RE: Latest DRAFT of SMTP bindings

["Moberg, Dale" <Dale_Moberg@stercomm.com>]

(apologies to dick and chris for the duplicate)


> The discussion concluded that the MSH need NOT concern itself with
> this since it is supposed to be transport neutral. The transport
> (e.g. the MTA) would deal with this as needed by encoding the WHOLE
> message package.


I don't think this will work very well in general. Assembling
multiparts, for example, is actually a user-agent and not
a MTA task. The User-agent puts the MIME headers into the 
message, for example, including the ones on CTEs. If the MTA
were to apply encodings, it would break multipart/signed
objects, for example, where the signature is over the headers
and the body part of the first bodypart in the multipart/signed.

Unless you want to preclude using PGP or SMIME, I would
very strongly urge that the CTEs be regarded as applied
prior to transport. Transport-neutrality among MIME-aware
transports is at most structural equivalence over transports,
and not bit-level equivalence; stronger transport neutrality
is not available at the bits-on-the-wire level.

Dale Moberg