When making a partial commit, git-commit uses git-ls-files with
the --error-unmatch option to expand and sanity check the user
supplied path patterns. When any path pattern does not match
with the paths known to the index, it errors out, in order to
catch a common mistake to say "git commit Makefiel cache.h"
and end up with a commit that touches only cache.h (notice the
misspelled "Makefile"). This detection however does not work
well when the path has already been removed from the index.
If you drop a path from the index and try to commit that
partially, i.e.
$ git rm COPYING
$ git commit -m 'Remove COPYING' COPYING
the command complains because git does not know anything about
COPYING anymore.
This introduces a new option --with-tree to git-ls-files and
uses it in git-commit when we build a temporary index to
write a tree object for the partial commit.
When --with-tree=<tree-ish> option is specified, names from the
given tree are added to the set of names the index knows about,
so we can treat COPYING file in the example as known.
Of course, there is no reason to use "git rm" and git-aware
people have long time done:
$ rm COPYING
$ git commit -m 'Remove COPYING' COPYING
which works just fine. But this caused a constant confusion.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git diff" does not record index lines for pure mode changes (i.e. no
lines changed). Therefore, apply --index-info would call out a bogus
error.
Instead, fall back to reading the info from the current index.
Incidentally, this fixes an error where git-rebase would not rebase a
commit including a pure mode change, and changes requiring a threeway
merge.
Noticed and later tested by Chris Shoemaker.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We used to rely on the fact that cherry-pick would trigger the code path
to set limited = 1 in handle_commit(), when an uninteresting commit was
encountered.
However, when cherry picking between two independent branches, i.e. when
there are no merge bases, and there is only linear development (which can
happen when you cvsimport a fork of a project), no uninteresting commit
will be encountered.
So set limited = 1 when --cherry-pick was asked for.
Noticed by Martin Bähr.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In order to (almost) always show the name of the file without
relying on "-H" option of GNU grep, we used to add /dev/null to
the argument list unless we are doing -l or -L. This caused
"/dev/null:0" to show up when -c is given in the output.
It is not enough to add -c to the set of options we do not pass
/dev/null for. When we have too many files, we invoke grep
multiple times and we need to avoid giving a widow filename to
the last invocation -- otherwise we will not see the name.
This keeps two filenames when the argv[] buffer is about to
overflow and we have not finished iterating over the index, so
that the last round will always have at least two paths to work
with (and not require /dev/null).
An obvious and the only exception is when there is only 1 file
that is given to the underlying grep, and in that case we avoid
passing /dev/null and let the external "grep -c" report only the
number of matches.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
An earlier commit fixed type-change case in "git add -u".
This adds a test to make sure we do not introduce regression.
At the same time, it fixes a stupid typo in the error message.
Signed-off-by: Benoit Sigoure <tsuna@lrde.epita.fr>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
On filesystems where it is appropriate to set core.filemode
to false, test 29 ("commitdiff(0): mode change") fails when
git-commit does not notice a file (execute) permission change.
A fix requires noting the new file execute permission in the
index with a "git update-index --chmod=+x", prior to the commit.
Add a function (note_chmod) which implements this idea, and
insert a call in each test that modifies the x permission.
Signed-off-by: Ramsay Jones <ramsay@ramsay1.demon.co.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Most of this patch code and message was written by Shawn O. Pearce.
I made some tests to know what the problem was, and then I changed
the code related with the SIGPIPE signal.
If the user has misconfigured `user.signingkey` in their .git/config
or just doesn't have any secret keys on their keyring and they ask
for a signed tag with `git tag -s` we better make sure the resulting
tag was actually signed by gpg.
Prior versions of builtin git-tag allowed this failure to slip
by without error as they were not checking the return value of
the finish_command() so they did not notice when gpg exited with
an error exit status. They also did not fail if gpg produced an
empty output or if read_in_full received an error from the read
system call while trying to read the pipe back from gpg.
Finally, we did not actually honor any return value from the do_sign
function as it returns ssize_t but was being stored into an unsigned
long. This caused the compiler to optimize out the die condition,
allowing git-tag to continue along and create the tag object.
However, when gpg gets a wrong username, it exits before any read was done
and then the writing process receives SIGPIPE and program is terminated.
