Date: Sat, 30 Jan 1999 15:39:42 -0500 (Eastern Standard Time)
From: John C Klensin <klensin@mci.net>
Subject: Re(2): stripped-down client configuration (Address, MX RRs, A RR-only?)
In-reply-to: <19990130201949.014005@relay.skynet.be>
To: Brad Knowles <blk@skynet.be>
Cc: drums@cs.utk.edu
Message-id: <SIMEON.9901301542.C@p6.mci.net>
MIME-version: 1.0
Content-type: TEXT/PLAIN; CHARSET=US-ASCII
Priority: NORMAL

Speaking as a participant, not as editor...

On Sat, 30 Jan 1999 20:19:49 +0100 Brad Knowles <blk@skynet.be> 
wrote:

>     As I said, I think this gets down to initial message injection versus
> transport of a previously injected message.  I feel that name->mx->a is
> the only correct method in the latter case, but I believe you're asking
> too much of (stupid) MUA authors to expect them to handle it in the
> former case.
>...
>     Any standard that makes "requirements" that are blatantly ignored for
> something like this risks simply alienating the programmers, system
> administrators, users, etc... that it is supposed to help serve.
>...

You are getting a little hyperbolic in applying this argument to
this situation.  Many MUA authors do use name->mx->a, some
don't.  I suggest that it is probably useful for us to provide
some guidance as to what is expected here, and maybe the reasons
why.

>     I think we need to make a clean and clear break between initial
> message injection and the transport of previously injected messages.  We
> can do this with separate sections, we can do this with specific language
> that calls out what is appropriate where, or we can do this with totally
> separate standards documents.  Frankly, I don't care which, but I think
> the time has come where we need to make that break.
>...

We can do this in the SUBMIT document?

    john