Git 2.4.5/2.5.0 RC0 发布

* The setup code used to die when core.bare and core.worktree are set

inconsistently, even for commands that do not need working tree.

* There was a dead code that used to handle "git pull --tags" and

show special-cased error message

* "color.diff.plain" was a misnomer; give it 'color.diff.context' as

a more logical synonym.

* The configuration reader/writer uses mmap(2) interface to access

the files; when we find a directory, it barfed with "Out of memory?".

* Recent "git prune" traverses young unreachable objects to safekeep

old objects in the reachability chain from them, which sometimes

showed unnecessary error messages that are alarming.

* "git rebase -i" fired post-rewrite hook when it shouldn't (namely,

when it was told to stop sequencing with 'exec' insn).

同时还包含常规修复，文档更新和代码清理。

UI, Workflows & Features

* The bash completion script (in contrib/) learned a few options that

"git revert" takes.

* Whitespace breakages in deleted and context lines can also be

painted in the output of "git diff" and friends with the new

--ws-error-highlight option.

* List of commands shown by "git help" are grouped along the workflow

elements to help early learners.

* "git p4" now detects the filetype (e.g. binary) correctly even when

the files are opened exclusively.

* git p4 attempts to better handle branches in Perforce.

* "git p4" learned "--changes-block-size <n>" to read the changes in

chunks from Perforce, instead of making one call to "p4 changes"

that may trigger "too many rows scanned" error from Perforce.

* More workaround for Perforce's row number limit in "git p4".

* Unlike "$EDITOR" and "$GIT_EDITOR" that can hold the path to the

command and initial options (e.g. "/path/to/emacs -nw"), 'git p4'

did not let the shell interpolate the contents of the environment

variable that name the editor "$P4EDITOR" (and "$EDITOR", too).

This release makes it in line with the rest of Git, as well as with

Perforce.

* A new short-hand <branch>@{push} denotes the remote-tracking branch

that tracks the branch at the remote the <branch> would be pushed

to.

* "git show-branch --topics HEAD" (with no other arguments) did not

do anything interesting.  Instead, contrast the given revision

against all the local branches by default.

* A replacement for contrib/workdir/git-new-workdir that does not

rely on symbolic links and make sharing of objects and refs safer

by making the borrowee and borrowers aware of each other.

Consider this as still an experimental feature; the UI will likely

to change.

* Tweak the sample "store" backend of the credential helper to honor

XDG configuration file locations when specified.

* A heuristic we use to catch mistyped paths on the command line

"git <cmd> <revs> <pathspec>" is to make sure that all the non-rev

parameters in the later part of the command line are names of the

files in the working tree, but that means "git grep $str -- /*.c"

must always be disambiguated with "--", because nobody sane will

create a file whose name literally is asterisk-dot-see.  Loosen the

heuristic to declare that with a wildcard string the user likely

meant to give us a pathspec.

* "git merge FETCH_HEAD" learned that the previous "git fetch" could

be to create an Octopus merge, i.e. recording multiple branches

that are not marked as "not-for-merge"; this allows us to lose an

old style invocation "git merge <msg> HEAD $commits..." in the

implementation of "git pull" script; the old style syntax can now

be deprecated (but not removed yet).

* Filter scripts were run with SIGPIPE disabled on the Git side,

expecting that they may not read what Git feeds them to filter.

We however treated a filter that does not read its input fully

before exiting as an error.  We no longer do and ignore EPIPE

when writing to feed the filter scripts.

This changes semantics, but arguably in a good way.  If a filter

can produce its output without fully consuming its input using

whatever magic, we now let it do so, instead of diagnosing it

as a programming error.

* Instead of dying immediately upon failing to obtain a lock, the

locking (of refs etc) retries after a short while with backoff.

* Introduce http.<url>.SSLCipherList configuration variable to tweak

the list of cipher suite to be used with libcURL when talking with

https:// sites.

* "git subtree" script (in contrib/) used "echo -n" to produce

progress messages in a non-portable way.

* "git subtree" script (in contrib/) does not have --squash option

when pushing, but the documentation and help text pretended as if

it did.

* The Git subcommand completion (in contrib/) no longer lists credential

helpers among candidates; they are not something the end user would

invoke interactively.

