chronologically) oldest tag that contains the
commit."
* "git clone" learned the "--shallow-submodules" option.
* HTTP transport clients learned to throw extra HTTP headers at the
server, specified via http.extraHeader configuration variable.
* The "--compaction-heuristic" option to "git diff" family of
commands enables a heuristic to make the patch output more readable
by using a blank line as a strong hint that the contents before and
after it belong to logically separate units. It is still
experimental.
* A new configuration variable core.hooksPath allows customizing
where the hook directory is.
* An earlier addition of "sanitize_submodule_env" with 14111fc4 (git:
submodule honor -c credential.* from command line, 2016-02-29)
turned out to be a convoluted no-op; implement what it wanted to do
correctly, and stop filtering settings given via "git -c var=val".
* "git commit --dry-run" reported "No, no, you cannot commit." in one
case where "git commit" would have allowed you to commit, and this
improves it a little bit ("git commit --dry-run --short" still does
not give you the correct answer, for example). This is a stop-gap
measure in that "commit --short --dry-run" still gives an incorrect
result.
* The experimental "multiple worktree" feature gains more safety to
forbid operations on a branch that is checked out or being actively
worked on elsewhere, by noticing that e.g. it is being rebased.
* "git format-patch" learned a new "--base" option to record what
(public, well-known) commit the original series was built on in
its output.
* "git commit" learned to pay attention to the "commit.verbose"
configuration variable and act as if the "--verbose" option
was given from the command line.
* Updated documentation gives hints to GMail users with two-factor
auth enabled that they need app-specific-password when using
"git send-email".
* The manpage output of our documentation did not render well in
terminal; typeset literals in bold by default to make them stand
out more.
* The mark-up in the top-level README.md file has been updated to
typeset CLI command names differently from the body text.
Performance, Internal Implementation, Development Support etc.
* The embedded args argv-array in the child process is used to build
the command line to run pack-objects instead of using a separate
array of strings.
* A test for tags has been restructured so that more parts of it can
easily be run on a platform without a working GnuPG.
* The startup_info data, which records if we are working inside a
repository (among other things), are now uniformly available to Git
subcommand implementations, and Git avoids attempting to touch
references when we are not in a repository.
* The command line argument parser for "receive-pack" has been
rewritten to use parse-options.
* A major part of "git submodule update" has been ported to C to take
advantage of the recently added framework to run download tasks in
parallel. Other updates to "git submodule" that move pieces of
logic to C continues.
* Rename bunch of tests on "git clone" for better organization.
* The tests that involve running httpd leaked the system-wide
configuration in /etc/gitconfig to the tested environment.
* Build updates for MSVC.
* The repository set-up sequence has been streamlined (the biggest
change is that there is no longer git_config_early()), so that we
do not attempt to look into refs/* when we know we do not have a
Git repository.
* Code restructuring around the "refs" API to prepare for pluggable
refs backends.
* Sources to many test helper binaries and the generated helpers
have been moved to t/helper/ subdirectory to reduce clutter at the
top level of the tree.
* Unify internal logic between "git tag -v" and "git verify-tag"
commands by making one directly call into the other.
* "merge-recursive" strategy incorrectly checked if a path that is
involved in its internal merge exists in the working tree.
* The test scripts for "git p4" (but not "git p4" implementation
itself) has been updated so that they would work even on a system
where the installed version of Python is python 3.
* As nobody maintains our in-tree git.spec.in and distros use their
own spec file, we stopped pretending that we support "make rpm".
* Move from "unsigned char[20]" to "struct object_id" continues.
* The code for warning_errno/die_errno has been refactored and a new
error_errno() reporting helper is introduced.
(merge 1da045f nd/error-errno later to maint).
* Running tests with '-x' option to trace the individual command
executions is a useful way to debug test scripts, but some tests
that capture the standard error stream and check what the command
said can be broken with the trace output mixed in. When running
our tests under "bash", however, we can redirect the trace output
to another file descriptor to keep the standard error of programs
being tested intact.
* t0040 had too many unnecessary repetitions in its test data. Teach
test-parse-options program so that a caller can tell what it
expects in its output, so that these repetitions can be cleaned up.
* Add perf test for "rebase -i".
* Common mistakes when writing gitlink: in our documentation are
found by "make check-docs".
* t9xxx series has been updated primarily for readability, while
fixing small bugs in it. A few scripted Porcelain commands have
also been updated to fix possible bugs around their use of
"test -z" and "test -n".
* CI test was taught to run git-svn tests.
* "git cat-file --batch-all" has been sped up, by taking advantage
of the fact that it does not have to read a list of objects, in two
ways.
* test updates to make it more readable and maintainable.
(merge e6273f4 es/t1500-modernize later to maint).
