Path: utzoo!mnetor!uunet!husc6!bloom-beacon!gatech!udel!rminnich
From: rminnich@udel.EDU (Ron Minnich)
Newsgroups: comp.sys.amiga
Subject: to IP or not to IP
Message-ID: <1739@louie.udel.EDU>
Date: 25 Mar 88 20:49:57 GMT
Reply-To: rminnich@udel.EDU (Ron Minnich)
Organization: University of Delaware
Lines: 25

OK, folks, here is the dilemma. TCP/IP, and as a consequence SLIP
are standard on pretty much everything. Even Terminal concentrator
boxes like CISCO support SLIP, at 38.4 Kbaud. There is going to 
be a SLIP standard soon. We have the availability of amigatcp, which 
supports SLIP.
   On the CON side, amigatcp has bugs. I just discovered that when a lot
of data pumps through the device, newtelnetp and amigatcp get stuck.
I don't know why and I can no longer put the time into it. ALso, amigatcp
is based on Karns' code which is at heart a single task pretending to
be a lot of tasks, because (of course) it started on a PC.
   ON the other hand, we have DNET, which works well, and fits into
the process/process model which we should be using. BUT, it talks
its own protocol, it can't get the efficiencies of SLIP, it 
can't get all the experience of the TCP/IP community; in other
words, a better implementation than amigatcp, but an outcast
protocl. What to do? I don't know. I use DNET for now, but am 
torn. Your comments would be appreciated.
   SHould we put a dnet support library into amigatcp ?> anybody
want to help fix amigatcp? I am out of time on this, i have to 
get my graduate work done, and dnet works now for me. I think
the problem with amigatcp is very simple, but i don't have time
to fix it. Any takers?

-- 
ron (rminnich@udel.edu)