The algorithm isn't right here: it accumulates any set of 8 spaces into
tabs even if they're separated by tabs, so
<four spaces><tab><four spaces><tab>
is converted to
<tab><tab><tab>
when it should be just
<tab><tab>
So teach git-apply that a tab hides any group of less than 8 previous
spaces in a row.
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
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>
Earlier, add_file_to_index() invalidated the path in the cache-tree
but remove_file_from_cache() did not, and the user of the latter
needed to invalidate the entry himself. This led to a few bugs due to
missed invalidate calls already. This patch makes the management of
cache-tree less error prone by making more invalidate calls from lower
level cache API functions.
The rules are:
- If you are going to write the index, you should either maintain
cache_tree correctly.
- If you cannot, alternatively you can remove the entire cache_tree
by calling cache_tree_free() before you call write_cache().
- When you modify the index, cache_tree_invalidate_path() should be
called with the path you are modifying, to discard the entry from
the cache-tree structure.
- The following cache API functions exported from read-cache.c (and
the macro whose names have "cache" instead of "index")
automatically call cache_tree_invalidate_path() for you:
- remove_file_from_index();
- add_file_to_index();
- add_index_entry();
You can modify the index bypassing the above API functions
(e.g. find an existing cache entry from the index and modify it in
place). You need to call cache_tree_invalidate_path() yourself in
such a case.
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>
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>
We attempt to remove directory that becomes empty after removal
of a file. We should do the same when we rename an existing
file away.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git apply" used to take check the whitespace in the wrong
direction.
Noticed by Daniel Barkalow.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The earlier code removed one newline too many from the hunk that
adds new lines at the end of the file. Also the way the code
counted the added blank lines was somewhat roundabout; I think
the way updated code does it is more direct and easier to
follow:
* We keep track of the number of blank lines added;
* While processing each line, we notice if it adds a blank
line, and increment the counter, or reset it to zero
otherwise;
* When actually we apply the data, we remove the empty lines we
counted earlier if we are applying it at the end of the
file.
Signed-off-by: Junio C Hamano <junkio@cox.net>
[jc: with an obvious microfix to avoid doing this unless --whitespace=strip]
Signed-off-by: Marco Costalba <mcostalba@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This patch fixes all calls to xread() where the return value is not
stored into an ssize_t. The patch should not have any effect whatsoever,
other than putting better/more appropriate type names on variables.
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
HPA noticed that git-rebase fails when changes involve symlinks
in the middle of the hierarchy. Consider:
* The tree state before the patch is applied has arch/x86_64/boot
as a symlink pointing at ../i386/boot/
* The patch tries to remove arch/x86_64/boot symlink, and
create bunch of files there: .gitignore, Makefile, etc.
git-apply tries to be careful while applying patches; it never
touches the working tree until it is convinced that the patch
would apply cleanly. One of the check it does is that when it
knows a path is going to be created by the patch, it runs
lstat() on the path to make sure it does not exist.
This leads to a false alarm. Because we do not touch the
working tree before all the check passes, when we try to make
sure that arch/x86_64/boot/.gitignore does not exist yet, we
haven't removed the arch/x86_64/boot symlink. The lstat() check
ends up seeing arch/i386/boot/.gitignore through the
yet-to-be-removed symlink, and says "Hey, you already have a
file there, but what you fed me is a patch to create a new
file. I am not going to clobber what you have in the working
tree."
We have similar checks to see a file we are going to modify does
exist and match the preimage of the diff, which is done by
directly opening and reading the file.
For a file we are going to delete, we make sure that it does
exist and matches what is going to be removed (a removal patch
records the full preimage, so we check what you have in your
working tree matches it in full -- otherwise we would risk
losing your local changes), which again is done by directly
opening and reading the file.
These checks need to be adjusted so that they are not fooled by
symlinks in the middle.
- To make sure something does not exist, first lstat(). If it
does not exist, it does not, so be happy. If it _does_, we
might be getting fooled by a symlink in the middle, so break
leading paths and see if there are symlinks involved. When
we are checking for a path a/b/c/d, if any of a, a/b, a/b/c
is a symlink, then a/b/c/d does _NOT_ exist, for the purpose
of our test.
This would fix this particular case you saw, and would not
add extra overhead in the usual case.
- To make sure something already exists, first lstat(). If it
does not exist, barf (up to this, we already do). Even if it
does seem to exist, we might be getting fooled by a symlink
in the middle, so make sure leading paths are not symlinks.
This would make the normal codepath much more expensive for
deep trees, which is a bit worrisome.
This patch implements the first side of the check "making sure
it does not exist". The latter "making sure it exists" check is
not done yet, so applying the patch in reverse would still
fail, but we have to start from somewhere.
