Back when the feature to use different HTTP authentication methods was
originally written, it needed an extra HTTP request for everything when
the feature was in effect, because we didn't reuse curl sessions.
However, b8ac923 (Add an option for using any HTTP authentication scheme,
not only basic, 2009-11-27) builds on top of an updated codebase that does
reuse curl sessions; there is no need to manually avoid the extra overhead
by making this configurable anymore.
Acked-by: Martin Storsjo <martin@martin.st> Acked-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Documentation: always respect core.worktree if set
The value of core.worktree in a ".git/config" is honored even when it
differs from the directory that has the ".git" directory as its
subdirectory. This is likely to be a misconfiguration, so warn users
about it. Also, drop the part of the documentation that incorrectly
claimed that we ignore such a misconfigured value.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Follow the argument convention of git-pack-objects, such that a
separate option (--preogress) is used to force progress reporting
instead of -v/--verbose.
-v/--verbose now does not force progress reporting. Make git-clone.txt
say so.
This should cover all the bases in 21188b1 (Implement git clone -v),
which implemented the option to force progress reporting.
Signed-off-by: Tay Ray Chuan <rctay89@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
check stderr with isatty() instead of stdout when deciding to show progress
Make transport code (viz. transport.c::fetch_refs_via_pack() and
transport-helper.c::standard_options()) that decides to show progress
check if stderr is a terminal, instead of stdout. After all, progress
reports (via the API in progress.[ch]) are sent to stderr.
Update the documentation for git-clone to say "standard error" as well.
Signed-off-by: Tay Ray Chuan <rctay89@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Previously, remote archive required internal (non remote-helper)
smart transport. Extend the remote archive to also support smart
transports implemented by remote helpers.
Signed-off-by: Ilari Liusvaara <ilari.liusvaara@elisanet.fi> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jk/1.7.0-status:
status/commit: do not suggest "reset HEAD <path>" while merging
commit/status: "git add <path>" is not necessarily how to resolve
commit/status: check $GIT_DIR/MERGE_HEAD only once
t7508-status: test all modes with color
t7508-status: status --porcelain ignores relative paths setting
status: reduce duplicated setup code
status: disable color for porcelain format
status -s: obey color.status
builtin-commit: refactor short-status code into wt-status.c
t7508-status.sh: Add tests for status -s
status -s: respect the status.relativePaths option
docs: note that status configuration affects only long format
commit: support alternate status formats
status: add --porcelain output format
status: refactor format option parsing
status: refactor short-mode printing to its own function
status: typo fix in usage
git status: not "commit --dry-run" anymore
git stat -s: short status output
git stat: the beginning of "status that is not a dry-run of commit"
read_index(): fix reading extension size on BE 64-bit archs
On big endian platforms with 8-byte unsigned long, the code reads the
size of the index extension section (which is a 4-byte network byte
order integer) incorrectly.
* maint:
Makefile: FreeBSD (both 7 and 8) needs OLD_ICONV
Start 1.6.6.X maintenance track
Add git-http-backend to command-list.
t4019 "grep" portability fix
t1200: work around a bug in some implementations of "find"
* jc/1.7.0-diff-whitespace-only-status:
diff.c: fix typoes in comments
Make test case number unique
diff: Rename QUIET internal option to QUICK
diff: change semantics of "ignore whitespace" options
* sr/vcs-helper:
tests: handle NO_PYTHON setting
builtin-push: don't access freed transport->url
Add Python support library for remote helpers
Basic build infrastructure for Python scripts
Allow helpers to report in "list" command that the ref is unchanged
Fix various memory leaks in transport-helper.c
Allow helper to map private ref names into normal names
Add support for "import" helper command
Allow specifying the remote helper in the url
Add a config option for remotes to specify a foreign vcs
Allow fetch to modify refs
Use a function to determine whether a remote is valid
Allow programs to not depend on remotes having urls
Fix memory leak in helper method for disconnect
Input to "grep" is supposed to be "text", but we deliberately feed output
from "git diff --color" to sift it into two sets of lines (ones with
errors, the other without). Some implementations of "grep" only report
matches with the exit status, without showing the matched lines in their
output (e.g. OpenBSD 4.6, which says "Binary file .. matches").
