NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/comp/mail/uucp/2469

Internet Message Format | 1992-12-30 | 3.7 KB

Xref: sparky comp.mail.uucp:2469 comp.mail.misc:4150
Path: sparky!uunet!psgrain!hippo!ee.und.ac.za!tplinfm
From: [e-mail] (Alan P Barrett)
Newsgroups: comp.mail.uucp,comp.mail.misc
Subject: Re: Mixed format addresses
Date: 30 Dec 1992 19:49:40 +0200
Organization: Dept. Elec. Eng., Univ. Natal, Durban, S. Africa
Lines: 64
Message-ID: <[e-mail]>
References: <[e-mail]> <[e-mail]>
NNTP-Posting-Host: daisy.ee.und.ac.za

In article <[e-mail]>,
[e-mail] (Willard Dawson) writes:
> [e-mail] (Alan P Barrett) writes:
> > An Internet host cannot *know* that .UUCP in invalid unless it either
> >     queries the DNS and gets back an error indication
> > or  has special-case information somehow installed in its configuration.
>
> Except that .UUCP is not a valid top-level domain (is it?).  I don't have
> all the RFC's, so I don't know from personal knowledge, but I've been
> told that.

I am going to try repeating myself using different words, because your
comment indicates that I did not make myself sufficiently clear.

.UUCP is not a valid top-level domain.  You know that and I know that.
But an Internet host cannot know that unless either
(a) it learns that .UUCP addresses are invalid by asking the DNS,
or
(b) it has some information stored somewhere outside the DNS, telling
    it that .UUCP addresses should not be looked up in the DNS.

Option (b) has several further possible refinements:  (i) The special
information may tell the host to reject all .UUCP mail, or (ii) to punt
all .UUCP mail to a smart host, or (iii) to look up the .UUCP addresses
in one or more places outside the DNS and take decisions based on the
results of the lookups.

I get the impression that you think that there is some other method --
let's call it (c) -- distinct from (a) and (b), but I don't know what
it might be.

> > A good argument can be made for avoiding special-case information, and
> > Internet hosts that believe such an argument have every right to query
> > the DNS for *everything*, including .UUCP sites.
>
> If .UUCP is not a part of the Internet, and if DNS only applies to intra-
> domain routing, what sense does it make to query DNS for site.UUCP?

The DNS is the Internet's official name registry.  That being the case,
I think that it is perfectly acceptable for the administrator of an
Internet host to decide that mail to destinations not registered in the
DNS will be rejected.  Such an administrator may elect to use method (a)
[ask the DNS] or method (b)(i) [special knowledge that .UUCP mail must be
rejected].  Many administrators elect to use method (a), and I consider
this to quite reasonable, although it's not what I would do myself.
If option (a) is chosen then the DNS will be the only method available
to the host for checking whether a .UUCP address is valid.  Asking the
DNS makes sense if that is the only method available.

I am not saying that a host SHOULD use method (a) [ask the DNS about
.UUCP sites], I am merely arguing against what I perceive to be your
position that method (b) [special knowledge outside the DNS] or method
(c) [if such a method exists] are the only acceptable methods.

I happen to be one of the folk who believes that storing special
knowledge about .UUCP outside the DNS is not such a bad thing, and I
use such special information -- methods (b)(ii) or (b)(iii) -- on all
the hosts under my control.  I figure that, if the .UUCP domain is ever
officially added to the DNS, I will easily be able to remove the special
information, and switch to method (a).

--apb
Alan Barrett, Dept. of Electronic Eng., Univ. of Natal, Durban, South Africa
RFC822: [e-mail]                    Bang: m2xenix!undeed!barrett