Re: Preferring shallower deltas on repack

From: Brian Downing <bdowning@lavos.net>
Date: Jul 9, 2007, 19:24 UTC
Message-ID: <20070709192403.GM4087@lavos.net>
In-Reply-To: <alpine.LFD.0.999.0707091503000.26459@xanadu.home>

On Mon, Jul 09, 2007 at 03:13:54PM -0400, Nicolas Pitre wrote:

Show 11 quoted lines

> Tuning for such an extreme without impacting normal cases is rather 
> hard.
> 
> My patch was meant to be used on top of yours.  Is that what you tested?
> 
> Also I'd suggest you do not use a max depth of 1000.  It is simply 
> insane and might possibly make the existing logic less effective.  Even 
> for runtime pack access you want it to be reasonably short, say 100 
> maximum, or even the current default of 50.  Any improvements you might 
> come with (like automatic depth determined on replay cost) should be 
> compared with that default which is known to work well already.

No, I didn't try it on top of mine; sorry.  I'll try that out.

I realize a depth of 1000 is nuts, I'm just using it to expose any suboptimal delta selection for the nasty versions.lisp-expr case, since I know that with a window size of 100 it should be easy to keep it from growing too deep.

-bcd