Unclear about dynamic binding

Question

I am not understanding the concept of dynamic binding and overriding properly:

Here is some code:

class Cake {
    public void taste (Cake c) {
        System.out.println("In taste of Cake class");
    }
}

class ChocolateCake extends Cake {
    public void taste(Cake c) {
        System.out.println("In taste (Cake version) of ChocolateCake class");
    }
    public void taste(ChocolateCake cc) {
        System.out.println("In taste (ChocolateCake version) of ChocolateCake class");
    }
}


public static void main(String[] args)
{
    ChocolateCake cc = new ChocolateCake();
    Cake c = new ChocolateCake();
    Cake c1 = new Cake();
    Cake c2 = new ChocolateCake();

    c1.taste(cc);
    c1.taste(c);

    c2.taste(cc);
    c2.taste(c);
}

I expected:

In taste of Cake class
In taste of Cake class
In taste (ChocolateCake version) of ChocolateCake class" <----
In taste (Cake version) of ChocolateCake class

Actual:

In taste of Cake class
In taste of Cake class
In taste (Cake version) of ChocolateCake class <----
In taste (Cake version) of ChocolateCake class

If the object is of type ChocolateCake and I call cc which is also ChocolateCake, how come the compiler shows it's getting Cake as a parameter?

Answer 1

This is because Java uses both static and dynamic binding to choose a method to call in this case.

The line in question is this, right?

c2.taste(cc);

The compiler first chooses which method to call (static binding). Since c2 is of compile time type Cake, the compiler only sees the taste(Cake) method. So it says "call taste(Cake)".

Now at runtime, the runtime needs to choose which implementation of taste(Cake) to call, depending on the runtime type of c2. This is dynamic binding. Does it choose the one in Cake? Or the one in ChocolateCake? Since c2 is of runtime type ChocolateCake, it calls the implementation of taste(Cake) in ChocolateCake.

As you can see, the method that you thought would be called - taste(ChocolateCake) - is not even mentioned! This is because that is a different overload of the taste method, and because it is in the ChocolateCake class, which the compiler can't see. Why can't the compiler see? Because c2 is of compile time type Cake.

In short, the compiler decides which overload, the runtime decides which implementation.

Responding to your statement:

if the object is of type ChocolateCake ...

Only you know the object is of type ChocolateCake. The compiler does not. It only knows c2 is of type Cake because that's what its declaration says.

Answer 2

Since the reference type of the c2 variable is Cake the taste method having the Cake type parameter will be called.

This is because the Cake type does not have the taste method which takes a ChocolateCake instance, so you can't invoke that method from a Cake type reference variable.

Now secondly, in Java due to the mechanism of runtime polymorphism the overridden taste method of the ChocolateCake is being called instead of the version declared in the parent Cake class. This is due to fact at runtime the object which the Cake reference is pointing to, will be examined and the taste version of that particular instance will be invoked.

So due to the combination of these two effects you see that output.

If you change the reference type of c2 to ChocolateCake you would see that the output is:

In taste (ChocolateCake version) of ChocolateCake class 

when you invoke c2.taste(cc);, since now both the compiler and runtime agrees to call that taste(ChocolateCake cc) method in particular.