C4 has to be in p1, not p2.

What “test case” are you running? Just for the hell of it, I put it into my IDE, and sure enough, C4.m() does not override C2.m() like I stated above.

You might want to read the link I posted above.

—————————-
package p1;

public class C1 {
int m() {return 1;}
}

—————————-
package p1;

public class C2 extends C1 {
int m() {return 2;}
}

—————————-
package p1;

import p2.C4;

public class Main {
public static void main(String[] args) {
C1 c = new C4();
System.out.println(c.m());
}

}

—————————-
package p2;

import p1.C2;

public class C3 extends C2 {
int m() {return 3;}
}

—————————-
package p2;

public class C4 extends C3 {

int m() {return 4;}
}
—————————-
Output:
2

I think there’s no way to call C2.m() from C4.m(). Granted, JVM will allow such a call via INVOKESPECIAL, but in the compiler super always refers to direct superclass and there is no other way to do that in Java.

Even if everything was public (instead of package private), you cannot access C2.m() from C4.m() (unless of course C3.m() does not exist or class C3 providers some sort of method to get it like C3.getC2m()). Regardless, if you’re trying to do anything like this, it’s time to rethink your design.

You still have a wrong mental model :) Which method is called does not depend on the caller, but rather on the target class in the call. Compiler resolved each call to the string Class.method(argsTypes)returnType. “Class” is a part of the message and vtable is selected by it. It’s actually not so different from C++ if you think about implementation, but getting to the vtable is quite harder.

do you want even more fun…jump into aspect-oriented progamming.