Sun Jan 20 11:18:58 CET 2013


Should phasors be implemented as unsigned integers, followed by
int->float conversion and a scaling operation, possibly some
bit-twiddling, or should we just stick to floats?

Another problem is delay lines: here there might need to be extra
precision over the 24 bit in a float.

Trouble with ints is that they make the code typed.  A single type is
the same as no type..