By ignoring this signal, anyway, the function write_or_die gets EPIPE from
write_in_full and exits returning 0 to the system without a message.
Here we better call to write_in_full directly so we can fail
printing a message and return safely to the caller.
With these issues fixed `git-tag -s` will now fail to create the
tag and will report a non-zero exit status to its caller, thereby
allowing automated helper scripts to detect (and recover from)
failure if gpg is not working properly.
Proposed-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Carlos Rica <jasampler@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The code to squelch empty diffs introduced by commit
fb13227e08 would inadvertently
populate filespec "two" of a submodule change using the uninitialized
(null) SHA1, thereby replacing the submodule SHA1 by 0{40} in the output.
This change teaches diffcore_skip_stat_unmatch to handle
submodule changes correctly.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Kristian Høgsberg pointed out that the two file modifications
we were doing during the 'creating initial files' step are not even
used within the test suite. This was actually confusing as we do
not even need these changes for the tests to pass. All that really
matters here is the specific commit dates are used so that these
appear in the branch's reflog, and that the dates are different so
that the branch will update when asked and the reflog entry is
also updated. There is no need for the file modification.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When the preimage we are patching is shorter than what the patch
text expects, we tried to match the buffer contents at the
"original" line with the fragment in full, without checking we
have enough data to match in the preimage. This caused the size
of a later memmove() to wrap around and attempt to scribble
almost the entire address space. Not good.
The code that follows the part this patch touches tries to match
the fragment with line offsets. Curiously, that code does not
have the problem --- it guards against reading past the end of
the preimage.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When a topic branch is rebased, some of whose commits are already
cherry-picked upstream:
o--X--A--B--Y <- master
\
A--B--Z <- topic
then 'git rebase -m master' would report:
Already applied: 0001 Y
Already applied: 0002 Y
With this fix it reports the expected:
Already applied: 0001 A
Already applied: 0002 B
As an added bonus, this change also avoids 'echo' of a commit message,
which might contain escapements.
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch removes certain behaviour of "git tag -l foo", currently
listing every tag name having "foo" as a substring. The same
thing now could be achieved doing "git tag -l '*foo*'".
This feature was added recently when git-tag.sh got the -n option
for showing tag annotations, because that commit also replaced the
old "grep pattern" behaviour with a more preferable "shell pattern"
behaviour (although slightly modified as you can see).
Thus, the following builtin-tag.c implemented it in order to
ensure that tests were passing unchanged with both programs.
Since common "shell patterns" match names with a given substring
_only_ when * is inserted before and after (as in "*substring*"), and
the "plain" behaviour cannot be achieved easily with the current
implementation, this is mostly the right thing to do, in order to
make it more flexible and consistent.
Tests for "git tag" were also changed to reflect this.
Signed-off-by: Carlos Rica <jasampler@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Although dcommit could detect if the first commit in the series
would conflict with the HEAD revision in SVN, it could not
detect conflicts in further commits it made.
Now we rebase each uncommitted change after each revision is
committed to SVN to ensure that we are up-to-date. git-rebase
will bail out on conflict errors if our next change cannot be
applied and committed to SVN cleanly, preventing accidental
clobbering of changes on the SVN-side.
--no-rebase users will have trouble with this, and are thus
warned if they are committing more than one commit. Fixing this
for (hopefully uncommon) --no-rebase users would be more complex
and will probably happen at a later date.
Thanks to David Watson for finding this and the original test.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
With this function, a commit filter can leave out unwanted commits
(such as temporary commits). It does _not_ undo the changeset
corresponding to that commit, but it _skips_ the revision. IOW
no tree object is changed by this.
If you like to commit early and often, but want to filter out all
intermediate commits, marked by "@@@" in the commit message, you can
now do this with
git filter-branch --commit-filter '
if git cat-file commit $GIT_COMMIT | grep '@@@' > /dev/null;
then
skip_commit "$@";
else
git commit-tree "$@";
fi' newbranch
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move the convenience functions to the top of git-filter-branch.sh, and
return from the script when the environment variable SOURCE_FUNCTIONS is
set.
By sourcing git-filter-branch with that variable set automatically, all
commit filters may access the convenience functions like "map".
