Make pull fetch whatever is specified, parse it to figure out what it is, and
then process it appropriately. This also supports getting tag objects, and
getting whatever they tag.
Signed-off-by: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Handle parsing a tag for a non-present object. This adds a function to lookup
an object with lookup_* for * in a string, so that it can get the right storage
based on the "type" line in the tag.
Signed-off-by: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The confusion between "revs" vs "refs" caused us to not find the branch
name, which in turn meant that we never switched the HEAD over to it.

Both of these scripts will end up resetting the index to some specific
head, and any unresolved merge will be forgotten.

We codify the following different heads (in addition to the main "HEAD",
which points to the current branch, of course):

 - FETCH_HEAD

   Populated by "git fetch"

 - ORIG_HEAD

   The old HEAD before a "git pull/resolve" (successful or not)

 - LAST_MERGE

   The HEAD we're currently merging in "git pull/resolve"

 - MERGE_HEAD

   The previous head of a unresolved "git pull", which gets committed by
   a "git commit" after manually resolving the result

We used to have "MERGE_HEAD" be populated directly by the fetch, and we
removed ORIG_HEAD and LAST_MERGE too aggressively.

Here is a patch that fixes several gcc4 warnings about different signedness,
all between char and unsigned char. I tried to keep the patch minimal
so resertod to casts in three places.
Signed-off-by: NMika Kukkonen <mikukkon@iki.fi>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The "${new=$old}" syntax only works for an undefined 'new', not for an
empty one. I knew that. Really. I'm not stupid.

Now "git checkout xyzzy" will check out branch "xyzzy" and
switch the HEAD to it.

And fix braino on testing "-f".

It is careful by default and refuses to overwrite old info, but if you
want to force everything to be re-read, use the "-f" flag.

Some day I'll make it take individual filenames too. Right now
it's all-or-nothing.

Current version would spin forever and exhaust memory while attempting
to sort all files from all revisions at once, until it dies before even
doing any real work.  This is especially noticeable when used on a big
repository like the imported bkcvs repo for the Linux kernel.

This patch allows for batching the sort to put a bound on needed
resources and making progress early, as well as including some small
cleanups.
Signed-off-by: NNicolas Pitre <nico@cam.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Diffs with only mode changes didn't pass through git-apply --stat.

[ Linus' note: they did for me, on my ppc64, where division by zero just
  silently returns zero.  Duh.  ]
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

It does a "git-update-cache" on the arguments, meaning that you can
commit files without doing a separate "git-update-cache".  This commit
was done with

	git commit git-commit-script

for example.

It suppresses the "needs update" message for subsequent "--refresh"
cases (but not other errors).

Will be useful for "git commit". Stay tuned.

The ".pN" thing might be a common ending of a tag, and in
contrast, ^ already is a special character for revisions
so use that instead.

You can say "HEAD.p" for the "parent of HEAD". It nests, so

	HEAD.p2.p

means parent of second parent of HEAD (which obviously depends
on HEAD being a merge).

We used to keep the parents in reverse order in the commit_list.  Most
users don't care, but it's wrong, and the next commit does care.

The unexpected thing is likely a pathname, we need the default for that
too.

Make git-resolve-script only write MERGE_HEAD if a merge actually
occurred. All merge failures leave ORIG_HEAD and LAST_MERGE
behind (instead of ORIG_HEAD and MERGE_HEAD).

Use git-rev-parse to expand arguments (and check for bad ones).
Signed-off-by: NDan Holmsand <holmsand@gmail.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This fixes "<hexsha1>..*", since get_sha1() will happily ignore any
garbage at the end and thus we never got to the ".." check before.

Put explicit parentheses around && in ||-expression.

This patch fixes a problem reported by Paul Mackerras regarding the interaction
of the --merge-order and --max-age switches of git-rev-list.

This patch applies to the current Linus HEAD. A cleaner fix for the same problem
in my current HEAD will follow later.

With this change, --merge-order produces the same result as no --merge-order
on the linux-2.6 git repository, to wit:

$> git-rev-list --max-age=1116330140 bcfff0b471a60df350338bcd727fc9b8a6aa54b2 | wc -l
655
$> git-rev-list --merge-order --max-age=1116330140 bcfff0b471a60df350338bcd727fc9b8a6aa54b2 | wc -l
655
Signed-off-by: NJon Seymour <jon.seymour@gmail.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

If b is reachable from a, then:

	 git-rev-list a b

argument would print one of the commits twice.

This patch fixes that problem. A previous problem fixed it for the
--merge-order switch.
Signed-off-by: NJon Seymour <jon.seymour@gmail.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

If a is reachable from b, then git-rev-list --merge-order b a would
produce a duplicate output of b.

