Moving "git-cmd" commands out of the path and into a special
git exec path, should include the builtins.
[jc: fixed the case where bindir == gitexecdir - ln -f fails
with a complaint that src and dst are the same, likewise for
the fallback cp.]
Signed-off-by: Junio C Hamano <junkio@cox.net>
For systems which lack inet_ntop(), this adds compat/inet_ntop.c,
and related build constant, NO_INET_NTOP. Older Cygwin(s) lack
inet_ntop().
Signed-off-by: Yakov Lerner <iler.ml@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This makes tar-tree a built-in. As an added bonus, you can now
say:
git tar-tree --remote=remote-repository <ent> [<base>]
This does not work with git-daemon yet, but should work with
localhost and git over ssh transports.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This changes semantics very subtly, because it adds a new atomicity
guarantee.
In particular, if you "git rm" several files, it will now do all or
nothing. The old shell-script really looped over the removed files one by
one, and would basically randomly fail in the middle if "-f" was used and
one of the files didn't exist in the working directory.
This C builtin one will not re-write the index after each remove, but
instead remove all files at once. However, that means that if "-f" is used
(to also force removal of the file from the working directory), and some
files have already been removed from the workspace, it won't stop in the
middle in some half-way state like the old one did.
So what happens is that if the _first_ file fails to be removed with "-f",
we abort the whole "git rm". But once we've started removing, we don't
leave anything half done. If some of the other files don't exist, we'll
just ignore errors of removal from the working tree.
This is only an issue with "-f", of course.
I think the new behaviour is strictly an improvement, but perhaps more
importantly, it is _different_. As a special case, the semantics are
identical for the single-file case (which is the only one our test-suite
seems to test).
The other question is what to do with leading directories. The old "git
rm" script didn't do anything, which is somewhat inconsistent. This one
will actually clean up directories that have become empty as a result of
removing the last file, but maybe we want to have a flag to decide the
behaviour?
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
c3c8835fbb broke the default template
location which is in builtin-init-db.o, by not supplying the
compilation-time constant to the right build commands.
Signed-off-by: Junio C Hamano <junkio@cox.net>
Basically this just renames init-db.c to builtin-init-db.c and makes
some strings const.
Signed-off-by: Timo Hirvonen <tihirvon@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Importing a quilt patch series into git is not very difficult
but parsing the patch descriptions and all of the other
minutia take a bit of effort to get right, so this automates it.
Since git and quilt complement each other it makes sense
to make it easy to go back and forth between the two.
If a patch is encountered that it cannot derive the author
from the user is asked.
Signed-off-by: Junio C Hamano <junkio@cox.net>
[jc: rewrote by stealing from what I run to update html and
man branches automatically]
Signed-off-by: Tilman Sauerbeck <tilman@code-monkey.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This was surprisingly easy. The diff is truly minimal: rename "main()" to
"cmd_rev_list()" in rev-list.c, and rename the whole file to reflect its
new built-in status.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
First try. Let's see how well this works.
In many ways, the hard parts of "git commit" are not so different from
this, and a builtin commit would share a lot of the code, I think.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Some implementations do not know what to do with -H; define
NO_H_OPTION_IN_GREP when you build git if your grep lacks -H.
Most of the time, it can be worked around by prepending
/dev/null to the argument list, but that causes -L and -c to
slightly misbehave (they both expose /dev/null is given), so
when these options are given, do not run external grep that does
not understand -H.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This moves the core directory traversal and filename exclusion logic
into the general git library, making it available for other users
directly.
If we ever want to do "git commit" or "git add" as a built-in (and we
do), we want to be able to handle most of git-ls-files as a library.
NOTE! Not all of git-ls-files is libified by this. The index matching
and pathspec prefix calculation is still in ls-files.c, but this is a
big part of it.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
After 567ffeb772 and
4bc87a28be, git-send-email no
longer requires any non-standard Perl modules, so there's no
reason to special-case it.
Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
NetBSD >=2.0 has iconv() in libc. A libiconv is not required and
does not exist.
See: http://netbsd.gw.com/cgi-bin/man-cgi?iconv+3+NetBSD-2.0
[jc: with a bit of simplification later discussed on the list.]
