Part #2:
Reza Rahman wrote:
> Marina,
>
> My input:
>
> # XXX marked items:
> <...>
> * We can pick either EJBException or NoSuchEJBException for bad
> singletons. I guess NoSuchEJBException would be more consistent with
> stateless and stateful (page 98).
It actually says javax.ejb.NoSuchEJBException in the referenced
chapters. I clarified. Thanks.
> * Transactions should be completed before the last AroundInvoke
> interceptor method completes (page 157).
Noted.
> * I don't understand the reasons for the restrictions on what
> singleton callback can be transactional (I think i should be enabled
> for both super-classes and interceptors). I also don't see why XML DDs
> values should not be applied to them (pages 166 and 169).
> * The module name stuff seems pretty clear to me (page 256).
Removed the XXX note.
> * Not sure what the issue with ejb-ref/mapped name is. Any decision we
> make is fine. In practice I think most people will use global JNDI
> names now anyways (page 259).
> * The dayOfMonth syntax is pretty clear to me. We don't go higher than
> 5th.
I added a note about the 5th to the expression rules.
> The expression rules also seem pretty clear, but I guess we cam always
> add more text (pages 321 and 325).
How about this:
"If the dayOfMonth attribute has a non-wildcard value and the dayOfWeek
attribute has a non-wildcard value, then the timer expires when the
current day matches either the dayOfMonth attribute or the dayOfWeek
attribute (i.e. the current day does not need to match of both attributes).