It's just as much a work-in-progress, but at least now it's gotten
enough technical review to shake out most of the really bad lies, so
hopefully it doesn't do any actual damage.  And if we encourage people
to read it, they'll be more likely to whine about it, which will help
get it fixed faster.
Signed-off-by: N"J. Bruce Fields" <bfields@citi.umich.edu>

I cherry-picked some additional miscellaneous fixes from those suggested
by Santi Béjar, including fixes to:

	- correct discussion of repository/HEAD->repository shortcut
	- add mention of git-mergetool
	- add mention of --track
	- mention "-f" as well as "+" for fetch

Cc: Santi Béjar <sbejar@gmail.com>
Signed-off-by: N"J. Bruce Fields" <bfields@citi.umich.edu>

Santi Béjar points out that when telling people how to "introduce
themselves" to git we're advising them to replace their entire
.gitconfig file.  Fix that.

Cc: "Santi Béjar <sbejar@gmail.com>
Signed-off-by: N"J. Bruce Fields" <bfields@citi.umich.edu>

The previous commit calls attention to the fact that we have two
sections each devoted to fast-forwards and to dangling objects.  Revise
and attempt to differentiate them a bit.  Some more reorganization may
be required later....

Signed-off-by: J. Bruce Fields

Any section lacking an id gets an annoying warning when you build
the manual.  More seriously, the table of contents then generates
volatile id's which change with every build, with the effect that
we get URL's that change all the time.

The ID's are manually generated and sometimes inconsistent, but
that's OK.

XXX: what to do about the preface?
Signed-off-by: N"J. Bruce Fields" <bfields@citi.umich.edu>

Nicolas Pitre pointed out a couple typos and some room for improvement
in the discussion of detached heads.
Signed-off-by: N"J. Bruce Fields" <bfields@citi.umich.edu>
Cc: Nicolas Pitre <nico@cam.org>

Check if a line of the header has enough characters to possibly
contain the requested prefix.
Signed-off-by: NAlex Riesen <raa.lkml@gmail.com>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Signed-off-by: NJames Bowes <jbowes@dangerouslyinc.com>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Signed-off-by: NJunio C Hamano <junkio@cox.net>

When you do this, existing "blame -C -C" would not find that the
latter half of the file2 came from the existing file1:

	... both file1 and file2 are tracked ...
	$ cat file1 >>file2
	$ git add file1 file2
	$ git commit

This is because we avoid the expensive find-copies-harder code
that makes unchanged file (in this case, file1) as a candidate
for copy & paste source when annotating an existing file
(file2).  The third -C now allows it.  However, this obviously
makes the process very expensive.  We've actually seen this
patch before, but I dismissed it because it covers such a narrow
(and arguably stupid) corner case.
Signed-off-by: NJunio C Hamano <junkio@cox.net>

The -C option to blame tries to find a section of a preimage
file by running diff against the lines whose origin is still
unknown, and excluding the different parts.  The code however
did not cover the case where the tail part of the section
matched, which we handle for the normal non-move/copy codepath.

This breakage was most visible when preimage file matches in its
entirety and failed to pass blame in such a case.
Signed-off-by: NJunio C Hamano <junkio@cox.net>

"git log --boundary" incorrectly honoured the option only when
"left-right" was enabled.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Add description of raw combined diff format to diff-formats.txt,
as "diff format for merges" section, before "Generating patches..."
section.
Signed-off-by: NJakub Narebski <jnareb@gmail.com>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Most other documentation will frequently be read from an installation
of git so will naturally be associated with the installed version.
But these two documents in particular are often read from web pages
while users are still exploring git. It's important to mention
version 1.5.1 since these documents provide example commands that
won't work with previous versions of git.
Signed-off-by: NJunio C Hamano <junkio@cox.net>

git-svn dcommit exports commits to Subversion, then imports them back
to git again, and last but not least rebases or resets HEAD to the
last of the new commits. I guess this rebasing is convenient when
using just git, but when the commits to be exported are managed by
StGIT, it's really annoying. So add an option to disable this
behavior. And document it, too!
Signed-off-by: NKarl Hasselström <kha@treskal.com>
Acked-by: NEric Wong <normalperson@yhbt.net>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Some of the existing markup was just plain broken, and some subcommand
options weren't indented properly.
Signed-off-by: NKarl Hasselström <kha@treskal.com>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Using decode() tries to decode data that is already UTF-8 and
borks, but decode_utf8 from Encode.pm has a built-in safety
against that.
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Fix t4200 so that it also works on OS X by not relying on gnu
extensions to sed.
Signed-off-by: NBryan Larsen <bryan@larsen.st>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Signed-off-by: NArjen Laarhoven <arjen@yaph.org>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

There is a mechanism PERL_PATH in the Makefile to specify path to
Perl binary, but sometimes it is convenient to let 'env' figure
out where Perl comes from, with PERL_PATH="/usr/bin/env perl".

