confusion in methdo overloading with var args

[Astha Sharma]

class Test
{
void A(Object... a){
	System.out.println("in Object");
}

void A(int... a){
	System.out.println("in int");
}
public static void main(String[]args)
{
	Test a=new Test();
	a.A(new Integer(3));
}
}

produces output-

in int

I have some questions regarding this code-

1. Why the call to method A is not ambigous?

2. Priority order for method call is Widening > Boxing > Varargs. Then why unboxing is prefered over widening here?

[Helen Ma]

1) When line 13 is commented out, the two overloaded methods are not ambiguous as one takes Object while the other one takes int.
2) When line 13 is uncommented out and executed, int... beats Object ... because unboxing costs less time to do than widening. But this is just my guess.

A(int...x) is equivalent to A(in[] x)
A (Object... x) is equivalent to A(Object[] x)

I think unboxing a set of Integer object, create an array of int cost less time for the compiler to do.
On the other hand, widening a set of Integer object, means creating an array of Object , and assign the Integer object as each element in this Object array, cost more time to do.

I hope someone else can help.

[Sagar Rohankar]

I think, here compiler choosing the most nearest matched object in the hierarchy, means Integer extends Object, and for "int" argument Integer matches first and thus it get called.

[Astha Sharma]

Thanks Helen and Sagar for the reply.
I'm getting confused because it is not following the priority order Widening > Boxing > Varargs.
Priority of widening is greater than boxing. So, shouldn't it be greater than unboxing also? Why compile is ignoring the priority order here?

[Helen Ma]

I guess unboxing has higher priority than widening in this case.

Can anyone help?

[Henry Wong]

Astha Sharma wrote:Thanks Helen and Sagar for the reply.
I'm getting confused because it is not following the priority order Widening > Boxing > Varargs.
Priority of widening is greater than boxing. So, shouldn't it be greater than unboxing also? Why compile is ignoring the priority order here?

I am more interested to know where you got the idea of priority of "Widening > Boxing > Varargs" -- as I don't recall ever seeing anything in the JLS that mentions this........ actually, admittedly, I find the section about method overloading, the "most specific" rule, and how it is related to variable arity to be very confusing.

Henry

[Astha Sharma]

Henry Wong wrote:I am more interested to know where you got the idea of priority of "Widening > Boxing > Varargs" -- as I don't recall ever seeing anything in the JLS that mentions this........ actually, admittedly, I find the section about method overloading, the "most specific" rule, and how it is related to variable arity to be very confusing.

It is written in k&b chapter 3, Assignment, page number 250

Widening beats boxing
Widening beats var-args
boxing beats var-args

Also see the thread