On 11/26/13 3:22 AM, Joseph Rushton Wakeling wrote: > So, as other people have suggested, really the only thing we can > reasonably do is to define a separate Imaginary type I agree. Let's move forward with this. Thanks, Andrei