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This clarifies the logic to omit fast-forward check and omit
trivial merge before running the specified strategy.
The "index_merge" variable started out as a flag to say "do not
do anything clever", but when recursive was changed to skip the
trivial merge, the semantics were changed and the variable alone
does not make sense anymore.
This splits the variable into two, allow_fast_forward (which is
almost always true, and avoids making a merge commit when the
other commit is a descendant of our branch, but is set to false
for ours and subtree) and allow_trivial_merge (which is false
for ours, recursive and subtree).
Unlike the earlier implementation, the "ours" strategy allows an
up-to-date condition. When we are up-to-date, the result will
be our commit, and by definition, we will have our tree as the
result.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Here is my attempt to fix this with a minimally intrusive patch.
* As "git --bare init" cannot tell if it was called with --bare or
just "GIT_DIR=. git init", I added an explicit assignment of
is_bare_repository_cfg on the codepath for "git --bare".
* GIT_WORK_TREE alone without GIT_DIR does not make any sense,
nor GIT_WORK_TREE with an explicit "git --bare". Catch that
mistake. It might make sense to move this check to "git.c"
side as well, but I tried to shoot for the minimum change for
now.
* Some scripts, especially from the olden days, rely on
traditional GIT_DIR behaviour in "git init". Namely, these
are some notable patterns:
(create a bare repository)
- mkdir some.git && cd some.git && GIT_DIR=. git init
- mkdir some.git && cd some.git && git --bare init
(create a non-bare repository)
- mkdir .git && GIT_DIR=.git git init
- mkdir .git && GIT_DIR=`pwd`/.git git init
This comes with a new test script and also passes the existing
test suite, but there may be cases that are still broken with
the current tip of master and this patch does not yet fix. I'd
appreciate help in straightening this mess out.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
After git-write-tree finishes computing the tree, it updates the
index so that later operations can take advantage of fully
populated cache tree.
However, anybody writing the index file has to mark the entries
that are racily clean. For each entry whose cached lstat(3)
data in the index exactly matches what is obtained from the
filesystem, if the timestamp on the index file was the same or
older than the modification timestamp of the file, the blob
contents and the work tree file, after convert_to_git(), need to
be compared, and if they are different, its index entry needs to
be marked not to match the lstat(3) data from the filesystem.
In order for this to work, convert_to_git() needs to work
correctly, which in turn means you need to read the config file
to get the settings of core.crlf and friends.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When squashing, rebase -i did not prevent fast forwards. This could
happen when picking some other commit than the first one, and then
squashing the first commit. So do not allow fast forwards when
squashing.
Noticed by Johannes Sixt.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When a thin pack wants to send a tree object at "sub/dir", and
the commit that is common between the sender and the receiver
that is used as the base object has a subproject at that path,
we should not try to use the data at "sub/dir" of the base tree
as a tree object. It is not a tree to begin with, and more
importantly, the commit object there does not have to even
exist.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The existing checkpoint command is very useful to force fast-import
to dump the branches out to disk so that standard Git tools can
access them and the objects they refer to. However there was not a
way to know when fast-import had finished executing the checkpoint
and it was safe to read those refs.
The progress command can be used to make fast-import output any
message of the frontend's choosing to standard out. The frontend
can scan for these messages using select() or poll() to monitor a
pipe connected to the standard output of fast-import.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
For the same reasons as the prior change we want to allow frontends
to omit the trailing LF that usually delimits commands. In some
cases these just make the input stream more verbose looking than
it needs to be, and its just simpler for the frontend developer to
get started if our parser is slightly more lenient about where an
LF is required and where it isn't.
To make this optional LF feature work we now have to buffer up to one
line of input in command_buf. This buffering can happen if we look
at the current input command but don't recognize it at this point
in the code. In such a case we need to "unget" the entire line,
but we cannot depend upon the stdio library to let us do ungetc()
for that many characters at once.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
A few fast-import frontend developers have found it odd that we
require the LF following a `data` command, especially in the exact
byte count format. Technically we don't need this LF to parse
the stream properly, but having it here does make the stream more
readable to humans. We can easily make the LF optional by peeking
at the next byte available from the stream and pushing it back into
the buffer if its not LF.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Several frontend developers have asked that some form of stream
comments be permitted within a fast-import data stream. This way
they can include information from their own frontend program about
where specific data was taken from in the source system, or about
a decision that their frontend may have made while creating the
fast-import data stream.
