From: John Stanley (stanley@peak.org)
Date: Mon Jul 08 2002 - 09:40:49 CDT
Stop mailing me copies of this stuff. How many times do I need to say
that?
On Mon, 8 Jul 2002, Clive D.W. Feather wrote:
> John Stanley said:
> > There is absolutely no reason for encoding a message id.
>
> If I am correctly reading the syntax, a message id is limited to US-ASCII
> characters. Therefore encoding is never necessary.
Like I said, there is absolutely no reason for encoding a message id, and
that is not just because a message id may be limited to US-ASCII. This
draft should not say that it may be encoded one way in one place and a
different way in another place.
> > In addition, to say that they MUST be identical except they may be
> > different is just loony. It's as if one is trying to redefine the
> > RFC-standardized meaning of MUST.
>
> No, "MUST be X, except case Y MAY be Z"
That is a contradiction. If mythical case Y means they don't have to be
the same, then the MUST is not a MUST.
Moot point, there is still NO REASON to encode a message id, nor is there
a reason for it to contain UTF-8 or anything except simple ASCII. The goal
seems to be to make the definitions of what a header can contain as
complex as possible, and it has been successful. Can a msg-id contain
UTF-8? Well, this change to the draft clearly includes Message-ID headers
as one of the headers that MAY be encoded when it does contain it, so any
claim that it cannot is just one interpretation. It was included
"specifically", so it wasn't just an accident that it was included.