True, but the question is also what steps need to be taken to insure
success.
I think you've already answered that. There is no means of ensuring
success, because the upconverter lacks the necessary information.
But the information he mentioned is all available. Determining used octet
values and line length is not rocket science.
*if* you know the nature of each content-type used. which, in general,
is not the case.
Except in the signed case, where you need to preserve details of the
original encoding, and *that* is broken.
true.
What the relevant standard *should* have done (of course it's a little
late now) was to sign the canonical form, and if they needed to sign the
headers, too, then describing a canonicalization method for those.
it's not too late to define another multipart that works properly.
Keith