This change introduces comments in the Bourne-shell/Tcl/Perl style.
Lines starting with '#' are ignored, up to and including the LF.
Unlike the above mentioned three languages however we do not look for
and ignore leading whitespace. This just simplifies the definition
of the comment format and the code that parses them.
To make comments work we had to stop using read_next_command() within
cmd_data() and directly invoke read_line() during the inline variant
of the function. This is necessary to retain any lines of the
input data that might otherwise look like a comment to fast-import.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Michael Haggerty <mhagger@alum.mit.edu> noticed while debugging a
Git backend for cvs2svn that fast-import was barfing when he tried
to use "TAG_FIXUP" as a branch name for temporary work needed to
cleanup the tree prior to creating an annotated tag object.
The reason we were rejecting the branch name was check_ref_format()
returns -2 when there are less than 2 '/' characters in the input
name. TAG_FIXUP has 0 '/' characters, but is technically just as
valid of a ref as HEAD and MERGE_HEAD, so we really should permit it
(and any other similar looking name) during import.
New test cases have been added to make sure we still detect very
wrong branch names (e.g. containing [ or starting with .) and yet
still permit reasonable names (e.g. TAG_FIXUP).
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
The t1301-shared-repo.sh testscript uses /usr/bin/stat to get the file
mode, which isn't portable. Implement the test in shell using 'ls' as
shown by Junio.
Signed-off-by: Arjen Laarhoven <arjen@yaph.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Apply "Subproject commit HEX" changes produced by git-diff.
As usual in the current git, only the superproject itself is actually
modified (possibly creating empty directories for new submodules).
Any checked-out submodule is left untouched and is not required to
be up-to-date.
With clean-ups from Junio C Hamano.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-add -u also takes the path limiters, but unlike the
command without the -u option, the code forgot that it
could be invoked from a subdirectory, and did not correctly
handle the prefix.
Signed-off-by: Salikh Zakirov <salikh@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This applies to 'maint' to fix a rather serious data corruption
issue. When "git add -u" affects a subdirectory in such a way
that the only changes to its contents are path removals, the
next tree object written out of that index was bogus, as the
remove codepath forgot to invalidate the cache-tree entry.
Reported by Salikh Zakirov.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Running git-svn log <ref> -r<rev> against a <ref> other than the
current HEAD did not work if the <rev> was exclusive to the
other branch.
Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This makes .gitattributes files to be read from the index when
they are not checked out to the work tree. This is in line with
the way we always allowed low-level tools to operate in sparsely
checked out work tree in a reasonable way.
It swaps the order of new file creation and converting the blob
to work tree representation; otherwise when we are in the middle
of checking out .gitattributes we would notice an empty but
unwritten .gitattributes file in the work tree and will ignore
the copy in the index.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When you try to merge a path that involves binary file-level
merge, merge-recursive died rudely without cleaning up its own
mess. A files added by the merge were left in the working tree,
but the index was not written out (because it just punted and
died), so it was cumbersome for the user to retry it by first
running "git reset --hard".
This changes merge-recursive to still warn but do the "binary"
merge for such a path; leave the "our" version in the working
tree, but still keep the path unmerged so that the user can sort
it out.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Windows / cygwin don't support HT, LF, or TAB in file name so this test
is meaningless there.
Signed-off-by: Mark Levedahl <mdl123@verizon.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This allows to refresh only a subset of the project files, based on
the specified pathspecs.
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Here-text to create fake-editor did not use <<\EOF but <<EOF,
but there was no point doing so, as it quoted all the variables
anyway. Simplify it.
Also futureproof the special mode to edit COMMIT_EDITMSG file;
it is interested in editing the COMMIT_EDITMSG file in any
GIT_DIR; GIT_DIR may be given as an absolute path.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* cr/tag:
Teach "git stripspace" the --strip-comments option
Make verify-tag a builtin.
builtin-tag.c: Fix two memory leaks and minor notation changes.
launch_editor(): Heed GIT_EDITOR and core.editor settings
Make git tag a builtin.
