Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!cbatt!ucbvax!hplabs!hplabsc!taylor
From: taylor@hplabsc.UUCP
Newsgroups: comp.mail.headers,comp.sys.ibm.pc
Subject: Re: Fidonet <-> usenet
Message-ID: <1589@hplabsc.UUCP>
Date: Sun, 12-Apr-87 17:42:27 EST
Article-I.D.: hplabsc.1589
Posted: Sun Apr 12 17:42:27 1987
Date-Received: Mon, 13-Apr-87 06:18:50 EST
References: <149@4gl.UUCP> <108@hobbes.UUCP>
Sender: taylor@hplabsc.UUCP
Organization: Hewlett-Packard Labs, Palo Alto, CA
Lines: 58
Xref: utgpu comp.mail.headers:169 comp.sys.ibm.pc:2881

A lot of people seem to be confused about how to address different systems
that have different address notations (e.g. Fidonet).  I've seen suggestions
that range from having each part of the Fido address being a different 
``hostname'' (e.g. gateway!node!net!user), ones that suggest .FIDO should
be a top-level domain (joe@44.24.FIDO), that should be a second-level
domain under .ORG (joe@44.24.FIDO.ORG) and a second-level domain under
.UUCP (!) (as in joe@44,24.FIDO.UUCP).  

Well . . . . As Tim Pozar pointed out in his message on this subject,
there are a number of existing standards that need to be considered,
not only on OUR end (e.g. RFC-822, RFC-976, etc), but on the other
end too, on the FIDO systems.  As he pointed out, the Fido system is
most likely going to be changing to a point and zone address 
notation since TJ never anticipated it catching on this well and
crushing the original addressing scheme.

So what does this all mean?  It means that:

1. All fidonet gateways should speak the same address format for mail
   TO fidonet (and from Fidonet too, but that's a subject we haven't 
   talked about much yet in this group).  This address should allow for
   future expansion of the Fidonet system and be a legal set of domains
   in the UUCP/ARPA world.

2. Since the conceptual model of Fido addresses is a heirarchical domain
   system (e.g. local-user/local-host-number/host-network) (or something
   like that) it is needlessly confusing to change the notation to appear
   to be a series of hostnames (e.g. gateway!host-network!local-host-num!user).
   This is a *BAD* scheme...

3. The existing heirarchical system is designed to have a *very* small
   number of different top-level domains (including .COM, .EDU, .MIL, and
   .ORG).  One of the existing domains is ".NET" and is meant to denote
   a gateway to a different network.  For example, the CSNET gateway for
   the ARPA Internet is at "RELAY.CS.NET".   There is  *NO*  reason why
   we shouldn't fit in with this scheme (e.g. .FIDO.NET) and this not only
   means we're compatible, but that the routing software (e.g. smail) will
   know what to do with mail to Fidonet (it will, correctly, send it to the
   'cheapest route' Fido gateway on the system).

4. As far as the individual node addresses and all, I strongly agree with
   Tim that we need to identify WHICH PART of the address each part is.
   So this means that, for example, his address is:

	Tim_Pozar@NODE137.ZONE65.FIDO.NET

   and we let the local mail system route the mail appropriately.  (Cheap
   plug: the Elm system supports the "domains" database too, so you can 
   have a simple one-line entry to allow users to directly type in this
   sort of address and have it routed as needed!)

I think that the other existing Fido gateways are great, but that we really
DO need to at least keep a consistent addressing format.

						-- Dave Taylor

ps: I don't think we should have a "FIDO" host - we do *NOT* want to tie 
    the entire network into having a single gateway, do we??