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>
This is a quick and dirty fix for the broken "git cherry-pick -n" on
some broken OS, which does not remove the directory entry after unlink
succeeded(!) if the file is still open somewher.
The entry is left but "protected": no open, no unlink, no stat.
Very annoying.
Signed-off-by: Alex Riesen <raa.lkml@gmail.com>
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>
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>
This abstracts away the size of the hash values when copying them
from memory location to memory location, much as the introduction
of hashcmp abstracted away hash value comparsion.
A few call sites were using char* rather than unsigned char* so
I added the cast rather than open hashcpy to be void*. This is a
reasonable tradeoff as most call sites already use unsigned char*
and the existing hashcmp is also declared to be unsigned char*.
[jc: Splitted the patch to "master" part, to be followed by a
patch for merge-recursive.c which is not in "master" yet.
Fixed the cast in the latter hunk to combine-diff.c which was
wrong in the original.
Also converted ones left-over in combine-diff.c, diff-lib.c and
upload-pack.c ]
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Most of the callers except the one in refs.c use the function to
update the index file. Among the index writers, everybody
except write-tree dies if they cannot open it for writing.
This gives the function an extra argument, to tell it to die
when it cannot create a new file as the lockfile.
The only caller that does not have to die is write-tree, because
updating the index for the cache-tree part is optional and not
being able to do so does not affect the correctness. I think we
do not have to be so careful and make the failure into die() the
same way as other callers, but that would be a different patch.
Signed-off-by: Junio C Hamano <junkio@cox.net>
The only change in behaviour should be having a "usage: " prefix
on the output string rather than "fatal: ", and an exit code of
129 rather than 128.
Signed-off-by: Ramsay Allan Jones <ramsay@ramsay1.demon.co.uk>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This changes the calling convention of built-in commands and
passes the "prefix" (i.e. pathname of $PWD relative to the
project root level) down to them.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
hold_lock_file_for_update() relies on that.
Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
The framework to create lockfiles that are removed at exit is
first used to reliably write the index file, but it is
applicable to other things, so stop calling it "cache_file".
This also rewords a few remaining error message that called the
index file "cache file".
Signed-off-by: Junio C Hamano <junkio@cox.net>
The "bind" commit can express an aggregation of multiple
projects into a single commit.
In such an organization, there would be one project, root of
whose tree object is at the same level of the root of the
aggregated projects, and other projects have their toplevel in
separate subdirectories. Let's call that root level project the
"primary project", and call other ones just "subprojects".
You would first read-tree the primary project, and then graft
the subprojects under their appropriate location using read-tree
--prefix=<subdir>/ repeatedly.
To write out a tree object from such an index for a subproject,
write-tree --prefix=<subdir>/ is used.
Signed-off-by: Junio C Hamano <junkio@cox.net>
When the extra "dryrun" parameter is true, cache_tree_update()
recomputes the invalid entry but does not actually creates
new tree object.
Signed-off-by: Junio C Hamano <junkio@cox.net>
The updated write-tree reads from $GIT_DIR/index.aux to pick up
subtree objects information, updates the cache-tree with the
index, and updates index.aux file after writing a tree out of
the index file.
Until update-index and other programs that modify the index are
updated to maintain index.aux file, the index.aux file written
by the last write-tree will become stale immediately after they
update the index, which will result in the whole tree
recomputation just like the original write-tree.
The idea is to convert those commands to invalidate cache-tree
whenever they touch the index entries, and write updated
index.aux out. After the index is updated with them, write-tree
will be able to reuse the parts of the cache-tree that have not
been touched.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This replaces occurences of "blob", "commit", "tag", and "tree",
where they're really used as type specifiers, which we already
have defined global constants for.
Signed-off-by: Peter Eriksen <s022018@student.dtu.dk>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This adds "assume unchanged" logic, started by this message in the list
discussion recently:
<Pine.LNX.4.64.0601311807470.7301@g5.osdl.org>
This is a workaround for filesystems that do not have lstat()
that is quick enough for the index mechanism to take advantage
of. On the paths marked as "assumed to be unchanged", the user
needs to explicitly use update-index to register the object name
to be in the next commit.
You can use two new options to update-index to set and reset the
CE_VALID bit:
git-update-index --assume-unchanged path...
git-update-index --no-assume-unchanged path...
These forms manipulate only the CE_VALID bit; it does not change
the object name recorded in the index file. Nor they add a new
entry to the index.
When the configuration variable "core.ignorestat = true" is set,
the index entries are marked with CE_VALID bit automatically
after:
- update-index to explicitly register the current object name to the
index file.
- when update-index --refresh finds the path to be up-to-date.
- when tools like read-tree -u and apply --index update the working
tree file and register the current object name to the index file.
