correct • elegant • free

△ Miscellany △

◅ Einstein



In article <>,
John Critchley <> wrote:
>>[ is ] obsolete, and should be removed.
>Really?  Since when?

Roughly, since UKnet ceased being the only service provider in the

In case you're not aware, UKnet (as they were then called) used to
publish two sets of UUCP maps for the UK.  The ones in comp.mail.maps
(u.gbr.[0-9]) showed that uknet had a direct link to every UK site,
and were intended for use by anyone outside the UK.  The ones posted
here (uk.gbr.[0-9]) showed real connectivity, and were for use by
anyone running pathalias within the UK.

This was a reasonably sensible tactic.  Updates to comp.mail.maps
could take weeks to get through, but by claiming a direct link to
everyone, UKnet controlled routing for all mail coming into the UK
(it all came through uknet anyway).  Once you'd initially got into
comp.mail.maps, everything worked smoothly, even if your connections
to other UK sites changed.

It all started crumbling when PIPEX appeared, followed by other
service providers, with their own international connectivity.
Obviously, if you are a customer of PIPEX, your mail has no business
sitting on EUnet relay hosts, or using EUnet's international

We tried to solve this by inventing u.gbr.100, where UUCP map
entries for customers of other service providers could live, but
it never really worked very well.  At the same time, EUnet were
pursuing a policy of really connecting directly to all their
customers, so the differences between u.gbr.[0-9] and uk.gbr.[0-9]
were slowly being eroded anyway.

I see that u.gbr.100 is still maintained by EUnet and posted to
comp.mail.maps.  However, EUnet haven't posted any maps to uk.mail.maps
for months and months.  I repeat: uk.mail.maps is obsolete, and
should be removed.

>>If you like that sort of thing, comp.mail.maps still has lots of
>I hope that this means that the UUCP mapping project is still healthy.

No, I don't think so.  (Apart from any other problems, every article
in our comp.mail.maps spool has an invalid `Expires:' header!)

Last time I looked in comp.mail.uucp (a long time ago) there were
endless complaints about the time it was taking for update to the
maps to happen.  You're the first person I've encountered who's
shown any positive interest in it for a couple of years.

>After all the .UUCP pseudo domain does give an alternative way of getting
>yourself a mail domain, without the risk of being black-balled or paying

Which a large and increasing number of mail users will be unable
to reach.  Until recently, we spent a large amount of (computer)
time every day running pathalias and munging the results into
something useable by PP, so we could support the .uucp pseudo-domain.
We did an audit on the use of pseudo domains, and over a period of
several weeks we saw not a single .uucp address.  This is despite
the fact that many PIPEX customers use our "" sendmail
config file which punts
.uucp and .bitnet to us.  We no longer run pathalias; if we ever see
another .uucp address (and it's not one that belongs to one of our
customers) it will be punted to

In reality, there is only one source of mail routing information
that 99% of the planet pays any attention to any more, and that's
the DNS.  Sorry, but if you are at all serious about email in 1996,
you need a domain name.

Please note that I am only talking about the UUCP mapping project,
and not UUCP as a protocol: the two are really completely orthogonal.
If you want to use UUCP to your service provider, please do so (I'm
slightly surprised that more people don't); just make sure that
they give you a domain name, and your UUCP name is kept private
between you and them.

Tim Goodwin   | "MS-DOS is dwindling, only to be replaced by something
Cambridge, UK | even more bizarre and incomprehensible." -- Steve Summit

Original headers:

From: (Tim Goodwin)
Subject: Re: News
Date: 4 Mar 1996 15:59:06 GMT
Organization: Unipalm PIPEX
Message-ID: <4hf40a$>
References: <>
  <4gkass$> <>

△ Miscellany △

◅ Einstein