This causes a problem for an upcoming version of gitk since it
breaks the --merge-order ordering invariant.

This patch fixes the problem for the --merge-order switch. A subsequent
patch will fix the problem for the non --merge-order switch.
Signed-off-by: NJon Seymour <jon.seymour@gmail.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Given that real trees in the wild include parents with duplicate parents, I have relaxed
over-zealous error checking in epoch.c and dealt with the problem a different way - duplicate
parents are now silently ignored.
Signed-off-by: NJon Seymour <jon.seymour@gmail.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

git-rev-list --merge-order HEAD ^HEAD was faulting rather than generating an empty output.

This patch fixes that problem.
Signed-off-by: NJon Seymour <jon.seymour@gmail.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

      git-rev-list --merge-order --show-breaks root

Was outputing:

| root

It now outputs:

= root

Which is consistent with the behaviour of other cases.
Signed-off-by: NJon Seymour <jon.seymour@gmail.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

1. --merge-order doesn't deal properly with a specified head that has no parent

   * FAIL 11: head has no parent

2. --merge-order doesn't deal properly with arguments of the form
     head ^head

   * FAIL 30: head ^head --merge-order git-rev-list --merge-order
     --show-breaks a3 ^a3

3. if one of the specified heads is reachable from the other, the
   head gets printed twice and this causes problems for upcoming
   versions of gitk. This is true for both --merge-order and non
   --merge-order style of invocations.
    * FAIL 24: one specified head reachable from another a4, c3, --merge-order
    * FAIL 26: one specified head reachable from another a4, c3, no --merge-order
    * FAIL 27: one specified head reachable from another c3, a4, no --merge-order

4. --merge-order aborts with commits that list the same parent twice...it should handle it more gracefully.
    * no longer unit testable

5. broken interaction between --merge-order and --max-age
   previously posted as:

    "[PATCH 1/2] Test case that demonstrates problem with --merge-order, --max-age interaction"
    * FAIL 23: --max-age=c3, --merge-order

Later patches in this patch set fix these problems.
Signed-off-by: NJon Seymour <jon.seymour@gmail.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Patch for a completely rewritten file detected by the -B flag
was shown as a pair of creation followed by deletion in earlier
versions.  This was an misguided attempt to make reviewing such
a complete rewrite easier, and unnecessarily ended up confusing
git-apply.  Instead, show the entire contents of old version
prefixed with '-', followed by the entire contents of new
version prefixed with '+'.  This gives the same easy-to-review
for human consumer while keeping it a single, regular
modification patch for machine consumption, something that even
GNU patch can grok.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This updates diff documentation to discuss --find-copies-harder,
and adds descriptions for options that were not described
earlier.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Like diff-tree, this patch makes -C option for diff-* brothers
to use only pre-image of modified files as rename/copy detection
by default.  Give --find-copies-harder to use unmodified files
to find copies from as well.

This also fixes "diff-files -C" problem earlier noticed by
Linus.  It was feeding the null sha1 even when the file in the
work tree was known to match what is in the index file.  This
resulted in diff-files showing everything in the project.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Noted by Jeff.

Otherwise you'll be bitten by a stale MERGE_HEAD like Jeff was..

Add -u option to indicate incremental conversion.

I wanted to be able to track CVS repositories in a GIT repository.  The
cvs2git program worked fine with the initial import but needed a tiny
modification to enable me to resync the GIT repository with the updated
CVS tree.

[ The original version of this patch failed to track the correct
  branch on the first new commit. Fixed and tested by Sven. ]
Signed-off-by: NPanagiotis Issaris <takis@lumumba.luc.ac.be>
Signed-off-by: NSven Verdoolaege <skimo@kotnet.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Previous patch to fix host:path notation broke
handling of ssh://host/path notation.
Signed-off-by: NSven Verdoolaege <skimo@liacs.nl>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

But warn about them.  If somebody really ends up later wanting to
explicitly add a note that something has the same parent twice (who
knows, there are strange people around), we can add a flag to say that
it's expected and ok.

This was brought on by a commit in the kernel tree, where a repeated
merge caused a duplicate parent.

Parent duplicates aren't "wrong" per se, they're just in practice not
something you are ever interested in.

As suggested by Junio

Pointed out by Junio, part of my debugging of the rewrite of the
file/dir conflict handling.

This is (imho) more readable, and is also a lot faster.  The expense of
looking up sub-directory beginnings was killing us on things like
"git-diff-cache", even though that one didn't even care at all about the
file vs directory conflicts.

We really only care when somebody tries to add a conflicting name to
stage 0.

We should go through the conflict rules more carefully some day.

Strangely, this warning only shows up when not compiling with "-O2",
which is why I didn't see it originally.