Given the following code (without virtual inheritance) :
class A
{
public:
virtual void f() = 0;
};
class B : public A
{
public:
virtual void f() {}
};
class C : public A
{
public:
virtual void f() {}
};
class D : public B, public C
{
/* some code */
};
int main()
{
D d;
return 0;
}
the code compile.
On the other hand , here :
class A
{
public:
virtual void f() = 0;
};
class B : virtual public A
{
virtual void f() {}
};
class C : virtual public A
{
virtual void f() {}
};
class D : public B, public C
{
/* some code */
};
int main()
{
D d;
return 0;
}
The compiler presents a compilation error:
no unique final overrider for 'virtual void A::f()' in 'D' .
Why is it different in the second code ?
Your first scenario hierarchy corresponds to:
F() F()
A A
| |
F() B C F()
\ /
D
Where D is not abstract, because there are two A subobjects in an object of type D: One that is made concrete by B through the lattice of B, and another that is made concrete through the lattice of C.
Unless you try to invoke the function F() on object of D there will not be any ambiguity.
Your second scenario hierarchy corresponds to:
F()
A
/ \
F() B C F()
\ /
D
In this scenario, the object D has a single Base class A sub object, and it must override and provide implementation of the pure virtual function in that subobject.
Herb Sutter's articles in Guru Of The Week(GOTW) are a nice read for Multiple Inheritance:
Multiple Inheritance Part I
Multiple Inheritance Part II
Multiple Inheritance Part III
With the virtual inheritance a D object has a single base-class A sub-object. This single sub-object can’t have two different implementations of a virtual function. In contrast, without virtual inheritance a D object has two distinct base-class A sub-objects, each with its own implementation of the function (which is OK until you try to call it on a D object, at which point you need to indicate which one you want).
Cheers & hth.
Related
Could I do this?
class A {
public:
virtual void aFoo() = 0;
};
class B : virtual public A {
public:
virtual void aFoo() { ... }
};
class D : public A {};
class C : public B, virtual public D {};
The issue is with the implementation of aFoo() in B and the lack of it in C. When it comes to compiling I see this error:
error: no unique final overrider for ‘virtual void A::aFoo()’ in ‘C’
Shouldn't it be possible to override a pure virtual method of a virtual base class in B?
Just edited the example to match the actual use case. Now looking at it in this simplified way I am not quite sure if this is even possible, let alone good design.
You need virtual inheritance in this way:
struct A {
virtual int f() = 0;
};
struct B : virtual A {
int f() { return 1; }
};
struct C : virtual A {};
struct D : B, C {};
int main() {
D d;
return d.f();
}
In the dupe I commented you can see this relation
A
/ \
B C
\ /
D
for virtual inheritance and
A A
| |
B C
\ /
D
without virtual inheritance.
In the second case D contains two function definitions. One is implemented and one isn't.
I have produced a minimal example to replicate the problem I am seeing with a more complex class hierarchy structure:
#include <string>
#include <iostream>
class A
{
protected:
virtual
~A() = 0;
};
inline
A::~A() {}
class B : public A
{
public:
virtual
~B()
{
}
std::string B_str;
};
class BB : public A
{
public:
virtual
~BB()
{
}
std::string BB_str;
};
class C : public A
{
protected:
virtual
~C()
{
}
virtual
void Print() const = 0;
};
class D : public B, public BB, public C
{
public:
virtual
~D()
{
}
};
class E : public C
{
public:
void Print() const
{
std::cout << "E" << std::endl;
}
};
class F : public E, public D
{
public:
void Print_Different() const
{
std::cout << "Different to E" << std::endl;
}
};
int main()
{
F f_inst;
return 0;
}
Compiling with g++ --std=c++11 main.cpp produces the error:
error: cannot declare variable ‘f_inst’ to be of abstract type ‘F’
F f_inst;
note: because the following virtual functions are pure within ‘F’:
class F : public E, public D
^
note: virtual void C::Print() const
void Print() const = 0;
^
So the compiler thinks that Print() is pure virtual.
But, I have specified what Print() should be in class E.
So, I've misunderstood some of the rules of inheritance.
What is my misunderstanding, and how can I correct this problem?
Note: It will compile if I remove the inheritance : public D from class F.
Currently your F is derived from C in two different ways. This means that an F object has two separate C bases, and so there are two instances of C::Print().
