Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!sundc!sun!decwrl!labrea!jade!ucbvax!ftp.UUCP!jbvb
From: jbvb@ftp.UUCP (James Van Bokkelen)
Newsgroups: comp.protocols.tcp-ip.ibmpc
Subject: Re: TCP/IP/UDP Driver Specification
Message-ID: <8710180251.AA10157@spdcc.COM>
Date: Sat, 17-Oct-87 22:25:43 EDT
Article-I.D.: spdcc.8710180251.AA10157
Posted: Sat Oct 17 22:25:43 1987
Date-Received: Sun, 18-Oct-87 23:01:54 EDT
References: <8710161618.aa20763@Louie.UDEL.EDU>
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The ARPA Internet
Lines: 13

The reason one can't use Netbios calls to access TCP/IP is that Netbios
has its own set of conventions about datagrams, sessions, and names, and
they are sufficiently limited in scope that you couldn't do much through
them.  A Netbios name looks nothing like either an IP address or a domain
name, a Netbios session has different shutdown characteristics than TCP, etc.

You can use TCP/IP as a transport layer for Netbios (see current Excelan or
U-B or the various RFC-1001 conforming products on the way).  You could hang
a lot of bags on the side of Int 5C so that you could effectively control
a TCP/IP through it.  However, my guess is that it would take enough new
hooks so that Netbios would be a pattern for the new hooks, and not much more.

jbvb