The flag is dropped upon read-tree that does not check out the index
entry. This happens regardless of the core.ignorestat settings.
Index entries marked with CE_VALID bit are assumed to be
unchanged most of the time. However, there are cases that
CE_VALID bit is ignored for the sake of safety and usability:
- while "git-read-tree -m" or git-apply need to make sure
that the paths involved in the merge do not have local
modifications. This sacrifices performance for safety.
- when git-checkout-index -f -q -u -a tries to see if it needs
to checkout the paths. Otherwise you can never check
anything out ;-).
- when git-update-index --really-refresh (a new flag) tries to
see if the index entry is up to date. You can start with
everything marked as CE_VALID and run this once to drop
CE_VALID bit for paths that are modified.
Most notably, "update-index --refresh" honours CE_VALID and does
not actively stat, so after you modified a file in the working
tree, update-index --refresh would not notice until you tell the
index about it with "git-update-index path" or "git-update-index
--no-assume-unchanged path".
This version is not expected to be perfect. I think diff
between index and/or tree and working files may need some
adjustment, and there probably needs other cases we should
automatically unmark paths that are marked to be CE_VALID.
But the basics seem to work, and ready to be tested by people
who asked for this feature.
Signed-off-by: Junio C Hamano <junkio@cox.net>
"git-write-tree junk" complains and dies, but it does not say
what option it supports. Die with the usage string in such a
case.
Signed-off-by: Junio C Hamano <junkio@cox.net>
These commands are converted to run from a subdirectory.
commit-tree convert-objects merge-base merge-index mktag
pack-objects pack-redundant prune-packed read-tree tar-tree
unpack-file unpack-objects update-server-info write-tree
Signed-off-by: Junio C Hamano <junkio@cox.net>
Simple whitespace-related tidyups ensuring style consistency.
This is carried over from my old git-pb branch.
Signed-off-by: Petr Baudis <pasky@ucw.cz>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This option allows a write-tree even if the referenced objects are not
in the database.
Signed-off-by: Bryan Larsen <bryan.larsen@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
git-write-tree failed when referenced objects only exist in the
GIT_ALTERNATE_OBJECT_DIRECTORIES path.
Signed-off-by: Jan Harkes <jaharkes@cs.cmu.edu>
Acked-by: Junio C Hamano <junkio@cox.net>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Fixes all in-code names that leaved during "big name change".
Signed-off-by: Alexey Nezhdanov <snake@penza-gsm.ru>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Kay Sievers noticed that you can have both path and path/file in
the cache and write-tree happily creates a tree object from such
a state. Since a merge can result in such situation and the
user should be able to see the situation by looking at the
cache, rather than forbidding add_cache_entry() to create such
conflicts, fix it by making write-tree refuse to write such an
nonsensical tree. Here is a test case.
-- test case --
$ ls -a
./ ../
$ git-init-db
defaulting to local storage area
$ date >path
$ git-update-cache --add path
$ rm path
$ mkdir path
$ date >path/file
$ git-update-cache --add path/file
$ git-ls-files --stage
100644 1738f2536b1201218c41153941da065cc26174c9 0 path
100644 620c72f1c1de15f56ff9d63d6d7cdc69e828f1e3 0 path/file
$ git-ls-tree $(git-write-tree) ;# using old one
100644 blob 1738f2536b1201218c41153941da065cc26174c9 path
040000 tree ec116937f223e3df95aeac9f076902ae1618ae98 path
$ ../git-write-tree ;# using new one
You have both path and path/file
fatal: write-tree: not able to write tree
$ exit
Signed-off-by: Junio C Hamano <junkio@cox.net>
Introduce xmalloc and xrealloc to die gracefully with a descriptive
message when out of memory, rather than taking a SIGSEGV.
Signed-off-by: Christopher Li<chrislgit@chrisli.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Of course, we can't even generate such an index yet, but give me
some time. This is a cunning plan. Let's see if it actually works.
(I feel like Wile E Coyote, waiting for the big rock to fall).
This allows using a git tree over NFS with different byte order, and
makes it possible to just copy a fully populated repository and have
the end result immediately usable (needing just a refresh to update
the stat information).
Now there is error() for "library" errors and die() for fatal "application"
errors. usage() is now used strictly only for usage errors.
Signed-off-by: Petr Baudis <pasky@ucw.cz>
It now requires the "--add" flag before you add any new files, and
a "--remove" file if you want to mark files for removal. And giving
it the "--refresh" flag makes it just update all the files that it
already knows about.
It's got some debugging printouts etc still in it, but testing on the
kernel seems to show that it does indeed fix the issue with huge tree
files for each commit.
The tool interface sucks (especially "committing" information, which is just
me doing everything by hand from the command line), but I think this is in
theory actually a viable way of describing the world. So copyright it.
Linus Torvalds