Jim Fulton wrote:
- for addition and subtraction of two "exact" (fixed-decimal) numbers, the result has the maximum of the scales.
One could argue that this is incorrect: if "3.1" means that I know the value to one decimal of precision, and "2.01" means that I know that value to two decimals of precision, stating the result of their sum as "5.11" suggests that I know the result to two decimals of precision, which is of course false: because I only knew one decimal of precision for one of the operands, I only know (at most!) one decimal of precision for the result.
Not arguing for this interpretation, just indicating that doing fixed precision arithmetic right is hard. I'm waiting for Tim Peters' contribution, but he's on vacation so it may be a while.
--Guido van Rossum (home page: http://www.python.org/%7Eguido/)