* "make DEVELOPER=1" worked as expected; setting DEVELOPER=1 in
config.mak didn't.
(merge 51dd3e8 mm/makefile-developer-can-be-in-config-mak later to maint).
* The way how "submodule--helper list" signals unmatch error to its
callers has been updated.
* A bash-ism "local" has been removed from "git submodule" scripted
Porcelain.
Also contains various documentation updates and code clean-ups.
Fixes since v2.8
----------------
Unless otherwise noted, all the fixes since v2.8 in the maintenance
track are contained in this release (see the maintenance releases'
notes for details).
* "git config --get-urlmatch", unlike other variants of the "git
config --get" family, did not signal error with its exit status
when there was no matching configuration.
* The "--local-env-vars" and "--resolve-git-dir" options of "git
rev-parse" failed to work outside a repository when the command's
option parsing was rewritten in 1.8.5 era.
* "git index-pack --keep[=<msg>] pack-$name.pack" simply did not work.
* Fetching of history by naming a commit object name directly didn't
work across remote-curl transport.
* A small memory leak in an error codepath has been plugged in xdiff
code.
* strbuf_getwholeline() did not NUL-terminate the buffer on certain
corner cases in its error codepath.
* "git mergetool" did not work well with conflicts that both sides
deleted.
* "git send-email" had trouble parsing alias file in mailrc format
when lines in it had trailing whitespaces on them.
* When "git merge --squash" stopped due to conflict, the concluding
"git commit" failed to read in the SQUASH_MSG that shows the log
messages from all the squashed commits.
* "git merge FETCH_HEAD" dereferenced NULL pointer when merging
nothing into an unborn history (which is arguably unusual usage,
which perhaps was the reason why nobody noticed it).
* When "git worktree" feature is in use, "git branch -d" allowed
deletion of a branch that is checked out in another worktree,
which was wrong.
* When "git worktree" feature is in use, "git branch -m" renamed a
branch that is checked out in another worktree without adjusting
the HEAD symbolic ref for the worktree.
* "git diff -M" used to work better when two originally identical
files A and B got renamed to X/A and X/B by pairing A to X/A and B
to X/B, but this was broken in the 2.0 timeframe.
* "git send-pack --all <there>" was broken when its command line
option parsing was written in the 2.6 timeframe.
* "git format-patch --help" showed `-s` and `--no-patch` as if these
are valid options to the command. We already hide `--patch` option
from the documentation, because format-patch is about showing the
diff, and the documentation now hides these options as well.
* When running "git blame $path" with unnormalized data in the index
for the path, the data in the working tree was blamed, even though
"git add" would not have changed what is already in the index, due
to "safe crlf" that disables the line-end conversion. It has been
corrected.
* A change back in version 2.7 to "git branch" broke display of a
symbolic ref in a non-standard place in the refs/ hierarchy (we
expect symbolic refs to appear in refs/remotes/*/HEAD to point at
the primary branch the remote has, and as .git/HEAD to point at the
branch we locally checked out).
* A partial rewrite of "git submodule" in the 2.7 timeframe changed
the way the gitdir: pointer in the submodules point at the real
repository location to use absolute paths by accident. This has
been corrected.
* "git commit" misbehaved in a few minor ways when an empty message
is given via -m '', all of which has been corrected.
* Support for CRAM-MD5 authentication method in "git imap-send" did
not work well.
* Upcoming OpenSSL 1.1.0 will break compilation by updating a few API
elements we use in imap-send, which has been adjusted for the change.
* The socks5:// proxy support added back in 2.6.4 days was not aware
that socks5h:// proxies behave differently from socks5:// proxies.
* "git config" had a codepath that tried to pass a NULL to
printf("%s"), which nobody seems to have noticed.
* On Cygwin, object creation uses the "create a temporary and then
rename it to the final name" pattern, not "create a temporary,
hardlink it to the final name and then unlink the temporary"
pattern.
This is necessary to use Git on Windows shared directories, and is
already enabled for the MinGW and plain Windows builds. It also
has been used in Cygwin packaged versions of Git for quite a while.
See https://lore.kernel.org/git/20160419091055.GF2345@dinwoodie.org/
* "merge-octopus" strategy did not ensure that the index is clean
when merge begins.
* When "git merge" notices that the merge can be resolved purely at
the tree level (without having to merge blobs) and the resulting
tree happens to already exist in the object store, it forgot to
update the index, which left an inconsistent state that would
break later operations.
* "git submodule" reports the paths of submodules the command
recurses into, but these paths were incorrectly reported when
the command was not run from the root level of the superproject.
