Functions that do many things are bad. We should basically
just parse the arguments in main(). We're not quite there
yet, but it's a step in the right direction.

People are making fun of me for being a bad speeler.

Make a separate helper for parsing the header of an object file
(really carefully) and for unpacking the rest. This means that
anybody who uses the "unpack_sha1_header()" interface can easily
look at the header and decide to unpack the rest too, without
doing any extra work.

It's for people who aren't necessarily interested in the whole
unpacked file, but do want to know the header information (size,
type, etc..)

For example, the delta code can use this to figure out whether
an object is already a delta object, and what it is a delta
against, without actually bothering to unpack all of the actual
data in the delta.

Both locally and remotely.

The core GIT repository has trees that record regular file mode
in 0664 instead of normalized 0644 pattern.  Comparing such a
tree with another tree that records the same file in 0644
pattern without content changes with git-diff-tree causes it to
feed otherwise unmodified pairs to the diff_change() routine,
which triggers a sanity check routine and barfs.  This patch
fixes the problem, along with the fix to another caller that
uses unnormalized mode bits to call diff_change() routine in a
similar way.

Without this patch, you will see "fatal error" from diff-tree
when you run git-deltafy-script on the core GIT repository
itself.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Use "git commit" instead of "git-commit-script", and talk about using
"git log" before introducing the more complex "git-whatchanged".

In short, try to make it feel a bit more normal to those poor souls
using CVS.

Do some whitspace edits too, to make the side notes stand out a bit
more.

The "git" script is just shorthand: "git xyz <args>" will just execute
"git-xyz-script <args>", which is useful for people used to the CVS
naming convention. So "git log" will run the new git-log-script, which
is just a wrapper around the new pretty-printing git-rev-list.

Cheesy.

That pretty-prints the resulting commit messages, so

	git-rev-list --pretty HEAD v2.6.12-rc5 | less -S

basically ends up being a log of the changes between -rc5
and current head.

It uses the pretty-printing helper function I just extracted
from diff-tree.c.

It's really just the header printign function from diff-tree.c,
and it's usable for other things too.

Pointed out by Junio. I kant't speel.

It had already tried to do that, but with the independent
rounding of the number of '+' and '-' characters, it would
sometimes do 80-char lines after all.

This fixes the problem with ls-tree which failed to show
"drivers/char" directory when the user asked for "drivers/char/"
from the command line.  At the same time, if "drivers/char" were
a non directory, "drivers/char/" would not show it.  This is
consistent with the way diffcore-pathspec has been recently
fixed.

This adds back the diffcore-pathspec test,dropped when my
earlier diffcore-pathspec fix was rejected.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This really is very basic stuff, no branches, no merging, no CVS
imports. Let's start small.

There were duplicate script pieces to help comparing diff
output, which this patch consolidates into the t/diff-lib.sh
library.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Removing (and ignoring) them is wrong, since that means
that a pathspec of "xxxx/" would match a regular filename
of "xxxx", which is obviously incorrect.

It just causes things like "git-apply --stat" to parse traditional
patch headers differently depending on what your index is, which
is nasty.

It makes rev-list show the list of parents, the same
way git-rev-tree does (but without the expense).

This mean sthat you can give a beginning/end pair to git-rev-list,
and it will show all entries that are reachable from the beginning
but not the end.

For example

	git-rev-list v2.6.12-rc5 v2.6.12-rc4

shows all commits that are in -rc5 but are not in -rc4.

This is useful for when we want to insert the next one after
this new one, for example.

The way broken deletes and creates are shown in the -p
(diff-patch) output format has become consistent with how
rename/copy edits are shown.  They will show "dissimilarity
index" value, immediately following the "deleted file mode" and
"new file mode" lines.

The git-apply is taught to grok such an extended header.
Signed-off-by: NJunio C Hamano <junkio@cox.net>

A new diffcore filter diffcore-order is introduced.  This takes
a text file each of whose line is a shell glob pattern.  Patches
that match a glob pattern on an earlier line in the file are
output before patches that match a later line, and patches that
do not match any glob pattern are output last.

A typical orderfile for git project probably should look like
this:

    README
    Makefile
    Documentation
    *.h
    *.c
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Should be obvious...

 - Use $VISUAL, $EDITOR, in this order if set, and fall back on
   vi.

 - Status R, C, D, N usually are followed by number, so adjust
   case arms to that pattern.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Not everybody necessarily even has diffstat installed.

Although it was described as such, git-mkdelta didn't really attempt to
find the best delta against any previous object in the list, but was
only able to create a delta against the preceeding object.  This patch
reworks the code to fix that limitation and hopefully makes it a bit
clearer than before, including fixing the delta loop detection which was
broken.

This means that

	git-mkdelta sha1 sha2 sha3 sha4 sha5 sha6

will now create a sha2 delta against sha1, a sha3 delta against either
sha2 or sha1 and keep the best one, a sha4 delta against either sha3,
sha2 or sha1, etc.  The --max-behind argument limits that search for the
best delta to the specified number of previous objects in the list.  If
no limit is specified it is unlimited (note: it might run out of
memory with long object lists).

Also added a -q (quiet) switch so it is possible to have 3 levels of
output: -q for nothing, -v for verbose, and if none of -q nor -v is
specified then only actual changes on the object database are shown.

Finally the git-deltafy-script has been updated accordingly, and some
bugs fixed (thanks to Stephen C. Tweedie for spotting them).

This version has been toroughly tested and I think it is ready
for public consumption.
Signed-off-by: NNicolas Pitre <nico@cam.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This is meant to make raw git not hugely less usable than something
like raw CVS. I want to make a 1.0 release of the plumbing, and the
actual commit part was just too intimidating.

