DevOps, CI/CD and Automation

Firstly, relevant details

OS/Arch: Solaris 11.2 SRU 15.5.1 on SPARC (T4)

Studio Version - Solaris Studio 12.4 (From solarisstudio IPS publisher, current)

Unless I am missing something, it seems when compiling using the gnu C++ ABI std::numeric_limits<T>infinity() isn't returning the correct values.

Sample code attached compiles and works correctly using g++ 4.8.2 (-std=c++11) and Solaris Studio 12.5 Beta (so it is a solved problem)

#include <cstdlib>
#include <cmath>
#include <cassert>
#include <limits>
#include <iostream>

int main() {
        double inf = std::numeric_limits<double>::infinity();
        double neginf = -std::numeric_limits<double>::infinity();
        float finf = std::numeric_limits<float>::infinity();
        float fneginf = -std::numeric_limits<float>::infinity();
        long double linf = std::numeric_limits<long double>::infinity();
        long double lneginf = -std::numeric_limits<long double>::infinity();
        std::cout << "Inf: " << inf << std::endl;
        std::cout << "-Inf: " << neginf << std::endl;
        std::cout << "float Inf: " << inf << std::endl;
        std::cout << "float -Inf: " << neginf << std::endl;
        std::cout << "long double Inf: " << inf << std::endl;
        std::cout << "long double -Inf: " << neginf << std::endl;

        assert(std::isinf(inf));
        assert(std::isinf(neginf));
        assert(std::isinf(finf));
        assert(std::isinf(fneginf));
        assert(std::isinf(linf));
        assert(std::isinf(lneginf));

        return EXIT_SUCCESS;
}

For 12.4 it returns

Inf: 1.06048e-314
-Inf: -1.06048e-314
float Inf: 1.06048e-314
float -Inf: -1.06048e-314
long double Inf: 1.06048e-314
long double -Inf: -1.06048e-314
Assertion failed: std::isinf(inf), file foo.cc, line 21
Abort (core dumped)

Its certainly not giving back HUGE_VAL{,L,F}

Stripping out the assertions and isinf checks the above also does what you expect it to do using RW libCstd, stlport4 and stdcxx4.

Is there any quick and dirty fix?