See following code:
#include <iostream>
struct A { // Interface
virtual void a() = 0;
virtual void x() = 0;
};
struct B {
virtual void a() { std::cout << "B" << std::endl; }
};
struct C : A, B {
void a() override { B::a(); }
void x() override { std::cout << "x" << std::endl; }
};
int main() {
C c;
c.a();
c.x();
}
This code works. The question is if it is the best/optimal solution.
I wonder if there is any trick which allow me not to create a() in C class.
Update: I corrected the code to show why B cannot inherit from A.
Update 2:
#include <iostream>
struct I1 {
virtual void i1() = 0;
};
struct I2 {
virtual void i2() = 0;
};
struct I3 : I1, I2 {};
struct A : I1 {
void i1() override { std::cout << "i1" << std::endl; }
};
struct B : A, I2 {
void i2() override { std::cout << "i2" << std::endl; }
};
int main() {
B b;
b.i1();
b.i2();
I3* ptr = dynamic_cast<I3*>(&b); // warning: dynamic_cast of ‘B b’ to ‘struct I3*’ can never succeed
std::cout << ptr << std::endl;
}
The question is: How to pass pointer to 'b' via interface? Why b cannot be casted to I3 interface?
CodePudding user response:
You can consider to split A
into 2 interfaces:
struct A1 { // Interface1
virtual void a() = 0;
virtual ~A1() {}
};
struct A2 { // Interface2
virtual void x() = 0;
virtual ~A2() {}
};
This way you can:
- Make
B
inherit fromA1
and implementa()
. - Make
C
inherit from A2 (to implementx()
) and fromB
to get the implementation fora()
.
See below:
#include <iostream>
struct B : public A1 {
virtual void a() override { std::cout << "B" << std::endl; }
};
struct C : public A2, public B {
virtual void x() override { std::cout << "x" << std::endl; }
};
int main() {
C c;
c.a();
c.x();
}
Note: I added virtual destructors to the base classes. See here why: When to use virtual destructors?.
CodePudding user response:
Depending on what you are trying to achieve you could use classes and make the function in B private so no one can try and use it. Something like:
class A { // Interface
public:
virtual ~A() {}
virtual void a() = 0;
virtual void x() = 0;
};
class B : public A{
private:
void x() { std::exception("Not Implemented"); };
public:
~B() override{}
void a() override { std::cout << "B" << std::endl; }
};
class C : public B {
public:
~C() override{}
void x() override { std::cout << "x" << std::endl; }
};
Other possible solutions are:
- Have B as a member of C, so you can call B.a() rather than multiple inheritance to make it a bit less confusing for readers
struct A { // Interface
virtual ~A() {}
virtual void a() = 0;
virtual void x() = 0;
};
struct B {
virtual void a() { std::cout << "B" << std::endl; }
};
struct C : A {
B b;
virtual ~C() override {}
void a() override { b.a(); }
void x() override { std::cout << "x" << std::endl; }
};
int main() {
C c;
c.a();
c.x();
}
You could use function pointers - but this means the functions must be static, and your base class becomes an abstract class
struct A { // Interface
virtual ~A() {}
virtual void x() = 0;
typedef void(*aFunction)();
aFunction a;
A() { a = nullptr; }
};
struct B {
static void a() { std::cout << "B" << std::endl; }
};
struct C : A {
virtual ~C() {}
void x() override { std::cout << "x" << std::endl; }
C() { a = B::a; }
};