float as a key in std::map

[S__B]

I think one of those return statements might be wrong.

Thanks Lindley for pointing out a problem.

How do you access the keys in the vector? What operations do you do to say "get me my value in the vector" and I will place it in the map"? Do not assume that this is even safe to do with floating point values.

To access the float, possibly a floating point CPU instruction is done which could cause a slight modification of the number, and that thing you never expected to happen just happens. Floating point trouble not only happens during calculation -- passing them by value, assignment, etc. could lead to these issues.

Paul, you are suggesting a serious flaw in this implementation. I thought that when a value is fetched into an internal register, it is just mirrored (a binary mirror) so that it won't change. Another question is that would this "problem" be more pronounced if it is compiled for 32-bit machines, but run on 64-bit machine. I am not sure, I know how it will handle the internal registers in such case. Or the behavior will be similar on either machines.
Thanks,
SB