> there seems to be some resistence among
> certain NetBSD developers against it.

 mb> Not against pf itself, but the way it interfaces with altq and
 mb> ipsec.  We want to be able to use others packet classifiers than
 mb> pf for theses.

I couldn't find the thread with details about the problems with PF in
the tech-net archives to which you referred, just a post from itojun
in October.  I don't want to reopen an old argument without the
background.  Do you remember the approximate date of the thread?

Anyway, will there be no option to use PF as an ALTQ classifier?  If
so, what's the case for the current ALTQ classifier?  

One of the ways the current situation sucks compared to OpenBSD is
that you cannot do things like prioritize TCP ACKs with NetBSD ALTQ.
It might also be nice to, for example, assign TCP flows to classes
using the NAT proxies like ftp, the ones used for ``keep state''
firewalls.  The ALTQ classifier is so primitive---is there some reason
it's preferred over PF?

-- 
Some people in the US don't like the UN and will never like the UN
because it's full of foreigners, which frankly can't be helped.
             -- Madeline Albright