Date: 20 Mar 1996 19:37:00 +0200
From: kai@khms.westfalen.de (Kai Henningsen)
To: drums@cs.utk.edu
Message-ID: <65Fq4YhjcsB@khms.westfalen.de>
In-Reply-To: <19960319234726.26756.qmail@koobera.math.uic.edu>
Subject: Re: allow <ext>?
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Organization: Organisation? Me?! Are you kidding?
Comment: Unsolicited commercial mail will incur an US$100 handling fee per received mail.

djb@koobera.math.uic.edu (D. J. Bernstein)  wrote on 19.03.96 in <19960319234726.26756.qmail@koobera.math.uic.edu>:

> > If an 822 address doesn't map to the target domain's local address
> > space then what DOES it map to?
>
> As I mentioned, the semantics should be user-defined but subject to
> preemptive standardization by the IETF.
>
> We already have one standard example: <>. The semantics of <> aren't
> defined by any particular domain.

Note that using <> is allowed *only* as a return addresses, and is defined  
as "undeliverable address".

Any address that doesn't include a domain cannot, obviously, be used as a  
destination address in the Internet. This means that *any* such <ext>  
address is undeliverable and is, thus, equivalent to <>; no new syntax is  
needed for this.

On the other hand, it _might_ be useful to allow some sort of not-naming-a- 
domain for useage between an UA and an MTA. (I think we want to eventually  
say something about UAs delivering mails to MTAs, don't we?)

An UA does not necessarily have a valid idea of the local domain name. So  
it might be useful for the UA to omit it when delivering mail, and let the  
MTA (which has to know it) fill it in. Using <username> is established  
practice for that; we might want to find a way to allow it syntactically.  
Then again, we might want to make it clear that this is forbidden once it  
has passed the first MTA.

And we might explicitely allow something like that for communication from  
the MTA to the UA, if we do consider that useful.

In any case, this should definitely be restricted to the MTA<->UA  
communication; it should never appear in the MTA<->MTA part of the world.  
This has to be made clear.

> In my experience, most of the hosts failing to support ``postmaster''
> are PCs owned by people who simply refuse to take on the responsibility
> imposed by RFC 822 even though they want to receive mail. I take it
> there have been similar problems with X.400 hosts.

It's easy to solve this, of course; the software I'm currently using does  
(IMHO) The Right Thing(tm) for single user machines. It's a DOS based  
UUCP+MTA+UA+News combined system, and it simply considers _every_ local  
part valid (no message will ever bounce) and delivers those to separate  
mail folders.

It supports "Postmaster" without even knowing about it, trivially.

> I agree that it's a shame. Quoting the qmail docs: ``Postmaster. You're
> not an Internet citizen if this address doesn't work.''

Yes. And it's really so easy to do!

MfG Kai