How to detect underflow in pow()?

[LuisG]

In this case, checking for FP_SUBNORMAL won't help much, since the pow function just returns a plain, old 0.0 double, in case of underflow.

Luis

[Wolfgang]

Dear Luis,

I agree that the question on FP_SUBNORMAL has only little to do with the original question, but when would it be set? Once that C99 if fully implemented? Or is it more an omission of the pow function?

[LuisG]

Full C99 compliance would require functions/macros such as these.

At this point, I'm not sure exactly what it would take for fpclassify to return a status of FP_SUBNORMAL. My understanding is that even the full C99 spec does not require the pow function to return anything other than 0.0 on underflow, and I would think that 0.0 is not a subnormal number but, to be honest, I'm not sure.

Luis

[Wolfgang]

Thanks, Luis!

Probably you are right, after reading more about subnormal numbers I would not expect this message to show up, as it is not really related to underflow. It is related to this thread.

So even if this thread did not solve my initial problem, once in a while I learn something new, about negative zeros, subnormal numbers, funny animals