Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site allegra.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!jdd
From: jdd@allegra.UUCP (John DeTreville)
Newsgroups: net.micro.pc,net.lan
Subject: Re: NCC
Message-ID: <2671@allegra.UUCP>
Date: Sun, 5-Aug-84 17:21:23 EDT
Article-I.D.: allegra.2671
Posted: Sun Aug  5 17:21:23 1984
Date-Received: Mon, 6-Aug-84 01:14:08 EDT
References: <419@sri-arpa.UUCP> <184@cbosgd.UUCP>, <3750@brl-tgr.ARPA>
Organization: AT&T Bell Laboratories, Murray Hill
Lines: 26


    From: ron@brl-tgr.ARPA (Ron Natalie )
    Newsgroups: net.micro.pc,net.lan
    Subject: Re: NCC
    Date: Thu, 2-Aug-84 11:34:22 EDT

    If 3Com was smart, they'd give up on UNET!  It's among the worst attempts
    at TCP I've seen.  Great if you want to talk to other UNET sites, not so
    hot at talking to real internetters.

The principal problem with UNET is that it was one of the earlier available
implementations of TCP/IP.  Such systems are typically debugged by seeing
whether they can talk with other systems (if they can't, it may be their
fault, or the other systems' fault, or both).  When UNET first came out,
there weren't an awful lot of other TCP/IP implementations around to test it
against.  As other implementations have come out, the process of testing
them has uncovered various bugs in UNET, which do get fixed.

The release of new implementations has also uncovered some unfortunate
overflexibilities in the TCP/IP specification, which allows for two
implementations, each of which conforms with the standard, to be unable to
communicate.  These problems are solved by arriving at a (weighted) consensus,
and once again UNET has to be changed.

Cheers,
John DeTreville
Bell Labs, Murray Hill