RM = Raul Miller To word fiddle a bit:
RM> The bit pattern which we currently associate with RM> "negative zero" would remain as a potential bit pattern. This is an endorsement of my position. RM> However, if execution time is not an issue, I do not see why J should RM> have to treat this as a valid bit pattern in the context of floating RM> point numbers. Then you subscribe to Roger's definition of "have" so our discussion is moot. -Dan ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