While "git bundle" was a useful way to sneakernet incremental
changes, we did not allow:
$ git bundle create v2.6.20.bndl v2.6.20
to create a bundle that contains the whole history to a
well-known good revision. Such a bundle can be mirrored
everywhere, and people can prime their repository with it to
reduce the load on the repository that serves near the tip of
the development.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The old behaviour was to unilaterally default to the cwd is the work tree
when GIT_DIR was set, but GIT_WORK_TREE wasn't, no matter if we are inside
the GIT_DIR, or if GIT_DIR is actually something like ../../../.git.
Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The hashed contents did not matter in the end result, but it passed
an uninitialized variable to printf, which caused it to emit empty
while giving an error/usage message.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This changes the behaviour of cloning from a repository on the
local machine, by defaulting to "-l" (use hardlinks to share
files under .git/objects) and making "-l" a no-op. A new
option, --no-hardlinks, is also added to cause file-level copy
of files under .git/objects while still avoiding the normal
"pack to pipe, then receive and index pack" network transfer
overhead. The old behaviour of local cloning without -l nor -s
is availble by specifying the source repository with the newly
introduced file:///path/to/repo.git/ syntax (i.e. "same as
network" cloning).
* With --no-hardlinks (i.e. have all .git/objects/ copied via
cpio) would not catch the source repository corruption, and
also risks corrupted recipient repository if an
alpha-particle hits memory cell while indexing and resolving
deltas. As long as the recipient is created uncorrupted, you
have a good back-up.
* same-as-network is expensive, but it would catch the breakage
of the source repository. It still risks corrupted recipient
repository due to hardware failure. As long as the recipient
is created uncorrupted, you have a good back-up.
* The new default on the same filesystem, as long as the source
repository is healthy, it is very likely that the recipient
would be, too. Also it is very cheap. You do not get any
back-up benefit, though.
None of the method is resilient against the source repository
corruption, so let's discount that from the comparison. Then
the difference with and without --no-hardlinks matters primarily
if you value the back-up benefit or not. If you want to use the
cloned repository as a back-up, then it is cheaper to do a clone
with --no-hardlinks and two git-fsck (source before clone,
recipient after clone) than same-as-network clone, especially as
you are likely to do a git-fsck on the recipient if you are so
paranoid anyway.
Which leads me to believe that being able to use file:/// is
probably a good idea, if only for testability, but probably of
little practical value. We default to hardlinked clone for
everyday use, and paranoids can use --no-hardlinks as a way to
make a back-up.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When calling "git rebase -i <upstream> <branch>", git should switch
to <branch> first. This worked before, but I broke it by my
"Shut git rebase -i up" patch.
Fix that, and add a test to make sure that it does not break again.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Acked-by: Alex Riesen <raa.lkml@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
[jc: adjusted t/t7501 as this makes -F and --amend compatible]
Signed-off-by: David Kastrup <dak@gnu.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Non-interactive rebase had this from the beginning -- match it by
using --cherry-pick option to rev-list.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The old version of work-tree support was an unholy mess, barely readable,
and not to the point.
For example, why do you have to provide a worktree, when it is not used?
As in "git status". Now it works.
Another riddle was: if you can have work trees inside the git dir, why
are some programs complaining that they need a work tree?
IOW it is allowed to call
$ git --git-dir=../ --work-tree=. bla
when you really want to. In this case, you are both in the git directory
and in the working tree. So, programs have to actually test for the right
thing, namely if they are inside a working tree, and not if they are
inside a git directory.
Also, GIT_DIR=../.git should behave the same as if no GIT_DIR was
specified, unless there is a repository in the current working directory.
It does now.
The logic to determine if a repository is bare, or has a work tree
(tertium non datur), is this:
--work-tree=bla overrides GIT_WORK_TREE, which overrides core.bare = true,
which overrides core.worktree, which overrides GIT_DIR/.. when GIT_DIR
ends in /.git, which overrides the directory in which .git/ was found.
In related news, a long standing bug was fixed: when in .git/bla/x.git/,
which is a bare repository, git formerly assumed ../.. to be the
appropriate git dir. This problem was reported by Shawn Pearce to have
caused much pain, where a colleague mistakenly ran "git init" in "/" a
long time ago, and bare repositories just would not work.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
