I hate to say this, but X.400 has the right solution! The original content
types are specified, and any performed conversions are indicated in the
Isn't this the equivalent of the much-attacked notion of encapsulating
the entire message that was floated (and roundly attacked as
UA-unfriendly) on this list a few weeks ago?
Unless I'm missing something, the expected action of a receiver would
be to serially un-do the conversions so as to be able to present the
message in the form of the original types, more or less. That is, of
course, in the case that goes one step beyond Erik's, which is an
ANSI/ISO charset machine sitting on the far side of the EBCDIC
environment and requiring a second conversion.
It seems to me that there are two philosophically quite different
approaches here, one of which is to assume careful and detail-level
conversions to the local environment (which is similar to the situation
with, and arguments for, transport encoding at the minimum-body-part
level only) and the other is to assume a something-centric
(ASCII-centric, 8859-1-centric, IA5-centric, SMTP-centric,
X.400-centric,...) world and try to make everyone else adapt to it. I
think there are, of course, arguments for both.
Or am I missing something?