When I wrote the orthopoly numerical stuff, I briefly thought about using
CLOS to do double-float, bigfloat, and complex floating point math.
If somebody got something like this working, I'd use it in orthopoly.
Without a unified scheme for floating point math, you either need
multiple versions of the numerical code, or you are stuck with
molasses---slow code (frequent calls to rectform on complex numbers).
Barton