The fix for 4514547 introduced extra COMPONENT_MOVED events for top-levels that weren't sent in previous JDKs:

* When dragging a Frame around the screen, we get two COMPONENT_MOVED events for every location that we put the Frame - one when we release the mouse at the new location (which is correct), and one when we start to drag the Frame to a new location (which is new). When this second event is set, technically the Frame has not moved anywhere.

* When a Frame is resized by the user, but the location has not changed (i.e. the upper-left corner hasn't moved) we get both a COMPONENT_RESIZED (which is correct) and a COMPONENT_MOVED event (which is new). Again, the Frame has not actually moved.

Don't get me wrong, 4514547 was a very important bug to fix. But if that problem could be solved without introducing these new events, that would be preferable.

I used the following test on Solaris 9 running CDE.

import java.awt.*;
import java.awt.event.*;

public class MovedTest extends Frame implements ComponentListener
{
    public MovedTest()
    {
        super( "MoveTest" );
        addComponentListener(this);
    }

    public void componentMoved(ComponentEvent e) {
        System.out.println("componentMoved: " + e);
    }
    public void componentResized(ComponentEvent e) {
        System.out.println("componentResized: " + e);
    }
    public void componentShown(ComponentEvent e) {}
    public void componentHidden(ComponentEvent e) {}


    public static void main( String[] args )
    {
        MovedTest t = new MovedTest();
        t.setSize(100,100);
        t.show();
    }
}