A new diffcore transformation, diffcore-break.c, is introduced.

When the -B flag is given, a patch that represents a complete
rewrite is broken into a deletion followed by a creation.  This
makes it easier to review such a complete rewrite patch.

The -B flag takes the same syntax as the -M and -C flags to
specify the minimum amount of non-source material the resulting
file needs to have to be considered a complete rewrite, and
defaults to 99% if not specified.

As the new test t4008-diff-break-rewrite.sh demonstrates, if a
file is a complete rewrite, it is broken into a delete/create
pair, which can further be subjected to the usual rename
detection if -M or -C is used.  For example, if file0 gets
completely rewritten to make it as if it were rather based on
file1 which itself disappeared, the following happens:

    The original change looks like this:

	file0     --> file0' (quite different from file0)
	file1     --> /dev/null

    After diffcore-break runs, it would become this:

	file0     --> /dev/null
	/dev/null --> file0'
	file1     --> /dev/null

    Then diffcore-rename matches them up:

	file1     --> file0'

The internal score values are finer grained now.  Earlier
maximum of 10000 has been raised to 60000; there is no user
visible changes but there is no reason to waste available bits.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The commit 15d061b4

    [PATCH] Fix the way diffcore-rename records unremoved source.

still leaves unneeded delete records in its output stream by
mistake, which was covered up by having an extra check to turn
such a delete into a no-op downstream.  Fix the check in the
diffcore-rename to simplify the output routine.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

When preparing data to feed the external diff, we should give
the mode we obtained from the caller, even when we are dealing
with a file with 0{40} SHA1 (i.e. the caller said "look at the
filesystem"), since the mode passed by the caller via
diff_addremove() or diff_change() is always trustworthy.

This is _not_ a bugfix --- the existing code stat() on the file
ifself and does the same computation on st.st_mode to compute
the mode the same way the caller did to give the original mode.
We cannot remove the stat() call from here, but the extra
computation to create the mode value is unnecessary.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

A new macro, DIFF_PAIR_RENAME(), is introduced to distinguish a
filepair that is a rename/copy (the definition of which is src
and dst are different paths, of course).  This removes the hack
used in the record_rename_pair() to always put a non-zero value
in the score field.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The three diff-* brothers had a sequence of calls into diffcore
that were almost identical.  Introduce a new diffcore_std()
function that takes all the necessary arguments to consolidate
it.  This will make later enhancements and changing the order of
diffcore application simpler.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The score number that follow R/C status were parsed but the
parse pointer was not updated, causing the entire line to become
unrecognized.  This patch fixes this problem.

There was a test missing to catch this breakage, which this
commit adds as t4009-diff-rename-4.sh.  The diff-raw tests used
in related t4005-diff-rename-2.sh (the same test without -z) and
t4007-rename-3.sh were stricter than necessarily, despite that
the comment for the tests said otherwise.  This patch also
corrects them.

The documentation is updated to say that the status can
optionally be followed by a number called "score"; it does not
have to stay similarity index forever and there is no reason to
limit it only to C and R.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Create .git/refs/{heads,tags} and make .git/HEAD be a symlink to
(the as yet non-existent) .git/refs/heads/master.

Instead of always assuming it can be read with a single
read() system call, loop around properly.

Pointed out by Pasky, but I ended up implementing it differently
from his suggested patch.

The count-delta routine sometimes overcounted the copied source
material which resulted in unsigned int wraparound.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Some source files were including "delta.h" without actually
needing it.  Remove them.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

A bug in the command line argument parsing code was making
pickaxe not to work at all in diff-cache and diff-files commands.
Embarrassingly enough, the working pickaxe in diff-tree tells me
that it was not working in these two commands from day one.
This patch fixes it.

Also updates the documentation to describe the --pickaxe-all option.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This is a complete rewrite of ls-tree to make it behave more
like what "/bin/ls -a" does in the current working directory.

Namely, the changes are:

 - Unlike the old ls-tree behaviour that used paths arguments to
   restrict output (not that it worked as intended---as pointed
   out in the mailing list discussion, it was quite incoherent),
   this rewrite uses paths arguments to specify what to show.

 - Without arguments, it implicitly uses the root level as its
   sole argument ("/bin/ls -a" behaves as if "." is given
   without argument).

 - Without -r (recursive) flag, it shows the named blob (either
   file or symlink), or the named tree and its immediate
   children.

 - With -r flag, it shows the named path, and recursively
   descends into it if it is a tree.

 - With -d flag, it shows the named path and does not show its
   children even if the path is a tree, nor descends into it
   recursively.

This is still request-for-comments patch.  There is no mailing
list consensus that this proposed new behaviour is a good one.

The patch to t/t3100-ls-tree-restrict.sh illustrates
user-visible behaviour changes.  Namely:

 * "git-ls-tree $tree path1 path0" lists path1 first and then
   path0.  It used to use paths as an output restrictor and
   showed output in cache entry order (i.e. path0 first and then
   path1) regardless of the order of paths arguments.

 * "git-ls-tree $tree path2" lists path2 and its immediate
   children but having explicit paths argument does not imply
   recursive behaviour anymore, hence paths/baz is shown but not
   paths/baz/b.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The latest change to diff-tree -z output adds an extra line
termination after non diff-raw material (the header and the
commit message).  To compensate for this change, stop adding the
output termination of our own.  "diff-tree -v -z" piped to
"diff-helper -z" would give different result from "diff-tree -v"
piped to "diff-helper" without this change.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>