Signed-off-by: Dennis Stosberg <dennis@stosberg.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
After 567ffeb772 and
4bc87a28be, git-send-email no
longer requires any non-standard Perl modules, so there's no
reason to special-case it.
Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This adds "binary patch" to the diff output and teaches apply
what to do with them.
On the diff generation side, traditionally, we said "Binary
files differ\n" without giving anything other than the preimage
and postimage object name on the index line. This was good
enough for applying a patch generated from your own repository
(very useful while rebasing), because the postimage would be
available in such a case. However, this was not useful when the
recipient of such a patch via e-mail were to apply it, even if
the preimage was available.
This patch allows the diff to generate "binary" patch when
operating under --full-index option. The binary patch follows
the usual extended git diff headers, and looks like this:
"GIT binary patch\n"
<length byte><data>"\n"
...
"\n"
Each line is prefixed with a "length-byte", whose value is upper
or lowercase alphabet that encodes number of bytes that the data
on the line decodes to (1..52 -- 'A' means 1, 'B' means 2, ...,
'Z' means 26, 'a' means 27, ...). <data> is 1 or more groups of
5-byte sequence, each of which encodes up to 4 bytes in base85
encoding. Because 52 / 4 * 5 = 65 and we have the length byte,
an output line is capped to 66 characters. The payload is the
same diff-delta as we use in the packfiles.
On the consumption side, git-apply now can decode and apply the
binary patch when --allow-binary-replacement is given, the diff
was generated with --full-index, and the receiving repository
has the preimage blob, which is the same condition as it always
required when accepting an "Binary files differ\n" patch.
Signed-off-by: Junio C Hamano <junkio@cox.net>
When inspecting a project whose build infrastructure used to
assume that .git/HEAD is a symlink ref, core.prefersymlinkrefs
in the config file of such a project would help to bisect its
history.
Signed-off-by: Junio C Hamano <junkio@cox.net>
(cherry picked from 9f0bb90d16 commit)
We used to depend on bignum from openssl for rev-list to compute
merge-order, but there is no reason to use different build
recipe from other programs anymore. Just build it with git-%$X
rule like everybody else.
Noticed by Alexey Dobriyan.
Signed-off-by: Junio C Hamano <junkio@cox.net>
When inspecting a project whose build infrastructure used to
assume that .git/HEAD is a symlink ref, core.prefersymlinkrefs
in the config file of such a project would help to bisect its
history.
Signed-off-by: Junio C Hamano <junkio@cox.net>
Remove the shell-script version, make the hardlink from the git
binary, and update the documentation to describe a new option.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This attempts to set up built-in "git grep" to further reduce
our dependence on the shell, while at the same time optionally
allowing to run grep against object database. You could do
funky things like these:
git grep --cached -e pattern ;# grep from index
git grep -e pattern master ;# or in a rev
git grep -e pattern master next ;# or in multiple revs
git grep -e pattern pu^@ ;# even like this with an
;# extension from another topic ;-)
git grep -e pattern master..next ;# or even from rev ranges
git grep -e pattern master~20:Documentation
;# or an arbitrary tree
git grep -e pattern next:git-commit.sh
;# or an arbitrary blob
Right now, it does not understand and/or obey many options grep
should accept, and the pattern must be given with -e option due
to the way the parameter parser is structured, both of which
obviously need to be fixed for usability.
But this is going in the right direction. The shell script
version is one of the worst Portability offender in the git
barebone Porcelainish; it uses xargs -0 to pass paths around and
shell arrays to sift flags and parameters.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This adds a builtin "push" command, which is largely just a C'ification of
the "git-push.sh" script.
Now, the reason I did it as a built-in is partly because it's yet another
step on relying less on shell, but it's actually mostly because I've
wanted to be able to push to _multiple_ repositories, and the most obvious
and simplest interface for that would seem be to just have a "remotes"
file that has multiple URL entries.
(For "pull", having multiple entries should either just select the first
one, or you could fall back on the others on failure - your choice).
And quite frankly, it just became too damn messy to do that in shell.
Besides, we actually have a fair amount of infrastructure in C, so it just
wasn't that hard to do.
Of course, this is almost totally untested. It probably doesn't work for
anything but the one trial I threw at it. "Simple" doesn't necessarily
mean "obviously correct".
