Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site cbosgd.UUCP
Path: utzoo!watmath!clyde!cbosgd!mark
From: mark@cbosgd.UUCP (Mark Horton)
Newsgroups: net.mail
Subject: Re: Address Questions
Message-ID: <526@cbosgd.UUCP>
Date: Thu, 29-Nov-84 17:15:50 EST
Article-I.D.: cbosgd.526
Posted: Thu Nov 29 17:15:50 1984
Date-Received: Fri, 30-Nov-84 07:32:14 EST
References: <14000003@smu.UUCP>
Organization: Bell Labs, Columbus
Lines: 64

In article <14000003@smu.UUCP> pedz@smu.UUCP writes:
>have three simple (and probably stupid) questions.  Suppose we
>setup a domain called "smu" and a host in that domain called
>"host".  Then my address would be pedz@host.smu.  Now suppose
>my domain is feed by uucp then I conclude that my full address
>would be pedz@host.smu.UUCP.

Note that smu probably would not qualify for a 2nd level domain
underneath UUCP, since 2nd level domains are intended for geographic
regions and huge organizations.  I suspect you would really fit in as
	host.smu.sw.uucp
where "sw" is a domain representing the southwestern USA.

> convex happens to feed us from the usenet and uucp mail net and of course
> uucp would choke if it was feed the above address.
Certainly some other hosts that aren't fixed will have trouble with it.

> Thus I suppose my address (which goes through convex) would
> be convex!smu!pedz@host.smu.UUCP.  Is this correct?
This is one way to do it, but not a very good way, because it's
ambiguous.  If convex suddenly converts from ! precedence to @
precedence, it will stop working.  What the UUCP project has settled
upon is this:
	convex!smu!host.smu.UUCP!pedz
which won't be stepped on by convex no matter what it's running.
Of course, this means that smu will have to recognize that
	host.smu.UUCP!pedz
has dots in the UUCP host name and should be flipped around the ! into
	pedz@host.smu.UUCP
This can be done with a 1 line addition to sendmail.cf.

> Assuming that it is at least close to correct (i.e. there
>is an @ and some .'s in the address, what is the probability that
>a user in a distant site would be able to get this address all the
>way through the net to me.  (Are some of the intermediate hosts
>going to choke on the address).
The chances are slim to none - if any of them run 4.2BSD they probably
use the 822 parsing (which gives @ priority.)  This is why we settled
on the domain!user syntax - we're confident it will get through unmunged.

>The third question is about !'s and @'s.  I have read that the
>!'s are processed before the @'s (at least I think I have).
Actually, it depends on your code.  The old UUCP code looks from
the left for a ! and ignores everything to the right.  This is not
documented anywhere but a lot of hosts do it.  Anything conforming to
RFC822 (including most 4.2BSD hosts, I believe) will parse it the other way.

>But what is the syntax to have the @'s processed before the !'s.
In general, there is no way to specify it.  You can't put parentheses
in because neither syntax understands them, and both use them for
something else.  (822 uses ()'s around comments, and the UNIX shell
will choke on them too.)

>For example, smu is also on csnet, and there is another host called
>smuvax which is linked by uucp.  What is the address for smuvax via csnet?
I suspect the preferred syntax would eventually be
	user@smuvax.smu.sw.uucp
In the meantime, within CSNET you'd probably use
	smuvax!user@smu.csnet
which has to be uglified into
	smuvax!user%smu.csnet@csnet-relay.ARPA
on the current ARPANET until CSNET becomes an official top level domain.

	Mark Horton