From: Jeff King <peff@peff.net>
Date: Tue, 31 Aug 2010 16:01:45 GMT
Subject: Re: [RFC] notes: avoid recommitting identical trees
Message-ID: <20100831160145.GC11014@sigill.intra.peff.net>
In-Reply-To: <bc44b3393db4018487bb956d00a12fa73f04ca9e.1283267564.git.git@drmicha.warpmail.net>

On Tue, Aug 31, 2010 at 05:16:17PM +0200, Michael J Gruber wrote:

> Currently, "git notes" behaves like "git commit --allow-empty" when
> committing notes trees. In particular, removing nonexisting notes leads
> to empty commits "commits with no diff".
> 
> Change this to avoid unnecessary notes commits.

Is this a sufficient check in the case of notes? Is it possible that we
re-balanced the fanout of the notes tree and got a different tree sha1,
even though there is nothing interesting to commit?

> +	if (!parent || parse_commit(parent->item) || parse_tree(parent->item->tree) ||
> +		hashcmp(parent->item->tree->object.sha1, tree_sha1)) {

I didn't check, but I can imagine you can drop the parse_tree here. We
should know the object sha1 once the commit is parsed.

-Peff