From owner-freebsd-hackers  Mon Sep  8 20:54:24 1997
Return-Path: <owner-freebsd-hackers>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id UAA29725
          for hackers-outgoing; Mon, 8 Sep 1997 20:54:24 -0700 (PDT)
Received: from whistle.com (s205m131.whistle.com [207.76.205.131])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id UAA29720
          for <freebsd-hackers@FreeBSD.ORG>; Mon, 8 Sep 1997 20:54:22 -0700 (PDT)
Received: (from smap@localhost) by whistle.com (8.7.5/8.6.12) id UAA17296; Mon, 8 Sep 1997 20:53:46 -0700 (PDT)
Received: from bubba.whistle.com(207.76.205.7) by whistle.com via smap (V1.3)
	id sma017294; Mon Sep  8 20:53:34 1997
Received: (from archie@localhost) by bubba.whistle.com (8.8.5/8.6.12) id UAA26952; Mon, 8 Sep 1997 20:53:34 -0700 (PDT)
From: Archie Cobbs <archie@whistle.com>
Message-Id: <199709090353.UAA26952@bubba.whistle.com>
Subject: Re: Divert sockets..
In-Reply-To: <19970909121148.26417@lemis.com> from Greg Lehey at "Sep 9, 97 12:11:48 pm"
To: grog@lemis.com (Greg Lehey)
Date: Mon, 8 Sep 1997 20:53:34 -0700 (PDT)
Cc: doconnor@ist.flinders.edu.au, freebsd-hackers@FreeBSD.ORG
X-Mailer: ELM [version 2.4ME+ PL31 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk


> >> IMHO, the proper way to do this is to let the PPP daemon handle it
> >> by installing a route first (so you let the routing code determine
> >> whether packets are supposed to go over the WAN link or not, as it
> >> should) and then checking each outgoing packet for suitability as
> >> "demand" (not all are, e.g., NTP packets). When "demand" is seen,
> >> it should start dialing, etc. The same "demand" test can also apply
> >> to idle timeout calculations. This is how mpd does it, anyway.
> >
> > Yeah, same with ijppp, but it means you can only use those packages to
> > do a connection, ie its not very general...
> 
> I don't understand.  Which packets do you want to use?  If they're not
> destined for that interface, they shouldn't cause a dialup.
> 
> > (So its too bad if you have dialon demand ISDN :)
> 
> I don't understand this statement, either.

I think what he's saying is that it would be nice to divorce the
"dial-on-demand" functionality from the thing that is actually
doing the dialing... you could do this with divert sockets, but
it would require some sort of "api" to the PPP process not only
to tell it to connect (eg, send SIGUSR1), but also to be able to
determine whether the link is already up (to avoid a flood of
such signals).

-Archie

___________________________________________________________________________
Archie Cobbs   *   Whistle Communications, Inc.  *   http://www.whistle.com