newbie willing to contribute...

[François-Xavier Coudert]

> A library part?  Can't you map complex(x,y) to
> the appropriate cmplx(X,Y,some_kind) where X and
> Y are properly converted to compatible types?

COMPLEX doesn't need a library part, but I sent the diff of my whole
current tree on my laptop since I'm a bit out of time (I'm supposed to
be working :)  It includes the other intrinsics, which have library
parts.

FX