From: Scott W. <swa...@my...> - 2002-02-18 19:29:15
|
> > everything looks ok to me. > > Can you send me your script? Or some more details? > > Just another remark : if the diff is little, all > is correct, so, back to my preceding exemple, 40.000 - 39.900 > gives the correct 100, but 40.000 - 0 gives -25536. > It seems you have an intermediate variable that can't > store more than a 16 bits integer... > > Fab. > Ah, ok. I think I found it then. I am only able to get strange behavior like this: Subtraction involving float and int, when the integer value requires more than 16 bits. Your guess is correct, during the floating point arithmetic operator, the integer was being passed to a function as a 16bit, instead of a 32 bit number. Marc, the change in double.c & double.h should correct this. -Scott |