Incorrect CBO for Generic Classes

[aaghamohammadi]

Suppose we have a generic class Box. The CBO returns 2 for the following class:

public class Coupling6 {
    private Box<Integer> integerBox = new Box<Integer>();
}

However, I think it should return one.

[mauricioaniche]

CK indeed returns 1 here. See the test I just created for this: 358b521

If that makes sense to you, I'll close the issue.

[aaghamohammadi]

This is interesting! When you only create Coupling6, the returned value is 2. However, when you create Coupling6 plus Box the returned value is 1.

[mauricioaniche]

Yeah, that's what happens when you are creating 'fake' examples. If there are things that can't really be resolved, I try to approximate.

There was a bug in inexistent generic classes. Hopefully, I just fixed it: a69e34f

[mauricioaniche]

Wrote more tests, it seems to work, @aaghamohammadi

2b63d1b

[aaghamohammadi]

This is strange. Really strange!

Consider the following code:

public class Coupling61 {
	private Box2<Integer> integerBox = new Box2<Integer>();
}

public class Coupling62 {
	private Box2<A> integerBox = new Box2<A>();
}

public class Coupling63 {
	private Box<A> integerBox = new Box<A>();
} 

CBO is correct (1,2 ,2).

However, for the following code:

public class Coupling61 {
	private Box<Integer> integerBox = new Box<Integer>();
}

public class Coupling62 {
	private Box2<Integer> integerBox = new Box2<Integer>();
}

public class Coupling63 {
	private Box<A> integerBox = new Box<A>();
}

CBO is incorrect (expected:1,1,2 --- output:1,1,1) .

[mauricioaniche]

Your Coupling63 example is the same in both snippets. I tried one more combination there, and it works.

Can you write a failing test?

[aaghamohammadi]

Yes. I have written a test that fails. I sent a pull request which includes that test.