You only override the one coming via E currently.
To solve this you must take one of the following options:
Also override the one coming via D, either by implementing D::Print() or F::Print()
Make Print non-pure
Use virtual inheritance so that there is only a single C base.
For the latter option, the syntax adjustments would be:
class E : virtual public C
and
class D : public B, public BB, virtual public C
This means that D and E will both have the same C instance as their parent, and so the override E::Print() overrides the function for all classes 'downstream' of that C.
For more information , look up "diamond inheritance problem". See also Multiple inheritance FAQ
In C++ multiple inheritance with virtual base
i understand why could this code be ambiguous , but why it still complains when i specifically call a derived class method ?
class A {
public:
virtual void f() { cout << 1 ;}
};
class B :virtual public A {
public:
virtual void f() { cout << 2; }
};
class C :virtual public A {
public:
virtual void f() { cout << 3; }
};
class D : public B, public C {};
void main(){
D* d = new D();
d->B::f();
getch();
}
i would expect that it will run the B:f() method but i get :
"ambiguous inheritance of 'void A::f(void)'
The problem is that your function is virtual which means it overrides address in vtable. You can't override one address with two functions.
Without virtual qualifier there would be just ambiguity during the call which can be avoided just like you did specifying base class B::f()
The problem is that your class D has two final overriders of function f. See http://en.cppreference.com/w/cpp/language/virtual about final overrider.
In the code below, I get the following warning and error:
test.cpp:15: warning: direct base 'B' inaccessible in 'D' due to ambiguity
test.cpp:15: error: no unique final overrider for 'virtual void A::f()' in 'D'
But if I remove the virtual inheritance of B from A (i.e. struct B : public A), I only get the warning, no error.
struct A
{
virtual void f() = 0;
};
struct B : public virtual A
{
void f() {}
};
class C : public B
{};
struct D : public C, virtual B
{};
int main()
{
return 0;
}
Why? Is this the dreaded diamond?
It's because C inherits in a non-virtual way from B while D inherits in a virtual way from B. This gives you B two times including two f().
Try virtual inheritance of B in C.
Update: So why does it work when you remove the virtual inheritance in B from A? Because it changes the "final overrider". Without virtual in B from A and in C from B you have A two times: once in C (with the final override of f() in B) and once in the virtual B in D (with the final override of f() in B). If you add back the virtual inheritance in B to A, A will be present only once and there will be two final overrides competing to implement the pure f() from A, both in B, once from C and once from the virtual B.
As a workaround you could add a using to D, that is using C::f; or using B::f.
See C++ 10.3/2
Let's look at the definition of 'final overrider' from 10.3[class.virtual]/2
A virtual member function C::vf of a class object S is a final overrider unless the most derived class of which S is a base class subobject (if any) declares or inherits another member function that overrides vf.
In a derived class, if a virtual member function of a base class subobject has more than one final overrider the program is ill-formed.
With virtual inheritance from A, there is only one base class subobject of type A, and its virtual member function f() has more than one final overrider (one in each subobject of type B)
Without virtual inheritance from A, there are two different base class subobjects of type A, and their virtual member functions f() each has its own final overrider (one in each B subobject)
Virtual base sub-objects are 'shared' between all base sub-objects in a complete object. Since A is is shared between D::C::B and D::B it can't tell which B object should have its f() called as the override for A::f().
Consider:
#include <iostream>
struct A {
virtual void f() = 0;
virtual ~A() {}
};
struct B : virtual A
{
void f() { std::cout << "B\n"; }
};
struct C : virtual A
{
void f() { std::cout << "C\n"; }
};
struct D : C, B {};
int main() {
D d;
A *a = dynamic_cast<A*>(&d); // single shared A between B and C
a->f(); // Should B::f() be called, or C::f()?
}
The B and C base sub-objects in D both share the same A base sub-object. When we call A::f() a virtual look-up is done for the overriding function. But both B and C are trying to override it, so which one 'wins'? Does x->f() print "B" or "C"? The answer is that a program that gets into the situation is ill-formed.
