> extended = no longer a keyword
> real = 80 bit floating point
> cfloat =  complex float
> cdouble = complex double
> creal = complex real
> ifloat = imaginary float
> idouble = imaginary double
> ireal = imaginary real

The "real" for extended sounds a bit strange. I would prefer something like
real32
real64
realMAX
complex32
complex64
complexMAX
imaginary32
imaginary64
imaginaryMAX

This will be consistent IMO. Anyway, as long as the keywords are there, the names doesnt really matter. With the array handling, attributes and these keywords along with the nice grammar in general, it looks like D could be THE choice for numerical programming (if the speed is up that is).