This uses the reworked diff interface to generate patches directly out
of show-diff when -p is specified. 
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This introduces three public functions for diff-cache and friends can
use to call out to the GIT_EXTERNAL_DIFF program when they wish to.

A normal "add/remove/change" entry is turned into 7-parameter process
invocation of GIT_EXTERNAL_DIFF program as before.  In addition, the
program can now be called with a single parameter when diff-cache and
friends want to report an unmerged path. 
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

We now modify the in-memory copy of the index file in "diff-cache", so
we need to add PROT_WRITE. 

We've always warned about them properly, but we would then do the
wrong thing if that filename existed in the tree we were comparing
against (we'd think the file has been deleted, because we forgot
about the unmerged cases).

This makes it match "show-diff" behaviour.

This makes diff-tree-helper handle ("warn about") unmerged path entries

    U <path> <record-terminator>

This is emitted once per unmerged path, no matter how many unmerged
stages there are. 
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

It really doesn't make sense, since there are potentially _many_ different
cache entries for an unmerged file. So just do the "U pathname" thing.

You'll need "diff-tree-helper" to show the full diff, but Junio is
dead set on adding a "-p" argument to all three to avoid it. That's
next..

When the trivial "read-tree" merge fails, fall back on the (equally
trivial) automatic merge script instead of forcing the user to do
it by hand.

When _that_ fails, you get to do a manual merge.

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: NLinus Torvalds <torvalds@osdl.org>

This makes it a lot more pleasant to use when you
interrupt a long-running operation.

diff-cache attempts to first remove all merge entries before letting the
diff_cache() do its work, but it incorrectly stops after the first
merge-entry it finds.

Fix by just replacing the "break" with a "continue".
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This reworks the diff-tree-helper and show-diff to further make external
diff command interface simpler.

These commands now honor GIT_EXTERNAL_DIFF environment variable which
can point at an arbitrary program that takes 7 parameters:

  name file1 file1-sha1 file1-mode file2 file2-sha1 file2-mode

The parameters for an external diff command are as follows:

  name        this invocation of the command is to emit diff
	      for the named cache/tree entry.

  file1       pathname that holds the contents of the first
	      file.  This can be a file inside the working
	      tree, or a temporary file created from the blob
	      object, or /dev/null.  The command should not
	      attempt to unlink it -- the temporary is
	      unlinked by the caller.

  file1-sha1  sha1 hash if file1 is a blob object, or "."
	      otherwise.

  file1-mode  mode bits for file1, or "." for a deleted file.

If GIT_EXTERNAL_DIFF environment variable is not set, the
default is to invoke diff with the set of parameters old
show-diff used to use.  This built-in implementation honors the
GIT_DIFF_CMD and GIT_DIFF_OPTS environment variables as before.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This patch introduces a new program, diff-tree-helper.  It reads
output from diff-cache and diff-tree, and produces a patch file.
The diff format customization can be done the same way the
show-diff uses; the same external diff interface introduced by
the previous patch to drive diff from show-diff is used so this
is not surprising.

It is used like the following examples:

   $ diff-cache --cached -z <tree> | diff-tree-helper -z -R paths...
   $ diff-tree -r -z <tree1> <tree2> | diff-tree-helper -z paths...

 - As usual, the use of the -z flag is recommended in the script
   to pass NUL-terminated filenames through the pipe between
   commands.

 - The -R flag is used to generate reverse diff.  It does not
   matter for diff-tree case, but it is sometimes useful to get
   a patch in the desired direction out of diff-cache.

 - The paths parameters are used to restrict the paths that
   appears in the output.  Again this is useful to use with
   diff-cache, which, unlike diff-tree, does not take such paths
   restriction parameters.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

With this patch, the non-core'ish part of show-diff command that
invokes an external "diff" comand to obtain patches is split
into a separate file.  The next patch will introduce a new
command, diff-tree-helper, which uses this common diff interface
to format diff-tree and diff-cache output into a patch form.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This makes the default fsck behaviour be quiet for a repository
that doesn't have any problems. Which is good.

Normally we don't care, we just check them for being valid tag
objects.

the name of the object it tags.

You need this if you actually want to build up a list of tags.

You use "git-mktag" to create them, and fsck-cache knows how to parse them.

Use "unsigned long" for the size, like we do everywhere else.

The write function now adds the header to the file by itself, so there
is no reason to duplicate it among all the users any more.

Now that we have hopefully converted all old archives, we
can consider it an error.

The git archives have some old-date-format commits with timezones
that the converter didn't recognize. Also, make it be quiet about
already-converted dates.

This adds an --ignore-missing option to update-cache, which makes it
ignore missing files.  Together with the "-n" option to checkout-cache,
it allows me to do

    checkout-cache -n -f -a && update-cache --ignore-missing --refresh

which only updates and refreshes the files I already have checked out.
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This adds the "-n" option to checkout-cache which tells it to not check
out new files, only refresh files already checked out.
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

That would SIGSEGV.

The type doesn't come from the parsing, the type also has to match the usage.

Otherwise we can have objects without a type, which is not good.

We should _not_ mark a blob object "parsed" just because we
looked it up: it gets marked that way only once we've actually
seen it. Otherwise we can never notice a missing blob.

Here is how to trigger it:

   echo blob 100 > .git/objects/00/ae4e8d3208e09f2cf7a38202a126f728cadb49

Then run fsck-cache. It will try to unpack after the header to calculate 
the hash, inflate returns total_out == 0 and memcpy() dies.

The patch below seems to work with ZLIB 1.1 and 1.2.
Signed-off-by: NAndreas Gal <gal@uci.edu>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This is based on a patch by David Woodhouse, but with the selection
tests much simplified and streamlined.

It makes diff-tree take extra arguments, specifying the files or         
directories which should be considered "interesting". Changes in
uninteresting directories are not reported.
Signed-off-by: NDavid Woodhouse <dwmw2@infradead.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

1) permissions aren't respected in the merge script (primarily because
they're never passed in to it in the first place).  Fix that and also
check for permission conflicts in the merge

2) the delete of a file in both branches may indeed be just that, but it
could also be the indicator of a rename conflict (file moved to
different locations in both branches), so error out and ask the
committer for guidance.
Signed-off-by: NJames Bottomley <James.Bottomley@SteelEye.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Make pop_most_recent_commit() return the same objects multiple times, but only
if called with different bits to mark.

This is necessary to make merge-base work again.
Signed-Off-By: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

seen them.

Otherwise any merges will make the parent list explode.

This is probably what you'd want to see for "git log".

This patch adds three similar and related programs. http-pull downloads
objects from an HTTP server; rpull downloads objects by using ssh and
rpush on the other side; and rpush uploads objects by using ssh and rpull
on the other side.

The algorithm should be sufficient to make the network throughput required
depend only on how much content is new, not at all on how much content the
repository contains.

The combination should enable people to have remote repositories by way of
ssh login for authenticated users and HTTP for anonymous access.
Signed-Off-By: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The old implementation was a nice algorithm, but, unfortunately, it could
be confused in some cases and would not necessarily do the obvious thing
if one argument was decended from the other. This version fixes that by
changing the criterion to the most recent common ancestor.
Signed-Off-By: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This adds two functions: one to check if an object is present in the local
database, and one to add an object to the local database by reading it
from a file descriptor and checking its hash.
Signed-Off-By: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>