When we eliminate the sharing by making B and C inherit non-virtually, then the separate A base sub-objects each have their functions overridden by unique base classes:
#include <iostream>
struct A {
virtual void f() = 0;
virtual ~A() {}
};
struct B : A
{
void f() { std::cout << "B\n"; }
};
struct C : A
{
void f() { std::cout << "C\n"; }
};
struct D : C, B {};
int main() {
D d;
// two different A objects
A *a1 = static_cast<A*>(static_cast<B*>(&d));
A *a2 = static_cast<A*>(static_cast<C*>(&d));
a1->f();
a2->f();
}
#include <iostream>
using namespace std;
class A { public: void eat(){ cout<<"A";} };
class B: public A { public: void eat(){ cout<<"B";} };
class C: public A { public: void eat(){ cout<<"C";} };
class D: public B,C { public: void eat(){ cout<<"D";} };
int main(){
A *a = new D();
a->eat();
}
I am not sure this is called diamond problem or not, but why doesn't this work?
I have given the defination for eat() for D. So, it doesn't need to use either B's or C's copy (so, there should be no problem).
When I said, a->eat() (remember eat() is not virtual), there is only one possible eat() to call, that of A.
Why then, do I get this error:
'A' is an ambiguous base of 'D'
What exactly does A *a = new D(); mean to the compiler??
and
Why does the same problem not occur when I use D *d = new D();?
The diamond results in TWO instances of A in the D object, and it is ambiguous which one you are referring to - you need to use virtual inheritance to solve this:
class B: virtual public A { public: void eat(){ cout<<"B";} };
class C: virtual public A { public: void eat(){ cout<<"C";} };
assuming that you actually only wanted one instance. I also assume you really meant:
class D: public B, public C { public: void eat(){ cout<<"D";} };
Imagine a slightly different scenario
class A { protected: int a; public: void eat(){ a++; cout<<a;} };
class B: public A { public: void eat(){ cout<<a;} };
class C: public A { public: void eat(){ cout<<a;} };
class D: public B,C { public: void eat(){ cout<<"D";} };
int main(){
A *a = new D();
a->eat();
}
If this would work, would it increment the a in B or the a in C? That's why it's ambiguous. The this pointer and any non-static data member is distinct for the two A subobjects (one of which is contained by the B subobject, and the other by the C subobject). Try changing your code like this and it will work (in that it compiles and prints "A")
class A { public: void eat(){ cout<<"A";} };
class B: public A { public: void eat(){ cout<<"B";} };
class C: public A { public: void eat(){ cout<<"C";} };
class D: public B, public C { public: void eat(){ cout<<"D";} };
int main(){
A *a = static_cast<B*>(new D());
// A *a = static_cast<C*>(new D());
a->eat();
}
That will call eat on the A subobject of B and C respectively.
Note that the compile error is on the "A *a = new D();" line, not on the call to "eat".
The problem is that because you used non-virtual inheritance, you end up with class A twice: once through B, and once through C. If for example you add a member m to A, then D has two of them: B::m, and C::m.
Sometimes, you really want to have A twice in the derivation graph, in which case you always need to indicate which A you are talking about. In D, you would be able to reference B::m and C::m separately.
Sometimes, though, you really want only one A, in which case you need to use virtual inheritance.
For a truly unusual situation, Neil's answer is actually wrong (at least partly).
With out virtual inheritance, you get two separate copies of A in the final object.
"The diamond" results in a single copy of A in the final object, and is produced by using virtual inheritance:
Since "the diamond" means there's only one copy of A in the final object, a reference to A produces no ambiguity. Without virtual inheritance, a reference to A could refer to either of two different objects (the one on the left or the one on the right in the diagram).
The error you're getting isn't coming from calling eat() - it's coming from the line before. It's the upcast itself that creates the ambiguity. As Neil Butterworth points out, there are two copies of A in your D, and the compiler doesn't know which one you want a to point at.
You want: (Achievable with virtual inheritance)
D
/ \
B C
\ /
A
And not: (What happens without virtual inheritance)
D
/ \
B C
| |
A A
Virtual inheritance means that there will be only 1 instance of the base A class not 2.
Your type D would have 2 vtable pointers (you can see them in the first diagram), one for B and one for C who virtually inherit A. D's object size is increased because it stores 2 pointers now; however there is only one A now.
So B::A and C::A are the same and so there can be no ambiguous calls from D. If you don't use virtual inheritance you have the second diagram above. And any call to a member of A then becomes ambiguous and you need to specify which path you want to take.
Wikipedia has another good rundown and example here