BTW, has there been considerations to use a "Content-Encoding" header
additional to the CTE?
In theory this is definitely the way to go. The problem is it doesn't
work in practice.
Current MIME agents look at content-type and content-transfer-encoding
and, if they understand what they see there, think they can remove
the transfer encoding and do something useful with the result. If you
added content-encoding, however, existing agents won't know to check it,
and will end up delivering piles of compressed binary trash to their
viewers. This is unacceptable.
This is why we defined MIME-Version -- we expected to be able to bump it when
introducing an incompatible change like this. But MIME-Version ended up botched
for political reasons, so it is no longer an option. (It turns out also to be
at the wrong level -- a technical botch, but not an insuperable one.)
So basically we're stuck with extending existing fields to express concepts
which I agree are largely (although not completely) orthogonal. Fortunately I
don't think this will prove to be much of a problem in practice, as long
as we can agree on what to do.
Ned