Date: Thu, 25 Apr 1991 10:35:25 -0400 (EDT)
From: Nathaniel Borenstein <nsb(_at_)thumper(_dot_)bellcore(_dot_)com>
Subject: Re: Comments on Draft RFC
... stuff deleted...
Ah, but in a multipart message, each "body part" is an encapsulated
message, and can therefore have its own message-id header.
I've been having trouble agreeing to some of the proposals being
floated with respect to just encapsulating everything recursively,
and I think that this sentence from Nathaniel captures where I
part ways with his conceptualization.
I think of body parts in the X.400 sense -- a body part is *not*
just a message, it is an entirely different data-type. The set
of headers that are "legal" in a message is much broader than
in a body part. In fact, "body part headers" is just a typographical
convenience to encode information; they really come from a different
name space and are interpreted on their own.
I will freely admit that, if we were starting with a clean sheet of
paper we could define body parts in either way, but since X.400
interoperability is (should be?) a major goal, it makes sense to
define the architecture of the message structure to be compatible
with X.400 body parts.
Does anyone actively want to do something that is not compatible with
X.400 in this area? Are the gains worth it?
Neil