return value from assignment

[Barkat Mardhani]

Hi All:
I guess I learned in this forum that any assignment within an expression will return the assigned value to the next context level so:
boolean a, b;
a = (b=true);
then a will also be assigned true. In the following code, the bold assignments should return nulls and hence most specific classes (B,B) should be invoked. But assignments return (a,b) and (A,B) method is invoked. Why?

class A {}
class B extends A {}
class C extends B {
  static void m(A x, A y) {System.out.print("AA");}
  static void m(A x, B y) {System.out.print("AB");}
  static void m(B x, A y) {System.out.print("BA");}
  static void m(B x, B y) {System.out.print("BB");}
  public static void main(String[] args) {
    A a;
    B b;
    m(null,null);
    m([B]a=null,b=null[/B]);
    m(b, a);
  }
}

Thanks
Barkat
[ September 19, 2002: Message edited by: Barkat Mardhani ]
[ September 19, 2002: Message edited by: Barkat Mardhani ]
[ September 19, 2002: Message edited by: Barkat Mardhani ]

[Valentin Crettaz]

First of all, bold tags do not appear within the CODE tags. What Barkat means by "bold assignment" is the line
m(a=null,b=null);
The reason why the output is BBABBA is because the most specific method is invoked when multiple method signature match the invocation expression. In this case, the most specific method (as per JLS 15.12.2.2 Choose the Most Specific Method) is the method taking two parameters of type B. B is more specific than A.

[Barkat Mardhani]

Hi Val:
Thanks for your reply. I understand first and third call for m method. It is the second call that is confusing:
m(a=null,b=null)
should result in:
m(null,null)
as I explained in previous post. Hence m(null,null) should call m(B,B) not m(A,B)... but it is calling m(A,B) why...

[Valentin Crettaz]

No, because in
m(a=null,b=null)
we have an indication of which method to invoke since a and b are of type A, respectively B, even though their value is null.

[Barry Gaunt]

I agree with Valentin. The only one where the compiler has to think a bit is the m(null,null) case. In the others the type of the passed arguments is well defined.
-Barry

[Barkat Mardhani]

Let me ask in another way:
boolean a, b;
a = (b=true);
Will you agree that (b=true) expression will yield true and that will be assigned to a?
[ September 20, 2002: Message edited by: Barkat Mardhani ]

[Valentin Crettaz]

yes I agree. And?

[Barry Gaunt]

Doesn't change the types of a and b, they're still
booleans

[Barkat Mardhani]

therefore (a=null) should result in null
and (b=null) should result in null
therefore (a=null,b=null) should result in (null,null)
therefore m(a=null,b=null) should result in m(null, null)
therefore m(null, null) should call most specific method m(B,B) not m(A,B)...Right? :roll:

[Barry Gaunt]

We are back to the first example with A a; B b; ?
When the compiler sees m(a=null, b=null) it sees
a call to m(A, B) because a is of type A and b is of type B. It has one of those to call so it does that.
The fact that a value is being assigned to a and b at the time is of no matter. Instead of assigning nulls you could be doing m(a=new A(), b= new B()).
-Barry

[Barkat Mardhani]

Thanks Barry and Val...

[Barry Gaunt]

If java was a true object oriented language (I have my head on offer saying this) we could do things like System.out.println((a=null).type)). That would help a lot in clarifying these cases.
Cheers
-Barry

[Valentin Crettaz]

class A{
public static final Class cl = A.class;
}
public class Test {
public static void main(String[] args) {
A a;
System.out.println((a=null).cl);
}
}

Output is:
class A
[ September 20, 2002: Message edited by: Valentin Crettaz ]

[Barry Gaunt]

Valentin, here it is ---> on a plate.
Hey Hang on! I take that back...
<------
You made that up. Boo! Unfair!
I want Java to tell me the type of result of an expression. Like built in...
On the other hand, it is rather neat.
-------->
[ September 20, 2002: Message edited by: Barry Gaunt ]

[Valentin Crettaz]

You made that up. Boo! Unfair!
Well, I did it with what Java provides me
I want Java to tell me the type of result of an expression. Like built in...
Sure it would be nice
You want built-ins, there, get some:

public class Test {
public static void main(String[] args) {
Integer i;
System.out.println((i=null).TYPE);
}
}

Actually all wrapper classes provide a constant called TYPE which resolves to the primitive type represented by the class type. Not quite what we wanted, but they are built-ins
On the other hand, it is rather neat.
isn't it? At least it helps clear Barkat's doubts

[Barkat Mardhani]

Thank you guys. That was enlightening. I guess I was thinking along the line that m(a=null,b=null) statement will try to resolve the VALUE of 'a=null' assignment. But there is another twist to it i.e. what is TYPE of 'a=null' assignment, and m(a=null,b=null) is more worried about TYPE than VALUE for correct method invocation.
Please tell me that I got it, because I hate to waste your time...
[ September 20, 2002: Message edited by: Barkat Mardhani ]
[ September 20, 2002: Message edited by: Barkat Mardhani ]

[Barry Gaunt]

Barkat, yes that's about it (I think )
-Barry