Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!cmcl2!arizona!rogerh
From: rogerh@arizona.edu (Roger Hayes)
Newsgroups: comp.sys.amiga
Subject: Re: tcp/ip for amiga
Message-ID: <2009@megaron.arizona.edu>
Date: Tue, 15-Sep-87 21:22:40 EDT
Article-I.D.: megaron.2009
Posted: Tue Sep 15 21:22:40 1987
Date-Received: Fri, 18-Sep-87 05:41:28 EDT
References: <503@louie.udel.EDU> <1929@umd5.umd.edu>
Reply-To: rogerh@arizona.edu (Roger Hayes)
Organization: University of Arizona, Tucson
Lines: 30

I have been thinking about this approach, too.  It seems useful to
embed TCP (and UDP?) in the filesystem.  For example,
	type <tcp:10.0.0.51/13
would get the time (tcp port 13, the daytime protocol) from sri-nic
(internet address 10.0.0.51).

I agree that it might not be possible to model all of TCP's semantics
this way.  Note that BSD sockets provide an incomplete but still useful
model of TCP; they don't handle urgent data very well.  They also
support the read & write calls, once they're connected; so someone
thought that it would be useful to be able to treat a connection like a
file.

It would be real slick to be able to use names instead of IP addresses:
	type <tcp:barrel.arizona.edu/79
	rogerh
to do a 'finger' on me.

Then there's a NFS device: perhaps mount each remote machine?
So 
	copy c:foobar to cholla:/usr/rogerh/amiga/archive
copies a file to a server's disk.  If this works, some of us
might be faced with a decision: hard disk or network interface?
What model does Ameristar present to the user?

		Roger Hayes
		rogerh@arizona.edu

I wish I had time to work on all this; maybe next year, after
dissertation.