Thanks Jason for your quick reply.

Unfortunately, Mr. Murphy put his dirty hands in my tests. I redefined the 
exception state variable in the PostConstruct method, thus the variable 
remained null. And my test exception has as root exception a 
NullPointerException. (I will never do that again!). So each invocation of the 
submitted test has only one exception with no other cause (thus no cause 
stack), a NullPointerException.

So the doc is correct, the invocations of the event handlers will bubble up the 
exception stack.

There is only one drawback. The invocation of the root exception will follow 
the class hierarchy and thus invoke the handlers  of the 'last resort' 
(Throwable, Exception) in the first turn before traversing the whole stack. 
Ideally they should be invoked only after the end of the turn. But maybe there 
is a simple solution for that I did not find out?

Again, my apologies
Regards, Rainer


Am 30.07.2014 15:01, schrieb Jason Porter:
You're correct, the docs are wrong, I'm sorry. handleAndContinue continues 
through the hierarchy chain. —
Sent from Mailbox

Reply via email to