Signed-off-by: Junio C Hamano <junkio@cox.net>
Noticed by applying two diffs of different contexts to the same file.
The check for existence of a file was wrong: the test assumed it was
a directory and reset the errno (twice: directly and by calling
lstat). So if an entry existed and was _not_ a directory no attempt
was made to rename into it, because the errno (expected by renaming
code) was already reset to 0. This resulted in error:
fatal: unable to write file file mode 100644
For Linux, removing "errno = 0" is enough, as lstat wont modify errno
if it was successful. The behavior should not be depended upon,
though, so modify the "if" as well.
The test simulates this situation.
Signed-off-by: Alex Riesen <raa.lkml@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This is an internal variable used to tell if we need to write
out the resulting index.
I'll be introducing write_index() function which would collide
with it.
Signed-off-by: Junio C Hamano <junkio@cox.net>
If the user is trying to apply a Git generated diff file and they
have specified a -p<n> option, where <n> is not 1, the user probably
has a good reason for doing this. Such as they are me, trying to
apply a patch generated in git.git for the git-gui subdirectory to
the git-gui.git repository, where there is no git-gui subdirectory
present.
Users shouldn't supply -p2 unless they mean it. But if they are
supplying it, they probably have thought about how to make this
patch apply to their working directory, and want to risk whatever
results may come from that.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
When defined, this allows plumbing commands that update the
index (add, apply, checkout-index, merge-recursive, mv,
read-tree, rm, update-index, and write-tree) to write their
resulting index to an alternative index file while holding a
lock to the original index file. With this, git-commit that
jumps the index does not have to make an extra copy of the index
file, and more importantly, it can do the update while holding
the lock on the index.
However, I think the interface to let an environment variable
specify the output is a mistake, as shown in the documentation.
If a curious user has the environment variable set to something
other than the file GIT_INDEX_FILE points at, almost everything
will break. This should instead be a command line parameter to
tell these plumbing commands to write the result in the named
file, to prevent stupid mistakes.
Signed-off-by: Junio C Hamano <junkio@cox.net>
When we write out the result of patch application, we sometimes
need to munge the data (e.g. under core.autocrlf). After doing
so, what we should free is the temporary buffer that holds the
converted data returned from convert_to_working_tree(), not the
original one.
This patch also moves the call to open() up in the function, as
the caller expects us to fail cheaply if leading directories
need to be created (and then the caller creates them and calls
us again). For that calling pattern, attempting conversion
before opening the file adds unnecessary overhead.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Some systems have sizeof(off_t) == 8 while sizeof(size_t) == 4.
This implies that we are able to access and work on files whose
maximum length is around 2^63-1 bytes, but we can only malloc or
mmap somewhat less than 2^32-1 bytes of memory.
On such a system an implicit conversion of off_t to size_t can cause
the size_t to wrap, resulting in unexpected and exciting behavior.
Right now we are working around all gcc warnings generated by the
-Wshorten-64-to-32 option by passing the off_t through xsize_t().
In the future we should make xsize_t on such problematic platforms
detect the wrapping and die if such a file is accessed.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Some file systems that can host git repositories and their working copies
do not support symbolic links. But then if the repository contains a symbolic
link, it is impossible to check out the working copy.
This patch enables partial support of symbolic links so that it is possible
to check out a working copy on such a file system. A new flag
core.symlinks (which is true by default) can be set to false to indicate
that the filesystem does not support symbolic links. In this case, symbolic
links that exist in the trees are checked out as small plain files, and
checking in modifications of these files preserve the symlink property in
the database (as long as an entry exists in the index).
Of course, this does not magically make symbolic links work on such defective
file systems; hence, this solution does not help if the working copy relies
on that an entry is a real symbolic link.
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
Signed-off-by: Junio C Hamano <junkio@cox.net>
We currently have two parallel notation for dealing with object types
in the code: a string and a numerical value. One of them is obviously
redundent, and the most used one requires more stack space and a bunch
of strcmp() all over the place.
This is an initial step for the removal of the version using a char array
found in object reading code paths. The patch is unfortunately large but
there is no sane way to split it in smaller parts without breaking the
system.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Internal function apply_line() is called to copy both context lines
and added lines to the output buffer, while possibly fixing the
whitespace breakages depending on --whitespace=strip settings.
However, it did its fix-up on both context lines and added lines.
This resulted in two symptoms:
(1) The number of lines reported to have been fixed up included
these context lines.
(2) However, the lines actually shown were limited to the added
lines that had whitespace breakages.
Signed-off-by: Junio C Hamano <junkio@cox.net>
When a patch fails, I automatically add '-v' to the command line
to see what fails.
This patch makes -v a synonym to --verbose, and actually tells
the user which text was not found.
Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This enhances the third point in the previous commit. When
applying a non-git patch that begins like this:
--- 2.6.orig/mm/slab.c
+++ 2.6/mm/slab.c
@@ -N,M +L,K @@@
...
and if you are in 'mm' subdirectory, we notice that -p2 is the
right option to use to apply the patch in file slab.c in the
current directory (i.e. mm/slab.c)
The guess function also knows about this pattern, where you
would need to use -p0 if applying from the top-level:
--- mm/slab.c
+++ mm/slab.c
@@ -N,M +L,K @@@
...
Signed-off-by: Junio C Hamano <junkio@cox.net>
Earlier one that tried to be too consistent with GNU patch by
not stripping the leading path when we _know_ we are in a
subdirectory and the patch is relative to the toplevel was a
mistake. This fixes it.
- No change to behaviour when it is run from the toplevel of
the repository.
- When run from a subdirectory to apply a git-generated patch,
it uses the right -p<n> value automatically, with or without
--index nor --cached option.
- When run from a subdirectory to apply a randomly generated
patch, it wants the right -p<n> value to be given by the
user.
The second one is a pure improvement to correct inconsistency
between --index and non --index case, compared with 1.5.0. The
third point could be further improved to guess what the right
value for -p<n> should be by looking at the patch, but should be
a topic of a separate patch.
Signed-off-by: Junio C Hamano <junkio@cox.net>
When (new_name == old_name), the previous one prefixed old_name
alone, leaving new_name untouched, and worse yet, left it
dangling pointing at an already freed memory location.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This mechanically converts strncmp() to use prefixcmp(), but only when
the parameters match specific patterns, so that they can be verified
easily. Leftover from this will be fixed in a separate step, including
idiotic conversions like
if (!strncmp("foo", arg, 3))
=>
if (!(-prefixcmp(arg, "foo")))
This was done by using this script in px.perl
#!/usr/bin/perl -i.bak -p
if (/strncmp\(([^,]+), "([^\\"]*)", (\d+)\)/ && (length($2) == $3)) {
s|strncmp\(([^,]+), "([^\\"]*)", (\d+)\)|prefixcmp($1, "$2")|;
}
if (/strncmp\("([^\\"]*)", ([^,]+), (\d+)\)/ && (length($1) == $3)) {
s|strncmp\("([^\\"]*)", ([^,]+), (\d+)\)|(-prefixcmp($2, "$1"))|;
}
and running:
$ git grep -l strncmp -- '*.c' | xargs perl px.perl
Signed-off-by: Junio C Hamano <junkio@cox.net>
git-apply running inside a subdirectory, with or without --index,
used to always assume that the patch is formatted in such a way
to apply with -p1 from the toplevel, but it is more useful and
consistent with the use of "GNU patch -p1" if it defaulted to
assume that its input is meant to apply at the level it is
invoked in.
This changes the behaviour. It used to be that the patch
generated this way would apply without any trick:
edit Documentation/Makefile
git diff >patch.file
cd Documentation
git apply ../patch.file
You need to give an explicit -p2 to git-apply now. On the other
hand, if you got a patch from somebody else who did not follow
"patch is to apply from the top with -p1" convention, the input
patch would start with:
diff -u Makefile.old Makefile
--- Makefile.old
+++ Makefile
and in such a case, you can apply it with:
git apply -p0 patch.file
Signed-off-by: Junio C Hamano <junkio@cox.net>
When a patch modifies (not deletes) the last file in a
directory, because we treat a modification just as deletion
followed by creation, and deleting the last file in a directory
automatically rmdir(2)'s that directory, we ended up removing
the directory, which can potentially be the cwd, and then
recreating the same directory to create the patch result.
Avoid the rmdir step when remove_file() is called only because
we are replacing it with the result by later calling
create_file().
Signed-off-by: Junio C Hamano <junkio@cox.net>
This teaches git-apply that the data read from and written to
the filesystem might need to get converted to adjust for local
line-ending convention.
Signed-off-by: Junio C Hamano <junkio@cox.net>
When neither --index nor --cached was used, git-apply did not
try calling setup_git_directory(), which means it did not look
at configuration files at all. This fixes it to call the setup
function but still allow the command to be run in a directory
not controlled by git.
The bug probably meant that 'git apply', not moving up to the
toplevel, did not apply properly formatted diffs from the
toplevel when you are inside a subdirectory, even though 'git
apply --index' would. As a side effect, this patch fixes it as
well.
Signed-off-by: Junio C Hamano <junkio@cox.net>
When we do not trust executable bit from lstat(2), we copied
existing ce_mode bits without checking if the filesystem object
is a regular file (which is the only thing we apply the "trust
executable bit" business) nor if the blob in the index is a
regular file (otherwise, we should do the same as registering a
new regular file, which is to default non-executable).
