IMHO any code that tries to perform floating-point arithmetic on integer values and then produce exact output should be considered suspect in and of itself...to

IMHO any code that tries to perform floating-point arithmetic on integer values and then produce exact output should be considered suspect in and of itself...too many edge cases.