Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!cbatt!ucbvax!ucbarpa.Berkeley.EDU!jordan
From: jordan@ucbarpa.Berkeley.EDU.UUCP
Newsgroups: comp.mail.uucp
Subject: Re: Use of domains by uucp sites
Message-ID: <17143@ucbvax.BERKELEY.EDU>
Date: Sat, 31-Jan-87 13:03:54 EST
Article-I.D.: ucbvax.17143
Posted: Sat Jan 31 13:03:54 1987
Date-Received: Sat, 31-Jan-87 20:29:28 EST
References: <3063@elroy.UUCP>
Sender: usenet@ucbvax.BERKELEY.EDU
Reply-To: jordan@ucbarpa.Berkeley.EDU (Jordan Hayes)
Organization: University of California, Berkeley
Lines: 26

David Robinson <david@elroy.UUCP> writes:

	Shouldn't there be new top level domains that dont confict with
	the DDN top level domains? With the old ".uucp" I would know to
	use pathalias to find the machine, now I have to hunt around
	between the DDN nameservers and pathalias to find out how to
	get somewhere.

Ay, thars the rub.

Cardinal rule of domains: names don't imply routes (let alone transport).

So, the problem becomes one where the main paradigm of mailer
resolution is now obsolete.  All previous attempts at having MTAs
provide a "switch" for multiple transport systems relied on the format
of the address.  This is no longer valid.  One good thing about it is
that it will encourage the design of better gateways.  Except this has
been known for 3 years and we all (myself included) have ignored the
writing on the wall (too harsh -- we all have been too busy for it, I
know).

Given, dynamic routing in a slow network given multi-homed nodes is
a really difficult problem ... any of you hot-shot professors out there
want to sponsor me for a PhD. project? :-) :-) :-) ...

/jordan