> If you apply the rules for arithmetic and comparison operations over mixed numbers, explained elsewhere in the manual, you have the semantics of mixed-number loops. What exactly do you think is missing?
My impression was that the original poster was unhappy about the "var > limit" condition, which, when limit is the max positive integer (or a larger floating point value), can never be satisfied. This is not obvious and arguably unexpected by programmers. Personally, I agree that having to consider that an intermediate value can overflow and derail the computation requires a discipline that would be difficult to instil upon myself or others.