This code compile, run and print 0 on Tru64 Unix. On Linux, it
crashes with a segfault. Is it legal to call nanl(0)? I tried
checking the C standard, but that didn't make me wiser. :(
#include <math.h>
#include <stdio.h>
int main() {
long double a = nanl(0);
printf("%Lf\n", a);
return 0;
}