Make a template type be more specific (T => Certain<X>) to help content-assist

Question

How to make a template type more specific internally to help content assist?

template<class T>class B{  //note: in real case, it has more template parameter
    public: void f(){}
};
template<class B1>class C{ //<-- I know for sure that B1 derived from "B<something>".    
    B1* b;
    void test(){
        b->
             ^ ctrl+space doesn't show f()
    }
};

My poor workaround is to create template specialization at class C, but it will confuse content-assist in another way.

Below is another workaround but it is quite tedious.
I have to reflect the template argument at B and use such reflect at C one-by-one.

template<class T>class B{  
    public: using reflectiveT=T;
    /* other T e.g. reflectiveT2=T2 , ... */
    public: void f(){}
};
template<class B1>class C{ 
    using BX=B<B1::reflectiveT>; //B<B1::reflectiveT1,..T2,...T3> ... tedious
    BX* b;
    void test(){
        b->
             ^ ctrl+space will show f()
    }
};

Issues:

• It suffer maintainabilty problem, when I want to refactor B to have more/less template arguments later.
• If BX is happen to be a class that derived from B<something>, BXT will != BX.

I dream for something like :

template<class B1>class C{ 
    using BX=B<...> as base of B1;   //????
};

I may too rely on content-assist, but it greatly helps me to code very complex classes.

Edit

I can't just pass Args inside B as template parameter of C , because C may act wrong.

For example, B<D>::callback will be call instead of D::callback in the below code (demo):-

class x{};
template<class T>class B{
    public: static void callback(){ std::cout<<"B<D>::callback()";       }
};
class D : public B<D>{ //
    public: static void callback(){ std::cout<<"D::callback()"; }
};
template<class... Args>class C{ 
    using BX=B<Args...>;
    BX* b;
    public: void test(){
        BX::callback(); 
        //^ will invoke B<D>::callback (wrong)
        //  instead of D::callback
    }
};
int main(){
    C<D> c;  c.test();  //print "B<D>::callback()"
}

Edit: Simplify question a lot.

Answer 1

You can separate the implementation and the interface for each function:

#include <iostream>

// Base class with default implementations and common interface
template<class T> class B {
    static void callback_impl() { std::cout<<"B<T>::callback()\n"; }
    void f_impl() { std::cout << "B<T>::f()\n"; }
public:
    static void callback() { T::callback_impl(); }
    void f() { static_cast<T*>(this)->f_impl(); }
};

class D1 : public B<D1>{ // D1 overrides callback
    friend class B<D1>;
    static void callback_impl(){ std::cout<<"D1::callback()\n"; }
};
class D2 : public B<D2> { // D2 overrides f
    friend class B<D2>;
    void f_impl() { std::cout << "D2::f()\n"; }
};

template<class... Args>class C{
    using BX=B<Args...>;
    BX b;
    public: void test(){
        BX::callback();
        b.f();
    }
};

int main(){
    C<D1> c1;  c1.test();  //print "D1::callback()\nB<T>::f()\n"
    C<D2> c2;  c2.test();  //print "B<T>::callback()\nD2::f()\n"
}