From: Jonathan Nieder <jrnieder@gmail.com>
Date: Sat, 20 Aug 2011 02:15:15 GMT
Subject: Re: update-index --index-info producing spurious submodule commits
Message-ID: <20110820021430.GA14281@elie.sbx02827.chicail.wayport.net>
In-Reply-To: <rmi39gxacp1.fsf@fnord.ir.bbn.com>

Greg Troxel wrote:
> Junio C Hamano <gitster@pobox.com> writes:

>> "ls-tree -r HEAD foo" is probably what you meant to say.
>
> Thanks very much for the clue - that works.  The update-index
> documentation should probably say that only blobs (or perhaps commits
> intended to be submodules??) are acceptable, and perhaps say "ls-tree
> -r" instead of ls-tree.

Makes sense.  Please make it so.

By the way, for this particular application I wonder if something like

	git ls-files -z <dir> | git update-index -z --force-remove --stdin
	git read-tree --prefix=<dir>/ <tree>

would be easier.  Or a commit-filter. :)

	tree=$1
	shift
	tree=$(
		git ls-tree -z "$tree" |
		perl -0ne '
			chop;
			my ($info, $name) = split(/\t/, $_, 2);
			if ($name eq "<dir>") {
				printf("040000 tree <good tree>\t<dir>\0");
			} else {
				printf("%s\0", $_);
			}
		' |
		git mktree -z
	)
	git commit-tree "$tree" "$@"

Thanks,
Jonathan