https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78939
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2016-12-28
CC| |jakub at gcc dot gnu.org,
| |jason at gcc dot gnu.org,
| |redi at gcc dot gnu.org
Component|c++ |libstdc++
Summary|<tuple> interferes with |[C++17] <tuple> interferes
|structured binding from |with structured binding
|struct |from struct
Ever confirmed|0 |1
--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
That is a consequence of the recent P0490R0 GB 20 change and <utility> having:
/// Finds the size of a given tuple type.
template<typename _Tp>
struct tuple_size;
// _GLIBCXX_RESOLVE_LIB_DEFECTS
// 2770. tuple_size<const T> specialization is not SFINAE compatible
template<typename _Tp, typename = void>
struct __tuple_size_cv_impl { };
template<typename _Tp>
struct __tuple_size_cv_impl<_Tp,
__void_t<decltype(tuple_size<_Tp>::value)>>
: integral_constant<size_t, tuple_size<_Tp>::value> { };
// _GLIBCXX_RESOLVE_LIB_DEFECTS
// 2313. tuple_size should always derive from integral_constant<size_t, N>
template<typename _Tp>
struct tuple_size<const _Tp> : __tuple_size_cv_impl<_Tp> { };
template<typename _Tp>
struct tuple_size<volatile _Tp> : __tuple_size_cv_impl<_Tp> { };
template<typename _Tp>
struct tuple_size<const volatile _Tp> : __tuple_size_cv_impl<_Tp> { };
http://www.open-std.org/jtc1/sc22/wg21/docs/lwg-defects.html#2770
So, either the GB 20 change has been added without considering the effect on
the existing <utility>, or there is some further <utility> change needed, or
libstdc++ doesn't do what the DR requests. But the current DR wording talks
about not having the value member, which is something GB 20 doesn't require any
more.