Fortunately, "grep -a" is often a way to force the command to treat its
input as text.
t1200: work around a bug in some implementations of "find"
"find path ..." command should exit with zero status only when all path
operands were traversed successfully. When a non-existent path is given,
however, some implementations of "find" (e.g. OpenBSD 4.6) exit with zero
status and break the last test in t1200.
Rewrite the test to check that there is no regular files in the objects
fan-out directories to work around this bug; it is closer to what we are
testing anyway.
Ever since 658f365 (Make git-rerere a builtin, 2006-12-20) rewrote it, it
kept this line-length limit regression, even after we started using strbuf
in the same function in 19b358e (Use strbuf API in buitin-rerere.c,
2007-09-06).
resolve-undo: teach "update-index --unresolve" to use resolve-undo info
The update-index plumbing command had a hacky --unresolve implementation
that was written back in the days when merge was the only way for users to
end up with higher stages in the index, and assumed that stage #2 must
have come from HEAD, stage #3 from MERGE_HEAD and didn't bother to compute
the stage #1 information.
There were several issues with this approach:
- These days, merge is not the only command, and conflicts coming from
commands like cherry-pick, "am -3", etc. cannot be recreated by looking
at MERGE_HEAD;
- For a conflict that came from a merge that had renames, picking up the
same path from MERGE_HEAD and HEAD wouldn't help recreating it, either;
- It may have been Ok not to recreate stage #1 back when it was written,
because "diff --ours/--theirs" were the only availble ways to review
conflicts and they don't need stage #1 information. "diff --cc" that
was invented much later is a lot more useful way but it needs stage #1.
We can use resolve-undo information recorded in the index extension to
solve all of these issues.
resolve-undo: "checkout -m path" uses resolve-undo information
Once you resolved conflicts by "git add path", you cannot recreate the
conflicted state with "git checkout -m path", because you lost information
from higher stages in the index when you resolved them.
Since we record the necessary information in the resolve-undo index
extension these days, we can reproduce the unmerged state in the index and
check it out.
resolve-undo: allow plumbing to clear the information
At the Porcelain level, operations such as merge that populate an
initially cleanly merged index with conflicted entries clear the
resolve-undo information upfront. Give scripted Porcelains a way
to do the same, by implementing "update-index --clear-resolve-info".
With this, a scripted Porcelain may "update-index --clear-resolve-info"
first and repeatedly run "update-index --cacheinfo" to stuff unmerged
entries to the index, to be resolved by the user with "git add" and
stuff.
Make sure that resolving a failed merge with git add records
the conflicted state, committing the result keeps that state,
and checking out another commit clears the state.
"git ls-files" learns a new option --resolve-undo to show the
recorded information.
resolve-undo: record resolved conflicts in a new index extension section
When resolving a conflict using "git add" to create a stage #0 entry, or
"git rm" to remove entries at higher stages, remove_index_entry_at()
function is eventually called to remove unmerged (i.e. higher stage)
entries from the index. Introduce a "resolve_undo_info" structure and
keep track of the removed cache entries, and save it in a new index
extension section in the index_state.
Operations like "read-tree -m", "merge", "checkout [-m] <branch>" and
"reset" are signs that recorded information in the index is no longer
necessary. The data is removed from the index extension when operations
start; they may leave conflicted entries in the index, and later user
actions like "git add" will record their conflicted states afresh.
Instead of using the low-level index_state interface, use the bog standard
active_cache and active_nr macros to access the cache entries when using the
default one.
git svn: branch/tag commands detect username in URLs
svn+ssh:// repositories often have userinfo embedded in the URL
which were stripped out of the "git-svn-id:" trailers. Since
the SVN::Client::copy function takes userinfo into account when
matching URLs for SVN repositories, we need to retrieve the full
URL with embedded userinfo in it to avoid mismatched URLs.
