Is there a way to know if a variable is NaN, +infinity, -infinity?

[zaikunzhang]

@CRquantum

If using IEEE_ARITHMETIC intrinsic module may lead to discrepancy due to -Ofast flag.
Then perhaps we can just use below to check NaN, +inf, -inf,

About if x is NaN,
like @vmagnin suggested, we can define a function to check if x = x or not.

About if x is +inf or -inf, can we use intrinsic HUGE function like below?

if x > huge(x), it means x is +inf
if x< -huge(x), it means is -inf

If you want some replacement that works (for the time being), you may try my GitHub repo mentioned above and in the other thread quoted before. They have been tested with 9 compilers (all that I could find) with aggressive optimization flags. I will be very happy to see where they do not work, and if possible, some suggested alternative implementation will be highly appreciated. Many thanks!