About template specialization (substitute, define_aggregate)

[Pespon]

Hi

I've been experimenting with p2996, and have encountered use cases for which I don't understand the behavior, and would appreciate a helping hand.

Use cases
Considering the following template:

template<int n>
struct generated_template_test;

Case *1:
If I specialize it in the root namespace via:

template<>
struct generated_template_test<42> {};

I observe that the return value of members_of(^::).size() increments by 1, which is what I expected.

Case *2:
On the other hand, if I do the specialization like this (in the root namespace):
static_assert(is_type(define_aggregate(substitute(^generated_template_test, {std::meta::reflect_value(42)}), {})));

I observe that the return value of members_of(^::).size() does not change, which I find strange, whereas the specialized version of template can be instantiated perfectly well.

Case *3:
Furthermore, if I specialize the template in a namespace (also works with a class):

namespace a_namespace
{
    static_assert(is_type(define_aggregate(substitute(^generated_template_test, {std::meta::reflect_value(42)}), {}))));
}

The situation seems to be perfectly similar to *2.

Questions
This raises the following questions:
A/ Is case *3 allowed (class specialization in a namespace)?
B/ Is it normal for the return of members_of(^::).size() and members_of(^a_namespace).size() not to change once the template has been specialized?
C/ Is there a way to browse all the specializations of a template class?

Thank you for your time, I really appreciate your work.