[Voted into the WP at the March, 2011 meeting as part of paper N3262.]

Proposed resolution (November, 2010) [SUPERSEDED]:

1. Add a new paragraph at the end of 13.8.3.4 [temp.dep.constexpr]:

An id-expression is value-dependent if it names a member of an unknown specialization.

2. Change 13.8.3.5 [temp.dep.temp] paragraphs 2-3 as follows:

An integral A non-type template-argument is dependent if its type is dependent or the constant expression it specifies is value-dependent.

A non-integral Furthermore, a non-type template-argument is dependent if its type is dependent or it has either of the following forms

qualified-id

& qualified-id

and contains a nested-name-specifier which specifies a class-name that names a dependent type the corresponding non-type template-parameter is of reference or pointer type and the template-argument designates or points to a member of the current instantiation or a member of a dependent type.

Given

    template <const char *N> struct A { static const char *p; };
    template <class T>
       struct B { static const char c[1]; typedef A<B<T>::c> C; };
    template <class T>
       struct D { static const char c[1]; typedef A<c> C; };

13.8.3.5 [temp.dep.temp] says that B<T>::c is dependent because it is used as a non-integral non-type template argument and it is a qualified-id with a nested-name-specifier that names a dependent type.

There doesn't seem to be anything to say that c in the definition of D<T>::C is dependent, which suggests that D<T>::C is the same type for all T, which is clearly false.

Instead of this special rule in 13.8.3.5 [temp.dep.temp], 13.8.3.4 [temp.dep.constexpr] should say that the address of a member of a dependent type is value-dependent, regardless of whether the address is written with a qualified-id.