Tested-by: Florian Köberle <florian@fkoeberle.de> Signed-off-by: Eric Wong <normalperson@yhbt.net>
Prevent git blame from segfaulting on a missing author name
The human-readable author and committer name can be missing from
commits imported from foreign SCM interfaces. Make sure we parse
the "author" and "committer" line a bit more leniently and avoid
segfaulting by assuming the name always exists.
Signed-off-by: David Reiss <dreiss@facebook.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change git-svn not to impose a limit of 16 parents on a merge.
This limit in git-svn artificially prevents cloning svn repositories
that contain commits with more than 16 merge parents.
The limit was removed from builtin-commit-tree.c for git v1.6.0 in commit ef98c5cafb3e799b1568bb843fcd45920dc62f16, so there is no need to check for it
it in git-svn.
Signed-off-by: Andrew Myrick <amyrick@apple.com> Acked-by: Eric Wong <normalperson@yhbt.net>
The old function was incorrect; in some instances it marks a cherry picked
range as a merged branch (because of an incorrect assumption that
'rev-list COMMIT --not RANGE' would work). This is replaced with a
function which should detect them correctly, memoized to limit the expense
of dealing with branches with many cherry picks to one 'merge-base' call
per merge, per branch which used cherry picking.
Signed-off-by: Sam Vilain <sam@vilain.net> Acked-by: Eric Wong <normalperson@yhbt.net>
git-svn: exclude already merged tips using one rev-list call
The old function would have to check all mentioned merge tips, every time
that the mergeinfo ticket changed. This involved 1-2 rev-list operation
for each listed mergeinfo line. If there are a lot of feature branches
being merged into a trunk, this makes for a very expensive operation for
detecting the new parents on every merge.
This new version first uses a single 'rev-list' to figure out which commit
ranges are already reachable from the parents. This is used to eliminate
the already merged branches from the list.
Signed-off-by: Sam Vilain <sam@vilain.net> Acked-by: Eric Wong <normalperson@yhbt.net>
git-svn: fix some mistakes with interpreting SVN mergeinfo commit ranges
SVN's list of commit ranges in mergeinfo tickets is inclusive, whereas
git commit ranges are exclusive on the left hand side. Also, the end
points of the commit ranges may not exist; they simply delineate
ranges of commits which may or may not exist. Fix these two mistakes.
Signed-off-by: Sam Vilain <sam@vilain.net> Acked-by: Eric Wong <normalperson@yhbt.net>
git-svn: memoize conversion of SVN merge ticket info to git commit ranges
Each time the svn mergeinfo ticket changes, we look it up in the rev_map;
when there are a lot of merged branches, this will result in many repeated
lookups of the same information for subsequent commits. Arrange the slow
part of the function so that it may be memoized, and memoize it. The more
expensive revision walking operation can be memoized separately.
[ew: changed "next" to "return" for function exit]
Signed-off-by: Sam Vilain <sam@vilain.net> Acked-by: Eric Wong <normalperson@yhbt.net>
git-svn: expand the svn mergeinfo test suite, highlighting some failures
As shown, git-svn has some problems; not all svn merges are correctly
detected, and cherry picks may incorrectly be detected as real merges.
These test cases will be marked as _success once the relevant fixes are in.
Signed-off-by: Sam Vilain <sam@vilain.net> Acked-by: Eric Wong <normalperson@yhbt.net>
git svn: fix --revision when fetching deleted paths
When using the -r/--revision argument to fetch deleted history,
calling SVN::Ra::get_log() from an SVN::Ra object initialized
to track the deleted URL will fail.
We now ignore errors from SVN::Ra::get_log() here because using
--revision will always override the value of $head here if
(and only if) we're tracking deleted directories.
* maint:
rebase -i: abort cleanly if the editor fails to launch
technical-docs: document hash API
api-strbuf.txt: fix typos and document launch_editor()
rebase -i: abort cleanly if the editor fails to launch
If the user's configured editor is emacsclient, the editor
will fail to launch if emacs is not running and the git
command that tried to lanuch the editor will abort. For most
commands, all you have to do is to start emacs and repeat
the command.
The "git rebase -i" command, however, aborts without cleaning
the "$GIT_DIR/rebase-merge" directory if it fails to launch the
editor, so you'll need to do "git rebase --abort" before
repeating the rebase command.
