Erik writes:
 
> I have not been following those discussions closely of late, but MIME used
> to be concerned only with the body of mail messages, and I tacitly assumed
> that were were talking about e-mail as a whole, not just the bodies, in
> fact, I do not consider the body part specifications in MIME to be very
> significant, but I also know I'm in the minority here.
 
I had never actually considered anything _except_ the body-part. It never
dawned on me that someone might want to express headers in SGML (why would
someone want to do that? unless building a textbase of email messages, which
is not at all what I was thinking of).
 
> However, the mail headers can relatively easily be expressed in SGML, and a
> DTD for that should be possible to design without much effort.  The biggest
> problem is the ability to add headers that are not specified in RFC 822 or
> updates, so that there are headers for which the semantics is known, and
> which should have their own element types to capture their structure, and
> headers for which no meaning is defined, and whose headers names needs to
> be preserved.  This distinction tends to complicate the picture somewhat.
 
Yes, this would certainly be possible. Looking a bit further into the future,
it might be more productive to tackle X.400 structures, which are better-
defined, and probably the way things will ultimately go. I know it doesn't
solve the problem of what to do with RFC822 headers, but if those are going
to die, do we need to sped a lot of effort on them? Unless, as said, one is
constructing a corpus of email...
 
> For the body part stuff, I'm sure a DTD can be written up for MIME, but I'm
> not so certain about the other stateful stuff that MIME talks about in
> those body parts.  Character sets is one of them.
 
My understanding (already shewn to be faulty :-) was that MIME and HTML held
a number of things in common. The spur for development was the perceived
need to be able to represent purely visual attributes (italics, bold and
the like) in email, and to add a hypertext dimension, enabling some form
of live x-reference between messages, or between a message and a URL elsewhere.
But I really haven't gotten into it: I downloaded a copy of metaMail because
MIME is turned ON in my version of elm, and I wanted to see what it did.
It failed to compile tho, so I left it.
 
> This is not a very good solution.  It is better to allow any DTD as long as
> it comes with a link process definition (LINK) that uses presentation
> attributes defined for the MIME MUA.  Then people can use whatever DTD they
> want, and still get useful things out of it.
 
I'd like to see this perform at an acceptable speed if it has to read
and act on a DTD in real time :-)
 
///Peter