[jdt-dev] Unexpected behaviour in anonymous classes management

[Antonio Cuomo <myskin83@xxxxxxxxx>]

Hi, I'm using jdt to extract informations from java source code. 
Basically, I organize informations given back by subtypes of 
IJavaElement.  I've found a strange behaviour when dealing with 
anonymous classes. As I expected, they come to the user as object of 
kind IType. The problem I noticed is with the fully qualified name. 
Suppose a situation like this:

package first
public class Example{

   public void methodA(){
       // just a random interface- does not matter for the example 
       Comparable c= new Comparable{
               public int compareTo(Object o) {
               //body does not matter here
               return 0;
               }
       }
       Comparable c1= new Comparable{
               public int compareTo(Object o) {
               //body does not matter here
               return 0;
               }
       }
   }
}

Here i get back for the first anonymous class (the one whose instance is 
assigned to c) the fully qualified name first.Example.1, while for the 
second (the one whose instance is assigned to c1) the fully qualified 
name is first.Example.2, a situation which i find correct.

But consider the case in which the two definitions lie in two different 
methods:

package first
public class Example{

   public void methodA(){
       // just a random interface- does not matter for the example 
       Comparable c= new Comparable{
               public int compareTo(Object o) {
               //body does not matter here
               return 0;
               }
       }
   }
   public void methodB(){
       Comparable c1= new Comparable{
               public int compareTo(Object o) {
               //body does not matter here
               return 0;
               }
       }
   }
}

Here i get back the same fully qualified name (first.Example.1) for both 
classes. Is this behaviour intentional?  I think the reason is in the 
way jdt deals with the
occurrenceCount field: since the parents  of the  classes  (the methods) 
are different, the occurrenceCount goes back from 1. But this gives back 
same fq name for the classes, and I'm not sure this is correct. Hope to 
hear you, thanks for the attention.