On the ###@###.### mailing list the following was reported.

Hi, Jemmy Folks -
We run a set of tests (> 700) that use Jemmy to exercise a
three-tiered client UI. We are running JDK 1.4.2_03 and Jemmy 2.2.1,
and intermittently, but consistently get this stack trace:

java.lang.NullPointerException
         at sun.awt.windows.WInputMethod.dispatchEvent(WInputMethod.java:253)
         at sun.awt.im.InputContext.dispatchEvent(InputContext.java:238)
         at sun.awt.im.InputMethodContext.dispatchEvent(InputMethodContext.java:180)
         at java.awt.Component.dispatchEventImpl(Component.java:3565)
         at java.awt.Container.dispatchEventImpl(Container.java:1627)
         at java.awt.Component.dispatchEvent(Component.java:3477)
         at java.awt.EventQueue.dispatchEvent(EventQueue.java:456)
         at java.awt.EventDispatchThread.pumpOneEventForHierarchy(EventDispatchThread.java:201)
         at java.awt.EventDispatchThread.pumpEventsForHierarchy(EventDispatchThread.java:151)
         at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:145)
         at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:137)
         at java.awt.EventDispatchThread.run(EventDispatchThread.java:100)

It occurs in a different thread from the main testing thread (note no
code of our own or JUnit's). I would guess that we see, depending on
the machine and user, the exception every ~1000 test runs. So not
every time through the test suite, but enough to be a real
problem. I'm trying to find a way to reproduce it consistently with
the test runs.

As far as I (and other engineers) know we have *never* seen it in all
the extensive manual testing and use of the client application and
framework, so we suspect that it has something to do with Jemmy. I
tried upgrading to Jemmy 2.2.4.9, but still saw it. Since we run
exclusively in "queue dispatching mode", I made some attempts at
running in Robot mode, but I'm not sure I actually succeeded in
kicking into in Robot mode (I was using
"-Djemmy.robot_dispatching=on") so that experiment is
inconclusive. I'm going to try it again.

But now it's time to appeal to the community at large - have any of you
seen this beastie? If so, were you able to do anything about it?

Thanks -

John Major
--------------------------------
Senior Software Architect
Cimarron Software, Inc.
175 S. West Temple
Suite 530
Salt Lake City, UT 84101
###@###.###
801 736 627


---------------------------------------------------------------------
To unsubscribe, e-mail: ###@###.###
For additional commands, e-mail: ###@###.###