I need to implement robust minkowski sum operations on polygons. I have seen that this could be implemented easily using boost::polygon, the tutorial is here http://www.boost.org/doc/libs/1_52_0/libs/polygon/doc/gtl_minkowski_tutorial... The example works fine, however i need more precision than "int" (32 bits on my plateform win32-msvc10-32bits). I tried to replace to replace int -> double typedef boost::polygon::point_data< int > point; with typedef boost::polygon::point_data< double > point; But this leads to strange errors deep into the code (operator< is "bad") if (!_Pred(_Left, _Right)) return (false); else if (_Pred(_Right, _Left)) _DEBUG_ERROR2("invalid operator<", _File, _Line); called after scan_data_.insert(*insert_itr); // line 1197 in scan_arbitrary.hpp What can i do to avoid it, if i use "long long" type it works but it sends a lot of warning at compile time. Any idea of what i should do ? Is "double" type supported ?