Change "git rebase -i" to terminate using "die_abort" (instead of
with "die") if the initial launch of the editor fails.
Signed-off-by: Björn Gustavsson <bgustavsson@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* maint:
Git 1.6.5.7
worktree: don't segfault with an absolute pathspec without a work tree
ignore unknown color configuration
help.autocorrect: do not run a command if the command given is junk
Illustrate "filter" attribute with an example
worktree: don't segfault with an absolute pathspec without a work tree
If a command is run with an absolute path as a pathspec inside a bare
repository, e.g. "rev-list HEAD -- /home", the code tried to run strlen()
on NULL, which is the result of get_git_work_tree(), and segfaulted. It
should just fail instead.
Currently the function returns NULL even inside .git/ in a repository
with a work tree, but that is a separate issue.
When parsing the config file, if there is a value that is
syntactically correct but unused, we generally ignore it.
This lets non-core porcelains store arbitrary information in
the config file, and it means that configuration files can
be shared between new and old versions of git (the old
versions might simply ignore certain configuration).
The one exception to this is color configuration; if we
encounter a color.{diff,branch,status}.$slot variable, we
die if it is not one of the recognized slots (presumably as
a safety valve for user misconfiguration). This behavior
has existed since 801235c (diff --color: use
$GIT_DIR/config, 2006-06-24), but hasn't yet caused a
problem. No porcelain has wanted to store extra colors, and
we once a color area (like color.diff) has been introduced,
we've never changed the set of color slots.
However, that changed recently with the addition of
color.diff.func. Now a user with color.diff.func in their
config can no longer freely switch between v1.6.6 and older
versions; the old versions will complain about the existence
of the variable.
This patch loosens the check to match the rest of
git-config; unknown color slots are simply ignored. This
doesn't fix this particular problem, as the older version
(without this patch) is the problem, but it at least
prevents it from happening again in the future.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
help.autocorrect: do not run a command if the command given is junk
If a given command is not found, then help.c tries to guess which one the
user could have meant. If help.autocorrect is 0 or unset, then a list of
suggestions is given as long as the dissimilarity between the given command
and the candidates is not excessively high. But if help.autocorrect was
non-zero (i.e., a delay after which the command is run automatically), the
latter restriction on dissimilarity was not obeyed.
In my case, this happened:
$ git ..daab02
WARNING: You called a Git command named '..daab02', which does not exist.
Continuing under the assumption that you meant 'read-tree'
in 4.0 seconds automatically...
The patch reuses the similarity limit that is also applied when the list of
suggested commands is printed.
Signed-off-by: Johannes Sixt <j6t@kdbg.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
filter-branch: remove an unnecessary use of 'git read-tree'
The intent of this particular call to 'git read-tree' was to fill an
index. But in fact, it only allocated an empty index. Later in the
program, the index is filled anyway by calling read-tree with specific
commits, and considering that elsewhere the index is even removed (i.e.,
it is not relied upon that the index file exists), this first call of
read-tree is completely redundant.
Signed-off-by: Johannes Sixt <j6t@kdbg.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit b4d1690 (Teach Git to respect skip-worktree bit (reading part))
fails to make "git commit -- a b c" respect skip-worktree
(i.e. not committing paths that are skip-worktree). This is because
when the index is reset back to HEAD, all skip-worktree information is
gone.
This patch saves skip-worktree information in the string list of
committed paths, then reuse it later on to skip skip-worktree paths.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
ie_match_stat(): do not ignore skip-worktree bit with CE_MATCH_IGNORE_VALID
Previously CE_MATCH_IGNORE_VALID flag is used by both valid and
skip-worktree bits. While the two bits have similar behaviour, sharing
this flag means "git update-index --really-refresh" will ignore
skip-worktree while it should not. Instead another flag is
introduced to ignore skip-worktree bit, CE_MATCH_IGNORE_VALID only
applies to valid bit.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The fast-forward logic is never being triggered because $common and
$MRC are never equivalent. $common is initialized to a commit id by
merge-base and MRC is initialized to HEAD. Fix this by initializing
$MRC to the commit id for HEAD so that its possible for $MRC and
$common to be equal.
