The problem with BINARYMIME is that unless everyone supports it, someone
may have to transform the message... and that will break DKIM signatures
and make life annoying. Hence my colleagues question about a mime aware
dkim canonicalization format as a pre-req for binarymime.
We considered it when we were doing the strict/loose c14n but decided
against it as a swamp. If a relay is going to change MIME structures,
it can recode parts and reorder them while keeping the "same" contents
which will make it quite a challenge to describe what one has to do to
put them into canonical form.
There is a c14n registry to which new values can be added, so if you
want to propose a MIME c14n, the first step is to write it up in a draft.
ietf-smtp mailing list