[Bug libstdc++/94545] std::to_integer(std::numeric_limits::max()) returns 0

[thomas.mercier.jr at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94545

--- Comment #2 from Thomas Mercier <thomas.mercier.jr at gmail dot com> ---
I thought that might be the response. Then why does it compile? The fact that
it does, and produces a result is surprising. I don't know what the standard
says, but just looking at cppreference it says that specializations are
provided for arithmetic types. We could have a more sane default than value
initialization
(https://github.com/gcc-mirror/gcc/blob/master/libstdc%2B%2B-v3/include/std/limits#L321)
that produces a compilation error if it's unspecified (something like below).

template<typename _Tp>
struct mylimits
{
  static constexpr _Tp max() noexcept {
    static_assert(sizeof(_Tp) == 0, "There is no specialization of mylimits for
this type."); 
  }
};

template<>
struct mylimits<bool>
{
  static constexpr bool max() noexcept { return true; }
};

// and so on...

Gives you:
$ g++ -std=c++17 test.cc
test.cc: In instantiation of ‘static constexpr _Tp mylimits<_Tp>::max() [with
_Tp = int]’:
test.cc:21:34:   required from here
test.cc:5:5: error: static assertion failed: There is no specialization of
mylimits for this type.
     static_assert(sizeof(_Tp) == 0, "There is no specialization of mylimits
for this type.");