[java] Closure in constructor breaks assumption

[singpolyma]

The following code says "Types that derive from a native class must have its super() call as the first statement in the constructor" and generates code where the extra variables generated for the closure come before the super call, even though the super call is clearly first in my code:

private enum Unit {
   Unit;
}

class AsyncTask<A> extends android.os.AsyncTask<A,Unit,Unit> {
   private var task:Array<A> -> Void;

   public function new(tsk:Void -> Void) {
      super();
      task = function(x) {
         tsk();
      };
   }
}

[waneck]

Can you provide an example with no dependencies?

[singpolyma]

Sure:

class Main<A> extends java.lang.Object {
   private var task:Array<A> -> Void;
   public function new(tsk:Void -> Void) {
      super();
      task = function(x) {
         tsk();
      };
   }
}

[Simn]

I guess that's an issue on all WrapRef targets?