Signed-off-by: Stephen Boyd <bebarino@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Its not very easy to understand what heads are being merged given
the current output of an octopus merge. Fix this by replacing the
sha1 with the (usually) better description in GITHEAD_<SHA1>.
Suggested-by: Jari Aalto <jari.aalto@cante.net> Signed-off-by: Stephen Boyd <bebarino@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Replace refs are useful to change some git objects after they
have started to be shared between different repositories. One
might want to ignore them to see the original state, and
"--no-replace-objects" option can be used from the command
line to do so.
This option simply sets the GIT_NO_REPLACE_OBJECTS environment
variable, and that is enough to make gitk ignore replace refs.
The GIT_NO_REPLACE_OBJECTS is set to "1" instead of "" as it is
safer on some platforms, thanks to Johannes Sixt and Michael J
Gruber.
Tested-by: Michael J Gruber <git@drmicha.warpmail.net> Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
Support the new options --all, --prune, and --dry-run for
'git fetch'.
As the --multiple option was primarily introduced to enable
'git remote update' to be re-implemented in terms of 'git fetch'
(16679e37) and is not likely to be used much from the command
line, it does not seems worthwhile to complicate the code
(to support completion of multiple remotes) to handle it.
Signed-off-by: Björn Gustavsson <bgustavsson@gmail.com> Acked-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
status/commit: do not suggest "reset HEAD <path>" while merging
Suggesting "'reset HEAD <path>' to unstage" is dead wrong if we are about
to record a merge commit. For either an unmerged path (i.e. with
unresolved conflicts), or an updated path, it would result in discarding
what the other branch did.
Note that we do not do anything special in a case where we are amending a
merge. The user is making an evil merge starting from an already
committed merge, and running "reset HEAD <path>" is the right way to get
rid of the local edit that has been added to the index.
Once "reset --unresolve <path>" becomes available, we might want to
suggest it for a merged path that has unresolve information, but until
then, just remove the incorrect advice.
We might also want to suggest "checkout --conflict <path>" to revert the
file in the work tree to the state of failed automerge for an unmerged
path, but we never did that, and this commit does not change that.
commit/status: "git add <path>" is not necessarily how to resolve
When the desired resolution is to remove the path, "git rm <path>" is the
command the user needs to use. Just like in "Changed but not updated"
section, suggest to use "git add/rm" as appropriate.
commit/status: check $GIT_DIR/MERGE_HEAD only once
The code checked for the MERGE_HEAD file to see if we were about
to commit a merge twice in the codepath; also one of them used a
variable merge_head_sha1[] which was set but was never used.
Just check it once, but do so also in "git status", too, as
we will be using this for status generation in the next patch.
Print the sha1 of the deleted tag (in addition to the tag name) so that
one can easily recreate a mistakenly deleted tag:
git tag -d tagname
Deleted tag 'tagname' (was DEADBEEF)
git tag 'tagname' DEADBEEF
We output the previous ref also in the case of forcefully overwriting
tags.
Signed-off-by: Michael J Gruber <git@drmicha.warpmail.net> Suggested-by: Jari Aalto <jari.aalto@cante.net> Helped-by: Björn Steinbrink <B.Steinbrink@gmx.de> Helped-by: Jeff King <peff@peff.net> Helped-by: Zoltán Füzesi <zfuzesi@eaglet.hu> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Giving --format from the command line, or using output file extention to
DWIM the output format, with a pathspec that is disambiguated with an
explicit double-dash on the command line, e.g.
git archive -o file --format=zip HEAD -- path
git archive -o file.zip HEAD -- path
didn't work correctly.
This was because the code reordered (when one was given) or added (when
the format was inferred) a --format argument at the end, effectively
making it to "archive HEAD -- path --format=zip", i.e. an extra pathspec
that is unlikely to match anything.
The command line argument list should always be "options, revs and then
paths", and we should set a good example by inserting the --format at the
beginning instead.
