 |
 |
>Well, in an ideal world, optimization wouldn't have semantics. It should Perhaps in an ideal world w/o floating point arithmetic as we know it. This "problem" is made worse if the consumer expects IEEE exceptions to occur "where written" and occur as many times "as written". As you noted in another part of the thread, interval arithmetic highlighted some of these issues. >the integrity of parenthesis). This is because I really would like to >move toward that ideal world where optimization was entirely a >pragmatic issue. I would be interested in seeing such an environment. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%