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This brings another small repacking speedup for sensibly the same pack
size. On the Linux kernel repo, git-repack -a -f is 3.7% faster for a
0.4% larger pack.
Credits to Geert Bosch who brought the Rabin's polynomial idea to my
attention.
This also eliminate the issue of adler32() reading past the data buffer,
as noticed by Johannes Schindelin.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Also it learned to do -v (verbose) to report:
- number of loose objects
- disk occupied by loose objects
- number of objects in local packs
- number of loose objects that are also in pack
- unrecognised garbage in .git/objects/??/.
Signed-off-by: Junio C Hamano <junkio@cox.net>
For now let's retire this and reintroduce it as part of the updated
pack-objects series from Geert when it is ready.
Signed-off-by: Junio C Hamano <junkio@cox.net>
Foolishly I renamed diff.o around which caused an old diff.o
taken out of libgit.a and got linked into resulting binary and
exhibited mysterious breakage for many people. This borrows
from the kernel Makefile (scripts/Makefile.build) to first remove
the target and then recreate.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This was useful in diagnosing the corrupt index.aux format
problem. But do not bother building or installing it by
default.
Signed-off-by: Junio C Hamano <junkio@cox.net>
The cache_tree data structure is to cache tree object names that
would result from the current index file.
The idea is to have an optional file to record each tree object
name that corresponds to a directory path in the cache when we
run write_cache(), and read it back when we run read_cache().
During various index manupulations, we selectively invalidate
the parts so that the next write-tree can bypass regenerating
tree objects for unchanged parts of the directory hierarchy.
We could perhaps make the cache-tree data an optional part of
the index file, but that would involve the index format updates,
so unless we need it for performance reasons, the current plan
is to use a separate file, $GIT_DIR/index.aux to store this
information and link it with the index file with the checksum
that is already used for index file integrity check.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This is the first installment to libify diff brothers.
The updated diff-files uses revision.c::setup_revisions()
infrastructure to parse its command line arguments, which means
the pathname arguments are checked more strictly than before.
The tests are adjusted to separate possibly missing paths from
the rest of arguments with double-dashes, to show the kosher
way.
As Linus pointed out, renaming diff.c to diff-lib.c was simply
stupid, so I am renaming it back. The new diff-lib.c is to
contain pieces extracted from diff brothers.
Signed-off-by: Junio C Hamano <junkio@cox.net>
Right now it split it into "builtin-log.c" for log-related commands
("log", "show" and "whatchanged"), and "builtin-help.c" for the
informational commands (usage printing and "help" and "version").
This just makes things easier to read, I find.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Retire git-unresolve and make it into "git-update-index --unresolve".
It processes all paths that follow.
During a merge, you would mark a path that is dealt with with:
$ git update-index hello
and you would "undo" it with:
$ git update-index --unresolve hello
Signed-off-by: Junio C Hamano <junkio@cox.net>
This is an attempt to address the issue raised on #git channel
recently by Carl Worth.
After a conflicted automerge, "git diff" shows a combined diff
to give you how the tentative automerge result differs from
what came from each branch. During a complex merge, it is
tempting to be able to resolve a few paths at a time, mark
them "I've dealt with them" with git-update-index to unclutter
the next "git diff" output, and keep going. However, when the
final result does not compile or otherwise found to be a
mismerge, the workflow to fix the mismerged paths suddenly
changes to "git diff HEAD -- path" (to get a diff from our
HEAD before merging) and "git diff MERGE_HEAD -- path" (to get
a diff from theirs), and it cannot show the combined anymore.
With git-unresolve <paths>..., the versions from our branch and
their branch for specified blobs are placed in stage #2 and
stage #3, without touching the working tree files. This gives
you the combined diff back for easier review, along with
"diff --ours" and "diff --theirs".
One thing it does not do is to place the base in stage #1; this
means "diff --base" would behave differently between the run
immediately after a conflicted three-way merge, and the run
after an update-index by mistake followed by a git-unresolve.
We could theoretically run merge-base between HEAD and
MERGE_HEAD to find which tree to place in stage #1, but
reviewing "diff --base" is not that useful so....
Signed-off-by: Junio C Hamano <junkio@cox.net>
Peter Eriksen