NOTE: This may be related to #123054, which I just reopened.

Using NetBeans IDE Dev (Java SE Build 201110160600) on Windows XP using Java 1.6.0_25, I observe that Quick Fix suggests a cast which is technically correct but seems misleading because of an inner class.

1.  Create a new Java class in NetBeans IDE

2.  Define a member variable of type Rectangle2D.Double.

3.  Invoke a method which returns Rectangle2D.Double 

4.  Observe that you will be asked to cast it to Double. This implies java.lang.Double to me, though the editor/compiler do correctly recognize this as Rectangle2D.Double.  

Here is code to reproduce the problem per the above steps:

package com.tomwheeler.foo;

import java.awt.geom.AffineTransform;
import java.awt.geom.Rectangle2D;

public class AssignmentBehavior {

    private Rectangle2D.Double original;
    private Rectangle2D.Double rotated;

    public void demonstrate() { 
        AffineTransform twist = new AffineTransform();
        twist.rotate(0.03);
        
        // To see what I describe, put your cursor inside the 
        // method name and hit Alt+Enter; you will be asked to 
        // cast to "Double" -- most people looking at the code
        // would assume that's java.lang.Double.
        rotated = twist.createTransformedShape(original);
    }
}

I believe the code would be more clear if the cast suggested was Rectangle2D.Double.  I realize that qualifying the enclosing class name might not always (or even usually) be desirable, but it's a special case when the unqualified name conflicts with an implicit type from java.lang.  

Offhand, the only special cases I can think of are:

  * Arc2D.Double
  * Arc2D.Float
  * CubicCurve2D.Double
  * CubicCurve2D.Float
  * Ellipse2D.Double
  * Ellipse2D.Float
  * Line2D.Double
  * Line2D.Float
  * Path2D.Double
  * Path2D.Float
  * Point2D.Double
  * Point2D.Float
  * QuadCurve2D.Double
  * QuadCurve2D.Float
  * Rectangle2D.Double
  * Rectangle2D.Float
  * RoundRectangle2D.Double
  * RoundRectangle2D.Float