* The "user.useConfigOnly" configuration variable makes it an error
if users do not explicitly set user.name and user.email. However,
its check was not done early enough and allowed another error to
trigger, reporting that the default value we guessed from the
system setting was unusable. This was a suboptimal end-user
experience as we want the users to set user.name/user.email without
relying on the auto-detection at all.
* "git mv old new" did not adjust the path for a submodule that lives
as a subdirectory inside old/ directory correctly.
* "git replace -e" did not honour "core.editor" configuration.
* "git push" from a corrupt repository that attempts to push a large
number of refs deadlocked; the thread to relay rejection notices
for these ref updates blocked on writing them to the main thread,
after the main thread at the receiving end notices that the push
failed and decides not to read these notices and return a failure.
* mmap emulation on Windows has been optimized and work better without
consuming paging store when not needed.
* A question by "git send-email" to ask the identity of the sender
has been updated.
* UI consistency improvements for "git mergetool".
* "git rebase -m" could be asked to rebase an entire branch starting
from the root, but failed by assuming that there always is a parent
commit to the first commit on the branch.
* Fix a broken "p4 lfs" test.
* Recent update to Git LFS broke "git p4" by changing the output from
its "lfs pointer" subcommand.
* "git fetch" test t5510 was flaky while running a (forced) automagic
garbage collection.
* Documentation updates to help contributors setting up Travis CI
test for their patches.
* Some multi-byte encoding can have a backslash byte as a later part
of one letter, which would confuse "highlight" filter used in
gitweb.
* "git commit-tree" plumbing command required the user to always sign
its result when the user sets the commit.gpgsign configuration
variable, which was an ancient mistake. Rework "git rebase" that
relied on this mistake so that it reads commit.gpgsign and pass (or
not pass) the -S option to "git commit-tree" to keep the end-user
expectation the same, while teaching "git commit-tree" to ignore
the configuration variable. This will stop requiring the users to
sign commit objects used internally as an implementation detail of
"git stash".
* "http.cookieFile" configuration variable clearly wants a pathname,
but we forgot to treat it as such by e.g. applying tilde expansion.
* Consolidate description of tilde-expansion that is done to
configuration variables that take pathname to a single place.
* Correct faulty recommendation to use "git submodule deinit ." when
de-initialising all submodules, which would result in a strange
error message in a pathological corner case.
* Many 'linkgit:<git documentation page>' references were broken,
which are all fixed with this.
* "git rerere" can get confused by conflict markers deliberately left
by the inner merge step, because they are indistinguishable from
the real conflict markers left by the outermost merge which are
what the end user and "rerere" need to look at. This was fixed by
making the conflict markers left by the inner merges a bit longer.
(merge 0f9fd5c jc/ll-merge-internal later to maint).
* CI test was taught to build documentation pages.
* "git fsck" learned to catch NUL byte in a commit object as
potential error and warn.
* Portability enhancement for "rebase -i" to help platforms whose
shell does not like "for i in <empty>" (which is not POSIX-kosher).
* On Windows, .git and optionally any files whose name starts with a
dot are now marked as hidden, with a core.hideDotFiles knob to
customize this behaviour.
* Documentation for "git merge --verify-signatures" has been updated
to clarify that the signature of only the commit at the tip is
verified. Also the phrasing used for signature and key validity is
adjusted to align with that used by OpenPGP.
* A couple of bugs around core.autocrlf have been fixed.
* Many commands normalize command line arguments from NFD to NFC
variant of UTF-8 on OSX, but commands in the "diff" family did
not, causing "git diff $path" to complain that no such path is
known to Git. They have been taught to do the normalization.
* "git difftool" learned to handle unmerged paths correctly in
dir-diff mode.
* The "are we talking with TTY, doing an interactive session?"
detection has been updated to work better for "Git for Windows".
* We forgot to add "git log --decorate=auto" to documentation when we
added the feature back in v2.1.0 timeframe.
(merge 462cbb4 rj/log-decorate-auto later to maint).
* "git fast-import --export-marks" would overwrite the existing marks
file even when it makes a dump from its custom die routine.
Prevent it from doing so when we have an import-marks file but
haven't finished reading it.
(merge f4beed6 fc/fast-import-broken-marks-file later to maint).
* "git rebase -i", after it fails to auto-resolve the conflict, had
an unnecessary call to "git rerere" from its very early days, which
was spotted recently; the call has been removed.
(merge 7063693 js/rebase-i-dedup-call-to-rerere later to maint).
* Other minor clean-ups and documentation updates
(merge cd82b7a pa/cherry-pick-doc-typo later to maint).
(merge 2bb73ae rs/patch-id-use-skip-prefix later to maint).
(merge aa20cbc rs/apply-name-terminate later to maint).
(merge fe17fc0 jc/t2300-setup later to maint).
(merge e256eec jk/shell-portability later to maint).