The rational behind the change is completely ok. I would be happy if you could also bring this to the EE6/7 board, because this is currently a real sick paradigma to only allow RuntimeExceptions everywhere (JPA lifecycle methods are another example) and to force wrapping the wrapper of an exceptionwrapper... until noone can see the real source of the problem anymore...
LieGrue, strub ----- Original Message ---- > From: David Blevins <[email protected]> > To: [email protected] > Sent: Thu, August 5, 2010 8:18:55 PM > Subject: Interceptor callback signature strictness > > This is a debatable chance, but the goal is to somehow make it possible for >integrators to be less strict with this particular spec rule. > > http://svn.apache.org/viewvc?view=revision&revision=982715 > > Basically, we opted not to strictly enforce this as the >invocationContext.proceed() method throws Exception so it forces all >interceptors to needlessly catch and convert that checked exception to a >runtime exception to meet the spec requirements: > > @PostConstruct > public void construct(InvocationContext context) { > try { > context.proceed(); > } catch (Exception e) { > throw new RuntimeException(e); > } > } > > The result is that if a callback exception is thrown by the bean, it will > end >up being wrapped and re-thrown N number of times. > > Not terrible, but it is a bit of bad API. > > Anyway, if anyone sees a better way to do this I'm open to solutions. I > just >went with simple and direct so it would be easily revertible and changeable >if >we wanted to do something different. > > > -David > >
