Re: [Boost-users] [math] fpclassify.hpp fails to compile with Intel icc 11.1

Subject: Re: [Boost-users] [math] fpclassify.hpp fails to compile with Intel icc 11.1
From: Gabriel Redner (gredner_at_[hidden])
Date: 2011-01-03 12:21:44

Further experiments show that the issue is present in boost 1.42, but
fixed in boost 1.45.  Sorry for the noise :)

Thanks,
-Gabe

On Mon, Jan 3, 2011 at 11:53 AM, Gabriel Redner <gredner_at_[hidden]> wrote:
> Hi all,
>
> The following code fails to compile with icc 11.1 under Linux:
>
> ==========
> #include <boost/math/special_functions/fpclassify.hpp>
> int main() {}
> ==========
>
> The full error is:
>
> ==========
> /usr/include/c++/4.4.5/cmath(500): error: identifier
> "__builtin_fpclassify" is undefined
>        return __builtin_fpclassify(FP_NAN, FP_INFINITE, FP_NORMAL,
>               ^
>          detected during:
>            instantiation of
> "__gnu_cxx::__enable_if<std::__is_arithmetic<_Tp>::__value,
> int>::__type std::fpclassify(_Tp) [with _Tp=long double]" at line 104
> of "/usr/include/boost/math/special_functions/fpclassify.hpp"
>            instantiation of "bool
> boost::math_detail::is_nan_helper(T, const boost::true_type &) [with
> T=long double]" at line 137 of
> "/usr/include/boost/math/special_functions/fpclassify.hpp"
>            instantiation of "int
> boost::math::detail::fpclassify_imp(T, const
> boost::math::detail::generic_tag<true> &) [with T=long double]" at
> line 242 of "/usr/include/boost/math/special_functions/fpclassify.hpp"
>
> compilation aborted for main.cpp (code 2)
> ==========
>
> The same code compiles fine under icc 12.0.
>
> I also tried explicitly adding:
> #define BOOST_MATH_DISABLE_STD_FPCLASSIFY
> before including the header, but got the same error.
>
> The math library docs say that icc 10.0 and later are fully supported
> with no serious caveats, so I presume this is a bug.  Should I file a
> ticket?  Please let me know if there is any other information about my
> system which could be useful to know.
>
> Thanks,
> -Gabe
>