I am not terribly concerned about lognumer in particular, though it seems
particularly poorly thought out. Worse are the options that might ought
[sic] to be useful, but aren't really. No examples off the top of my head.
Also, it would be nice to have a clearer, cleaner general design philosophy
against which to evaluate things like this. For example, does a float
actually imply inexact results? (I think yes.) Bt does it imply *complex*
inexactness? For example, is log(-1.0) =3D 3.14*%i (as if it were -1.0+0.0*%i)
or is it %pi*%i? Is 0.0 the same thing as 0.0+%i*0.0?
-s