<AWT Dev> <Swing Dev> Request for review, JDK-8009883, REGRESSION: test/closed/javax/swing/AbstractButton/4246045/bug4246045.java fails
alexandr.scherbatiy at oracle.com
Fri May 16 11:45:24 UTC 2014
Is there any difference between AccessibleAWTFocusHandler and
On 5/7/2014 5:45 AM, Pete Brunet wrote:
> Hi Swing and AWT teams, (and Artem since you were involved in the
> 7179482 fix),
> I'd appreciate your review for a fix to a regression in
> java.awt.Component and javax.swing.JComponent.
> Original bug: https://bugs.openjdk.java.net/browse/JDK-7179482
> Orignal CCC: 7179482
> Regression Bug: https://bugs.openjdk.java.net/browse/JDK-8009883
> Regression Fix Webrev:
> Problem Description:
> jdk/test/closed/javax/swing/AbstractButton/4246045/bug4246045.java tests
> to make sure the focus state changes whenever a JButton or JToggleButton
> property change listener is called. The problem in this case is that
> for each change of focus the listener is called twice. The first time
> there is no problem because for example the component is not focused and
> then it is, but the second time the listener is called the component is
> already focused and the new state is also focused so the test case
> detects that the focus state didn't change.
> Root cause:
> There are two focus gained and two focus lost events fired when
> JComponents gain/loose focus. Part of the fix for JDK-7179482 was to
> deprecate the JComponent.AccessibleJComponent.accessibleFocusHandler
> field noting that
> Component.AccessibleAWTComponent.accessibleAWTFocusHandler should be
> used instead. The problem is that this was not done when 7179482 was
> fixed. The end result is that two focus listeners are added when
> JComponent.AccessibleJComponent.addPropertyChangeListener is called.
> The first listener is added in that property change method and the
> second one is added when the superclass method is called, i.e.
> Component.AccessibleAWTComponent.addPropertyChangeListener. The two
> listeners cause two focus events with the second one not causing a state
> Proposed fix:
> The proposed fix is for
> JComponent.AccessibleJComponent.addPropertyChangeListener to set its
> listener into Component.AccessibleAWTComponent.accessibleAWTFocusHandler
> so when the superclass method,
> Component.AccessibleAWTComponent.addPropertyChangeListener is called the
> listener at that level will already be set.
> A side effect of the fix is that logic was needed to be added in
> Component.AccessibleAWTComponent.addPropertyChangeListener to deal with
> the fact that the
> Component.AccessibleAWTComponent.accessibleAWTFocusHandler field can be
> set either from within the object's subclass method
> JComponent.AccessibleJComponent.addPropertyChangeListener or via an
> external call to Component.AccessAWTComponent.addPropertyChangeListener
> for cases other than when there is a JComponent subclass of Component.
> In the former case the listener is already set by the subclass so a
> focus listener isn't added. In the latter case the listener is not yet
> added so a focus listener is new'd and added.
> There are similar considerations when removing a listener; this can be
> seen in the webrev.
> There may be a better way to fix this but the proposed fix does resolve
> the issue that the test case exposed.
> Best regards,
More information about the awt-dev