The missing characters could also be coded in the private use zones
of 10646. 10646 already have the mechanisms, why not use it?
If we use a 10646 private zone for the unconvertible characters, what
happens when the next version of 10646 comes out and it contains the
previously unconvertible characters?
This is similar to having two versions of ISO 646 (or UNICODE or
ISO 6937 or X0208). A mechanism you could use is to register and
number each version (like the ECMA registry) and then specify
which version is used in the RFC. When a new revision of our
omnipotent character set is released (they found life on Pluto),
a new RFC is released (some years after...) with the new specification
and a way to retain interoperability with the old specs.