Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sun-barr!olivea!mintaka!yale!think.com!zaphod.mps.ohio-state.edu!usc!elroy.jpl.nasa.gov!ncar!gatech!bloom-beacon!eru!hagbard!sunic!mcsun!ukc!mucs!cns!umaida!jf
From: jf@ap.co.umist.ac.uk (John Forrest)
Newsgroups: comp.mail.sendmail
Subject: When to reverse domain addresses?
Message-ID: <1990Nov16.155757@ap.co.umist.ac.uk>
Date: 16 Nov 90 15:57:57 GMT
Sender: usenet@cns.umist.ac.uk (News System)
Reply-To: jf@ap.co.umist.ac.uk (John Forrest)
Organization: UMIST Computation, Manchester, UK.
Lines: 44

Neil Rickert has very kindly done some mods to the IDA cf file for me. The aim of
this was to be able to conditionally reverse incomming mail addresses if required
[remember, traditionally in the UK we do this backwards]. As an example, our
relay might send us local names looking like "jf@uk.ac.umist.co.ap" which we
reverse to give "jf@ap.co.umist.ac.uk". I suggested two specifications to him
(depending on the required functionality locally).

	Spec 1 - Cope with incomming mail with reversed names.

	It after looking up the name in the various forms, the last
	element of the address is not a "top domain" (in CT?), and the
	first element of the address (after @) is, reverse and try
	again.

	Spec 2 - Allow local users to type mail reversed addresses.

	If after looking up the name in the various forms, the last
	element is not a "top domain", reverse and try again.

He has implemented Spec 1, explaining that Spec 2 requires non-trivial
modifications. We have so far not used the mods in anger, but they have worked in
testing (both in -bt mode, and sample outgoing mail). However, we have suddenly
realised that the function described is flawed, for the reasons several people
here are finding out. Basically, consider:

	uk.ac.ucl.cs

If we had this as an incomming mail, we would interpret this as being somewhere
in Czechoslovakia, not as Computer Science at University College, London. Note we
don't have direct access to the Internet, and our local name server contains (or
will soon contain):

*.cs.	MX 0 ......

so we just assume anything in "cs." is to be relayed. Has anyone got a better
algorithm that the above - really to detect whether the name is valid. The only
things I can think of are:

	1) Hardwire all the problem cases in one of the databases.
	2) Let's give up sending reversed addresses around the UK. (please?)

John Forrest
Dept of Computation
UMIST.