Reported-by: Ilari Liusvaara <ilari.liusvaara@elisanet.fi> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* git://repo.or.cz/git-gui:
git-gui: suppress RenderBadPicture X error caused by Tk bug
git-gui: Increase blame viewer usability on MacOS.
git-gui: search 4 directories to improve statistic of gc hint
git gui: make current branch default in "remote delete branch" merge check
* git://git.kernel.org/pub/scm/gitk/gitk:
gitk: Fix selection of tags
gitk: Default to the system colours on Windows
gitk: Update Japanese translation
gitk: Fix "git gui blame" invocation when called from top-level directory
gitk: Disable checkout of remote branches
gitk: Improve appearance of radiobuttons and checkbuttons
gitk: Skip translation of "wrong Tcl version" message
gitk: Add Japanese translation
gitk: Use the --submodule option for displaying diffs when available
gitk: Fix diffing committed -> staged (typo in diffcmd)
gitk: Add configuration for UI colour scheme
gitk: Don't compare fake children when comparing commits
gitk: Show diff of commits at end of compare-commits output
gitk: Add a user preference to enable/disable use of themed widgets
gitk: Fix errors in the theme patch
gitk: Use themed tk widgets
gitk: Restore scrolling position of diff pane on back/forward in history
Refactor the transport options parsing so that protocols that aren't
directly smart transports (file://, git://, ssh:// & co) can record
the smart transport options for the case if it turns that transport
can actually be smart.
Signed-off-by: Ilari Liusvaara <ilari.liusvaara@elisanet.fi> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Pass unknown protocols to external protocol handlers
Change URL handling to allow external protocol handlers to implement
new protocols without the '::' syntax if helper name does not conflict
with any built-in protocol.
foo:// now invokes git-remote-foo with foo:// as the URL.
Signed-off-by: Ilari Liusvaara <ilari.liusvaara@elisanet.fi> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add support for marking capability as mandatory for hosting git version
to understand. This is useful for helpers which require various types
of assistance from main git binary.
Signed-off-by: Ilari Liusvaara <ilari.liusvaara@elisanet.fi> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Merge branch 'mm/maint-merge-ff-error-message-fix' into maint
* mm/maint-merge-ff-error-message-fix:
builtin-merge: show user-friendly error messages for fast-forward too.
merge-recursive: make the error-message generation an extern function
If a clone errors out because of a missing author, or user interrupt,
this allows `git svn fetch` to resume seamlessly, rather than forcing
the user to re-provide the path to the authors file.
[ew: shortened subject] Signed-off-by: Alex Vandiver <alex@chmrr.net> Acked-by: Eric Wong <normalperson@yhbt.net>
git-svn: Set svn.authorsfile to an absolute path when cloning
If --authors-file is passed a relative path, cloning will work, but
future `git svn fetch`es will fail to locate the authors file
correctly. Thus, use File::Spec->rel2abs to determine an absolute
path for the authors file before setting it.
Signed-off-by: Alex Vandiver <alex@chmrr.net> Acked-by: Eric Wong <normalperson@yhbt.net>
They caused a "git pull" (without any arguments, and without any local
commits---only to update to the other side) to warn that commit log
message is ignored because the merge resulted in a fast-forward.
Another possible solution is to add an extra option to "git merge" so that
"git pull" can tell it that the message given is not coming from the end
user (the canned message is passed just in case the merge resulted in a
non-ff and caused commit), but I think it is easier _not_ to deprecate the
old syntax.
Commit 24ab81a fixed the deletion of empty files, but broke
deletion of non-empty files. The approach it took was to
factor out the "deleted" line from the patch header into its
own hunk, the same way we do for mode changes. However,
unlike mode changes, we only showed the special "delete this
file" hunk if there were no other hunks. Otherwise, the user
would annoyingly be presented with _two_ hunks: one for
deleting the file and one for deleting the content.
This meant that in the non-empty case, we forgot about the
deleted line entirely, and we submitted a bogus patch to
git-apply (with "/dev/null" as the destination file, but not
marked as a deletion).
Instead, this patch combines the file deletion hunk and the
content deletion hunk (if there is one) into a single
deletion hunk which is either staged or not.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>