Re: 1.3.0 creating bigger packs than 1.2.3

From: Junio C Hamano <junkio@cox.net>
Date: Apr 20, 2006, 17:24 UTC
Message-ID: <7vk69kyxbj.fsf@assigned-by-dhcp.cox.net>
In-Reply-To: <Pine.LNX.4.64.0604200954440.3701@g5.osdl.org>

Linus Torvalds <torvalds@osdl.org> writes:

> You can try "--depth=50" (slogan: more "hot delta on delta action"), but 
> it's looking less and less like a delta selection issue, and more and more 
> like the deltas themselves are deproved.

I do not think I have time to look into this today until late night, but one thing I noticed is that trying to delta more things sometimes tend to produce bigger result X-<.

At the end of pack-objects.c::find_deltas(), there is a code that is commented out, which is remnant from a failed experiment. What it tried to do was to avoid placing an object whose delta depth is already window-size back into the candidates list, which means the next object gets compared with one object more than otherwise would be (the extra one being the oldest one in the window -- which might not produce better delta than the maxed out one, but the delta with the maxed out one would not be used anyway). The result was noticeably worse overall packsize with more deltified objects.

We might be better off favoring compressed undeltified representation over deltified representation a bit more aggressively. Currently we allow delta as big as half the uncompressed size minus 20-byte overhead or something like that; tweaking that limit might show improvements.