On Mon, Jun 30, 2003 at 08:57:36AM -0700, Jason Thorpe wrote:
> 
> Even if it is in "serious flux" (which I don't think it is), there has 
> traditionally not been a problem with checking in functional 
> works-in-progress into NetBSD-current.

Usually, work in progress is checked in after the API issues have been sorted
out. IMHO it's not yet the case. The way altq can interract with
packet classifiers isn't clear.

Don't misunderstand me, I'll be happy to see PF in NetBSD.
It's a big win to have 2 different filtering engine available.
But one should not be mandatory to use others parts (altq, ipsec) of the
kernel because the API between these different parts don't allow it;
especially when these other parts have been abailable for a long time
as standalone features.
In this specific case, if the code doesn't exists to use altq or ipsec
from ipf, it's not a big deal (just write the code) as long as the
ipsec and altq API allows it. Getting the API right before is much easier
than tweaking it after.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 24 ans d'experience feront toujours la difference
--