* The index file can be taught with "update-index --untracked-cache"

to optionally remember already seen untracked files, in order to

speed up "git status" in a working tree with tons of cruft.

* "git mergetool" learned to drive WinMerge as a backend.

* "git upload-pack" that serves "git fetch" can be told to serve

commits that are not at the tip of any ref, as long as they are

reachable from a ref, with uploadpack.allowReachableSHA1InWant

configuration variable.

* "git cat-file --batch(-check)" learned the "--follow-symlinks"

option that follows an in-tree symbolic link when asked about an

object via extended SHA-1 syntax, e.g. HEAD:RelNotes that points at

Documentation/RelNotes/2.5.0.txt.  With the new option, the command

behaves as if HEAD:Documentation/RelNotes/2.5.0.txt was given as

input instead.

Consider this as still an experimental and incomplete feature:

- We may want to do the same for in-index objects, e.g.

asking for :RelNotes with this option should give

:Documentation/RelNotes/2.5.0.txt, too

- "git cat-file --follow-symlinks blob HEAD:RelNotes"

may also be something we want to allow in the future.

* "git send-email" learned the alias file format used by the sendmail

program (in a simplified form; we obviously do not feed pipes).

* "git am" learned am.threeWay configuration variable.

* Traditionally, external low-level 3-way merge drivers are expected

to produce their results based solely on the contents of the three

variants given in temporary files named by %O, %A and %B on their

command line.  Additionally allow them to look at the final path

(given by %P).

* "git blame" learned blame.showEmail configuration variable.

* "git apply" cannot diagnose a patch corruption when the breakage is

to mark the length of the hunk shorter than it really is on the

hunk header line "@@ -l,k +m,n @@"; one special case it could is

when the hunk becomes no-op (e.g. k == n == 2 for two-line context

patch output), and it learned to do so in this special case.

* Add the "--allow-unknown-type" option to "cat-file" to allow

inspecting loose objects of an experimental or a broken type.

* Many long-running operations show progress eye-candy, even when

they are later backgrounded.  Hide the eye-candy when the process

is sent to the background instead.

(merge a4fb76c lm/squelch-bg-progress later to maint).

Performance, Internal Implementation, Development Support etc.

* "unsigned char [20]" used throughout the code to represent object

names are being converted into a semi-opaque "struct object_id".

This effort is expected to interfere with other topics in flight,

but hopefully will give us one extra level of abstraction in the

end, when completed.

* for_each_ref() callback functions were taught to name the objects

not with "unsigned char sha1[20]" but with "struct object_id".

* Catch a programmer mistake to feed a pointer not an array to

ARRAY_SIZE() macro, by using a couple of GCC extensions.

* Some error messages in "git config" were emitted without calling

the usual error() facility.

* When "add--interactive" splits a hunk into two overlapping hunks

and then let the user choose only one, it sometimes feeds an

incorrect patch text to "git apply".  Add tests to demonstrate

this.

I have a slight suspicion that this may be $gmane/87202 coming back

and biting us (I seem to have said "let's run with this and see

what happens" back then).

* More line-ending tests.

* An earlier rewrite to use strbuf_getwholeline() instead of fgets(3)

to read packed-refs file revealed that the former is unacceptably

inefficient.  It has been optimized by using getdelim(3) when

available.

* The refs API uses ref_lock struct which had its own "int fd", even

though the same file descriptor was in the lock struct it contains.

Clean-up the code to lose this redundant field.

* There was a dead code that used to handle "git pull --tags" and

show special-cased error message, which was made irrelevant when

the semantics of the option changed back in Git 1.9 days.

(merge 19d122b pt/pull-tags-error-diag later to maint).

* Help us to find broken test script that splits the body part of the

test by mistaken use of wrong kind of quotes.

(merge d93d5d5 jc/test-prereq-validate later to maint).

* Developer support to automatically detect broken &&-chain in the

test scripts is now turned on by default.

(merge 92b269f jk/test-chain-lint later to maint).

* Error reporting mechanism used in "refs" API has been made more

consistent.

* "git pull" has more test coverage now.

* "git pull" has become more aware of the options meant for

underlying "git fetch" and then learned to use parse-options

parser.

此外还有各种文档更新和代码清理。详细改进请看 更新日志 。