eh? The results are predictable and intentional (on a platform with support for negative zeroes), and they exist to support small values. | [reply] |
When I said, results can be unpredictable...that meant that source code on machine X may not generate the same results as on machine Y. Machine X will probably do what Machine X does repeatably!
I think the whole idea of this thread was that when you are dealing with either very big or very small numbers, there are a lot of ways to "get into trouble", meaning that numeric results for the same source just may not be equal.
| [reply] |
that meant that source code on machine X may not generate the same results as on machine Y. Machine X will probably do what Machine X does repeatably!
"platform-specific" or "not portable" are common terms to describe this.
I think the whole idea of this thread was that when you are dealing with either very big or very small numbers, there are a lot of ways to "get into trouble"
I don't know to what large numbers, small numbers and trouble you are referring. I did not get any such idea from this thread. Quite the opposite, the OP's question was about a feature that *prevents* you from getting into trouble in a particular situation.
| [reply] |