Allowing this would make things easier to MacPorts, where we wish
to work with the MacPorts perl if it is installed, but fall back
to the system perl if it isn't.
Signed-off-by: NBryan Larsen <bryan@larsen.st>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

This also improves the implementation to match how strndup is
specified (by GNU): if the length given is longer than the string,
only the string's length is allocated and copied, but the string need
not be null-terminated if it is at least as long as the given length.
Signed-off-by: NDaniel Barkalow <barkalow@iabervon.org>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

We broke the size-cache handling when we changed the function
signature of sha1_object_info() in 21666f1a.  We obviously
wanted to cache the size we obtained when sha1_object_info()
succeeded, not when it failed.
Signed-off-by: NJunio C Hamano <junkio@cox.net>

curl_multi_remove_handle() is broken in libcurl < 7.16, in that it
doesn't correctly update the active handles count when a request is
aborted. This causes the transfer to hang forever waiting for the
handle count to become less than the number of active requests.
Signed-off-by: NAlexandre Julliard <julliard@winehq.org>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

* 'maint' of git://repo.or.cz/git/fastimport:
  Teach import-tars about GNU tar's @LongLink extension.

We can't unconditionally assign revision 1.1 to
newly added files. In case the file did exist in the
past and was deleted we need to honor the old
revision number.
Signed-off-by: NFrank Lichtenheld <frank@lichtenheld.de>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

This extension allows GNU tar to process file names in excess of the 100
characters defined by the original tar standard. It does this by faking a
file, named '././@LongLink' containing the true file name, and then adding
the file with a truncated name. The idea is that tar without this
extension will write out a file with the long file name, and write the
contents into a file with truncated name.

Unfortunately, GNU tar does a lousy job at times. When truncating results
in a _directory_ name, it will happily use _that_ as a truncated name for
the file.

An example where this actually happens is gcc-4.1.2, where the full path
of the file WeThrowThisExceptionHelper.java truncates _exactly_ before the
basename. So, we have to support that ad-hoc extension.

This bug was noticed by Chris Riddoch on IRC.
Signed-off-by: NJohannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: NShawn O. Pearce <spearce@spearce.org>

The code used write_in_full() without pulling its declarations from the
header file.  When header is included, usage[] collides with usage()
function.
Signed-off-by: NMartin Koegler <mkoegler@auto.tuwien.ac.at>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Signed-off-by: NJunio C Hamano <junkio@cox.net>

It can be either hostname/address, or a full path to a
local executable.
Signed-off-by: NJari Aalto <jari.aalto@cante.net>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Signed-off-by: NJari Aalto <jari.aalto@cante.net>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Separate things to be checked when making commits, and things
to be checked when sending patches.
Signed-off-by: NJari Aalto <jari.aalto@cante.net>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

This patch adds a new 'find-rev' command to git-svn that lets you easily
translate between SVN revision numbers and git tree-ish.
Signed-off-by: NAdam Roben <aroben@apple.com>
Acked-by: NEric Wong <normalperson@yhbt.net>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

After reading the leading contents from a symlink data obtained
from subversion, which we expect to begin with 'link ', the code
forked to hash the remainder (which should match readlink()
result) using git-hash-objects, by redirecting its STDIN from
the filehandle we read that 'link ' from.  This was Ok with Perl
on modern Linux, but on Mac OS, the read in the parent process
slurped more than we asked for in stdio buffer, and the child
did not correctly see the "remainder".

This attempts to fix the issue by using lower level sysseek and
sysread instead of seek and read to bypass the stdio buffer.
Signed-off-by: NJunio C Hamano <junkio@cox.net>
Acked-by: NEric Wong <normalperson@yhbt.net>
Acked-by: NSeth Falcon <sethfalcon@gmail.com>

Calling http_init after calling http_cleanup causes a segfault.  This
is due to the pragma_header curl_slist being freed but not being set
to NULL.  The subsequent call to http_init tries to setup the slist
again, but it now points to an invalid memory location.
Signed-off-by: NJulian Phillips <julian@quantumfyre.co.uk>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Signed-off-by: NJosh Triplett <josh@freedesktop.org>
Signed-off-by: NJunio C Hamano <junkio@cox.net>

This heeds advice from our resident Perl expert to make sure
the script is not confused with a string that ends with /\n
Signed-off-by: NJunio C Hamano <junkio@cox.net>

Signed-off-by: NJunio C Hamano <junkio@cox.net>

Reflog message is primarily about easier identification, and
leaving truncated entry is much better than dying.
Signed-off-by: NJunio C Hamano <junkio@cox.net>

* 'maint' of git://repo.or.cz/git/fastimport:
  Don't allow empty pathnames in fast-import
  import-tars: be nice to wrong directory modes

Released versions of fast-import have been able to create a tree that
contains files or subtrees that contain no name.  Unfortunately these
trees aren't valid, but people may have actually tried to create them
due to bugs in import-tars.perl or their own fast-import frontend.

We now look for this unusual condition and warn the user if at
least one of their tree objects contains the problem.
Signed-off-by: NShawn O. Pearce <spearce@spearce.org>
Signed-off-by: NJunio C Hamano <junkio@cox.net>