[Bug c++/46382] New: constexpr vs. static_assert in constexpr ctors

[bkoz at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46382

           Summary: constexpr vs. static_assert in constexpr ctors
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: bkoz@gcc.gnu.org


The current implementation has the nice feature that in addition to a return
statement, a constexpr function can also contain the following:

1) typedefs
2) static_asserts

Both make sense. Hurray! However, this is different for member functions and
so-called "special member functions." In particular, constructors. Here's an
example of something that would be nice if it worked.


---


#include <type_traits>

template<typename _Tp>
 struct A
  {
    int _M_i;

    constexpr A(int i) : _M_i(i)
    {
#if BUT_WE_WANT_IT_HERE_TOO
      static_assert(std::is_fundamental<_Tp>::value, "no"); // not really
#endif
    }


    constexpr bool notwhatyouwant() 
    { 
      typedef _Tp type; // ok
      static_assert(std::is_fundamental<_Tp>::value, "no"); // ok
      return _M_i == 0; 
    }
  };


int main()
{
  constexpr A<int> obj(5);

  constexpr bool b = obj.notwhatyouwant();
  return 0;
}