Hi Prasanta,
This issue does not occur with getComponentBefore().
The components can be traversed consistently in correct
sequence using getComponentBefore().
Also have verified the JCK tests related to focus traversal
system.
There are no failures due to this change.
Thanks,
Ambarish
From: prasanta sadhukhan
Sent: Wednesday, February 10, 2016 2:28 PM
To: Ambarish Rapte; Semyon Sadetsky; awt-dev@openjdk.java.net
Subject: Re: Review Request for 8025001: setFocusTraversalPolicy() to
ContainerOrderFocusTraversalPolicy results in an infinite loop
Looks ok to me logically. Also, don't we need to do the same for
getComponentBefore() to check for same component? Did you run JCK
java/awt/containerorderfocustraversalpolicy tests just to be safe?
Regards
Prasanta
On 2/1/2016 7:18 PM, Ambarish Rapte wrote:
Hi All,
Please review the fix for jdk9,
Bug: https://bugs.openjdk.java.net/browse/JDK-8025001
Webrev: http://cr.openjdk.java.net/~arapte/8025001/webrev.00/
Issue:
ContainerOrderFocusTraversalPolicy.getComponentAfter() returns
the same component on each call,
which results in an infinite loop in the test provided in bug.
Cause:
By default, ContainerOrderFocusTraversalPolicy implicitly transfers focus
down-cycle.
That is, during normal forward focus traversal, the Component traversed after a
focus-cycle-root
will be the focus-cycle-root's default Component to focus.
For a scenarios for above behavior, i.e.
When focus-cycle-root & focus-cycle-root's default Component are the same
component in traversal cycle.
If getComponentAfter() is called with focus-cycle-root, it returns
focus-cycle-root itself.
Fix:
Added a check to verify if the getComponentDownCycle() returns
same component as the one passed to it.
Then do not return this component and continue traversing.
Verification:
Verified that FocusTraversal tests PASS.
DefaultFocusTraversalPolicy is directly inherited from
ContainerOrderFocusTraversalPolicy.
DefaultFocusTraversalPolicy also behaves well with above change.
Thanks,
Ambarish
