Why the need for a using-declaration naming a constructor?

Question

Both paragraphs, 7.3.3.p1 and p3, in the C++11 Standard, make reference to a using-declaration naming a constructor. Why is this necessary? The code below shows that the base class A's constructors are seen in the derived class B as expected.

class A
{
    int i;

    public:
    A() : i(0) {}
    A(int i) : i(i) {}
};

class B : public A
{
    public:
//  using A::A;
    A f1() { return A(); }
    A f2() { return A(1); }
};

int main()
{
    B b;
    A a1 = b.f1();
    A a2 = b.f2();
}

If I comment out using A::A; above nothing changes in the program execution.

Answer 1

That's meant for inheriting the non-default constructors from the parent class, which is A(int i) in this case. Change your declaration in main like so:

int main()
{
    B b(42);  // Want to forward this to A::A(int)
    ...
}

Without the using A::A clause, you'll get the following compiler error (at least from g++ 4.8.0):

co.cpp: In function ‘int main()’:
co.cpp:20:11: error: no matching function for call to ‘B::B(int)’
     B b(42);
           ^
co.cpp:20:11: note: candidates are:
co.cpp:10:7: note: B::B()
 class B : public A
       ^
co.cpp:10:7: note:   candidate expects 0 arguments, 1 provided
co.cpp:10:7: note: constexpr B::B(const B&)
co.cpp:10:7: note:   no known conversion for argument 1 from ‘int’ to ‘const B&’
co.cpp:10:7: note: constexpr B::B(B&&)
co.cpp:10:7: note:   no known conversion for argument 1 from ‘int’ to ‘B&&’

but if you add the using A::A declaration back, it compiles cleanly. b(42) ends up calling A::A(int i).