I have yet to see a justication for signed zero in the real domain which did not rely on arguments based on the complex domain.
Also, I think therre should be a more sensible way of approaching this for complex domains. For example, polar complex numbers are perfectly capable of representing "signed zero". However, these would not deal with the earlier example where the sign of the complex element of the value _1 matters for the result. In other words, that problem demands two vectors -- a vector from the origin to the point in question, and a second vector to orient a direction from that point. I wonder if the earlier Borda example might be more comfortably expressed using quaternions (which are capable of representing a sequence of rotations)? -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
