Re: Gconfd and non-resident operation..

[Tom Lisjac <netdxr adelphia net>]

On Sun, 2004-01-25 at 09:23, Havoc Pennington wrote:
> On Tue, 2004-01-20 at 04:16, Tom Lisjac wrote:
> > 
> > This is a crude solution at best, but the prototype systems perform well
> > and have all run for several weeks without any problems.

> In principle this isn't needed on a couple of levels, one is that the
> kernel will swap out unused data, and the other is that gconfd itself
> will drop its cache over time. Neither of these things is necessarily
> reflected in "top" though.

Hmmm... didn't realize that! Perhaps we're putting too much emphasis on
getting the "numbers" down and not being objective enough about
observing the real performance.

> What you're doing probably works fine, though it's a fairly major hack
> ;-)

Yes, we know!

Thanks for the reply... and comforting words! :)