Tim Peters wrote: > from trig functions doing argument reduction as if pi were represented > with infinite precision, That sounds like an interesting trick! Can you provide pointers to any literature describing how it's done? Not doubting it's possible, just curious. -- Greg