From owner-freebsd-hackers  Sat Jan  4 15:05:06 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id PAA21631
          for hackers-outgoing; Sat, 4 Jan 1997 15:05:06 -0800 (PST)
Received: from cheops.anu.edu.au (avalon@cheops.anu.edu.au [150.203.76.24])
          by freefall.freebsd.org (8.8.4/8.8.4) with ESMTP id PAA21612;
          Sat, 4 Jan 1997 15:05:02 -0800 (PST)
Message-Id: <199701042305.PAA21612@freefall.freebsd.org>
Received: by cheops.anu.edu.au
	(1.37.109.16/16.2) id AA248799098; Sun, 5 Jan 1997 10:04:58 +1100
From: Darren Reed <avalon@coombs.anu.edu.au>
Subject: Re: New Networking framework for BSD
To: julian@freefall.freebsd.org (Julian Elischer)
Date: Sun, 5 Jan 1997 10:04:58 +1100 (EDT)
Cc: hackers@freefall.freebsd.org
In-Reply-To: <199701022236.OAA09044@freefall.freebsd.org> from "Julian Elischer" at Jan 2, 97 02:36:49 pm
X-Mailer: ELM [version 2.4 PL23]
Content-Type: text
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk


I have one _big_ concern with this:

Currently we have very well known and understood networking code inside the
kernel which has been around for years and has matured over that time to
become a very reliable backbone.  Rewriting this (or replacing it) with new
code will effectively put FreeBSD back many years in so far as being sure
that it has reliable networking.  Two excellent cases which show this are
Solaris 2 & Linux (both of which have their own TCP/IP stack as opposed to
BSD's) and amusingly the BSD socket interface is planned for Solaris 2.6
for software performance reasons :)

So, as long as we're prepared to have an "options TESTINET" (or similar)
for the kernel - which would be mutually exclusive with "options INET" -
for some years to come, it looks like an interesting idea worth pursuing.

Darren