Noticed by Johannes Sixt.
Signed-off-by: Junio C Hamano <junkio@cox.net>
The static variable "prefix" was shadowed by an unused parameter
of the same name. In case of execution in a subdirectory, the
static variable was accessed, leading to a crash.
Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Returning negative value from there does not stop the caller from using
the earlier part.
Noticed by Linus.
Signed-off-by: Junio C Hamano <junkio@cox.net>
find_header() function is used to read and parse the patchfile
and it detects errors in the patch, but one place ignored the
error and went ahead, which was quite bad.
Noticed by Jeff Garzik.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This is a mechanical clean-up of the way *.c files include
system header files.
(1) sources under compat/, platform sha-1 implementations, and
xdelta code are exempt from the following rules;
(2) the first #include must be "git-compat-util.h" or one of
our own header file that includes it first (e.g. config.h,
builtin.h, pkt-line.h);
(3) system headers that are included in "git-compat-util.h"
need not be included in individual C source files.
(4) "git-compat-util.h" does not have to include subsystem
specific header files (e.g. expat.h).
Signed-off-by: Junio C Hamano <junkio@cox.net>
This patch fixes a sparse warning about inaccurate_eof being a
"dubious one-bit signed bitfield", makes three more binary
variables members of this (now unsigned) bitfield and adds a
short comment to indicate the nature of two ternary variables.
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <junkio@cox.net>
We do not even know number of lines so showing it as 0 0 is
lying. This would also help Porcelains like cvsexportcommit.
Signed-off-by: Junio C Hamano <junkio@cox.net>
At least one (older) version of the Solaris C compiler won't allow
'unsigned long x = -1' without explicitly casting -1 to a type of
unsigned long. So instead use ULONG_MAX, which is really the
correct constant anyway.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
We deduced a GNU diff output that does not use /dev/null convention
as creation (deletion) diff correctly by looking at the lack of context
and deleted lines (added lines), but forgot to reset the new (old) name
field properly.
This was a regression when we added a workaround for --unified=0 insanity.
Signed-off-by: Junio C Hamano <junkio@cox.net>
The latest GNU diff from CVS emits an empty line to express
an empty context line, instead of more traditional "single
white space followed by a newline". Do not get broken by it.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Most callers of write_sha1_file_prepare() are only interested in the
resulting hash but don't care about the returned file name or the header.
This patch adds a simple wrapper named hash_sha1_file() which does just
that, and converts potential callers.
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Somebody was wondering on #git channel why a git generated diff
does not apply with GNU patch when the filename contains a SP.
It is because GNU patch expects to find TAB (and trailing timestamp)
on ---/+++ (old_name and new_name) lines after the filenames.
The "diff --git" output format was carefully designed to be
compatible with GNU patch where it can, but whitespace
characters were always a pain.
We can make our output a bit more GNU patch friendly by adding an
extra TAB (but not trailing timestamp) to old/new name lines when
the filename as a SP in it. This updates git-apply to prepare
ourselves to accept such a patch, but we still do not generate
output that is patch friendly yet. That change needs to wait
until everybody has this change.
When a filename contains a real tab, "diff --git" format
always c-quotes it as discussed on the list with GNU patch
maintainer previously:
http://marc.theaimsgroup.com/?l=git&m=112927316408690&w=2
so there should be no downside.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This makes --whitespace={warn,error,strip} option to also notice
the leading whitespace errors in addition to the trailing
whitespace errors. Spaces that are followed by a tab in indent
are detected as errors, and --whitespace=strip option fixes them.
Signed-off-by: Junio C Hamano <junkio@cox.net>
In "git-apply", we have a few sanity checks and heuristics that
expects that the patch fed to us is a unified diff with at least
one line of context.
* When there is no leading context line in a hunk, the hunk
must apply at the beginning of the preimage. Similarly, no
trailing context means that the hunk is anchored at the end.
* We learn a patch deletes the file from a hunk that has no
resulting line (i.e. all lines are prefixed with '-') if it
has not otherwise been known if the patch deletes the file.
Similarly, no old line means the file is being created.
And we declare an error condition when the file created by a
creation patch already exists, and/or when a deletion patch
still leaves content in the file.
These sanity checks are good safety measures, but breaks down
when people feed a diff generated with --unified=0. This was
recently noticed first by Matthew Wilcox and Gerrit Pape.
This adds a new flag, --unified-zero, to allow bypassing these
checks. If you are in control of the patch generation process,
you should not use --unified=0 patch and fix it up with this
flag; rather you should try work with a patch with context. But
if all you have to work with is a patch without context, this
flag may come handy as the last resort.
Signed-off-by: Junio C Hamano <junkio@cox.net>
Pierre Habouzit