Path: utzoo!attcan!uunet!ncc!augate!lyndon
From: lyndon@auvax.uucp (Lyndon Nerenberg)
Newsgroups: comp.mail.uucp
Subject: Re: Another example why not to re-route
Summary: .uucp is not a domain ...
Message-ID: <419@aurora.auvax.uucp>
Date: 12 Dec 88 17:29:08 GMT
References: <14430@comp.vuw.ac.nz> <831@mailrus.cc.umich.edu> <38@microsoft.UUCP>
Reply-To: lyndon@auvax.UUCP (Lyndon Nerenberg)
Organization: Athabasca University
Lines: 30

In article <38@microsoft.UUCP> w-colinp@microsoft.UUCP (Colin Plumb) writes:
>How does this re-routing rule sound:
>
>If you see a domain-style name (even .uucp) in the bang path, it's okay
>to route to that site.  Sites without domains are not necessarily distinct,
>and should not be routed to, even if they appear before the domain-style name.

The purpose of a domain name is to *guarantee* a globally unique name for
some entity. This is done through enforced coordination of the domain
namespace. The .uucp "domain" has no central organization, therefore
you cannot guarantee the existence of one, and ONLY one, foo.uucp.

Preaching aside :-), there is a practical reason for this as well. Various
pieces of usenet software (Bnews, rn, smail 2.X come to mind) provide a
default "domain" of .uucp. Odds are quite good that when a new site comes
on the net, the system administrator will not understand the implications
of his site name duplicating that of another site. Until he finds out,
your scheme guarantees that one of the two sites won't be getting its
mail. You CANNOT reliably force a route to anything.uucp inside of a
bang path.

[ I speak from experience. We have a site here named "aurora" that
  duplicates a registered site at NASA. Until AU gets a registered
  domain, I have to carve up the mailers and news software to lie
  about traffic originating from *our* aurora. ]


-- 
Lyndon Nerenberg  --  Computing Services  --  Athabasca University
{alberta,attvcr,ncc}!auvax!lyndon