Getting RFC2047 encoding "right"

[Keith Moore]

I've been talking about doing the "right" thing wrt RFC 2047. It would
have been more accurate for me to say "this is how 2047 was intended to
be used" or even "this is what I had in mind".

I think I can speak with reasonable authority on how 2047 was intended
to be used. Whether or not those intentions were naive or shortsighted
or less scalable than we'd like - those are of course separate
questions. Still, when a protocol is used in a way that is different
than was intended it does seem likely that some problems will result.
One thing that seems obvious in hindsight (and with ~12 years more
experience) is that protocols are quite frequently used in ways that
are different than their designers intended.