Migrate from boost::math::isnan to std::isnan

[SunBlack]

There is a single boost::math::isnan left, where I don't have an idea what boost returns

pcl/apps/in_hand_scanner/src/input_data_processing.cpp

Line 249 in ee3fe6f

if (!boost::math::isnan (it_n->getNormalVector4fMap ()))

Error during compiling with Clang 6, if I replace this line:

/media/sf_pcl/apps/in_hand_scanner/src/input_data_processing.cpp:249:10: error: no matching function for call to 'isnan'
    if (!std::isnan (it_n->getNormalVector4fMap ()))
         ^~~~~~~~~~
/usr/bin/../lib/gcc/x86_64-linux-gnu/7.3.0/../../../../include/c++/7.3.0/cmath:618:3: note: candidate function not viable: no known conversion from 'pcl::Vector4fMapConst' (aka 'const Map<const Matrix<float, 4, 1>, Eigen::Aligned>') to 'float' for 1st argument
  isnan(float __x)
  ^
/usr/bin/../lib/gcc/x86_64-linux-gnu/7.3.0/../../../../include/c++/7.3.0/cmath:626:3: note: candidate function not viable: no known conversion from 'pcl::Vector4fMapConst' (aka 'const Map<const Matrix<float, 4, 1>, Eigen::Aligned>') to 'double' for 1st argument
  isnan(double __x)
  ^
/usr/bin/../lib/gcc/x86_64-linux-gnu/7.3.0/../../../../include/c++/7.3.0/cmath:631:3: note: candidate function not viable: no known conversion from 'pcl::Vector4fMapConst' (aka 'const Map<const Matrix<float, 4, 1>, Eigen::Aligned>') to 'long double' for 1st argument
  isnan(long double __x)
  ^
/usr/bin/../lib/gcc/x86_64-linux-gnu/7.3.0/../../../../include/c++/7.3.0/cmath:639:5: note: candidate template ignored: substitution failure [with _Tp = Eigen::Map<const Eigen::Matrix<float, 4, 1, 0, 4, 1>, 16, Eigen::Stride<0, 0> >]: no type named '__type' in '__gnu_cxx::__enable_if<false, bool>'
    isnan(_Tp __x)
    ^
1 error generated.

Any idea what do to with this line? (I have no OpenNI on my debug machine, so I cannot step into and see what the code is doing)

[taketwo]

I think the line in question can be replaced with:

if (!it_n->getNormalVector4fMap ().hasNaN ())

[SunBlack]

Sure we don't change behavior with this change (compared to now)?

[taketwo]

I think what I suggested will work the way the author intended. I'm not sure if the current code behaves the same though. I don't understand how it's possible that boost::math::isnan supports an Eigen type argument.