Hi, David :
Seems that we could do the samething for AroundTimeout, if no one has
begun the work, could you assign the JIRA OpenEJB-1172 to me ? I could like
to work on it.
Thanks !
2010/6/9 Ivan <[email protected]>
>
>
> 2010/6/9 David Blevins <[email protected]>
>
> On Jun 7, 2010, at 4:40 PM, Ivan wrote:
>>
>> > Thanks for the comment, I attach another patch to the JIRA, if any
>> problem,
>> > please let me know :-)
>>
>> Wow. Wonderful patch, Ivan! Doesn't really get any better than that :)
>> Very thorough test case as well.
>>
>> I closed these JIRAs and marked them completed by you:
>>
>> OPENEJB-1147 javax.ejb.AfterBegin
>> OPENEJB-1148 javax.ejb.AfterCompletion
>> OPENEJB-1149 javax.ejb.BeforeCompletion
>> OPENEJB-1215 <after-begin-method> element
>> OPENEJB-1216 <before-completion-method> element
>> OPENEJB-1217 <after-completion-method> element
>> OPENEJB-1218 @AfterBegin overriding via <after-begin-method>
>> OPENEJB-1219 @AfterCompletion overriding via <after-completion-method>
>> OPENEJB-1220 @BeforeCompletion overriding via <before-completion-method>
>>
>>
>> A few more fun things we can tinker with:
>>
>> We can probably get rid of AppModulePreProcessor. This chunk of code can
>> go right into the AnnotationDeployer next to the related annotation:
>>
>> if (SessionSynchronization.class.isAssignableFrom(clazz)) {
>> sessionBean.getAfterBegin().add(new
>> LifecycleCallback(clazz.getName(), "afterBegin"));
>> sessionBean.getBeforeCompletion().add(new
>> LifecycleCallback(clazz.getName(), "beforeCompletion"));
>> sessionBean.getAfterCompletion().add(new
>> LifecycleCallback(clazz.getName(), "afterCompletion"));
>> } else {
>>
>> And this little chunk we can cleverly move into the SessionBean jaxb
>> class:
>>
>> if (sessionBean.getAfterBeginMethod() != null) {
>> sessionBean.getAfterBegin().add(new
>> LifecycleCallback(clazz.getName(),
>> sessionBean.getAfterBeginMethod().getMethodName()));
>> }
>> if (sessionBean.getBeforeCompletionMethod() != null) {
>> sessionBean.getBeforeCompletion().add(new
>> LifecycleCallback(clazz.getName(),
>> sessionBean.getBeforeCompletionMethod().getMethodName()));
>> }
>> if (sessionBean.getAfterCompletionMethod() != null) {
>> sessionBean.getAfterCompletion().add(new
>> LifecycleCallback(clazz.getName(),
>> sessionBean.getAfterCompletionMethod().getMethodName()));
>> }
>>
>> Take a look at the sneaky little trick we use in the MessageDrivenBean
>> jaxb class for handling the deprecated message-selector and
>> acknowledge-mode. :)
>>
>
> Actually that is what I did in the beginning, but after thinking it for a
> while, it might not work.
> a. AnnotationDeployer will not be invoked while META-COMPLETE is with
> value true, but we still need to configure those three methods, that is why
> I add the AppModulePreProssor class.
> b. Currently, we treat the session synchronization methods the same way as
> other lifecycle methods, but it has a little different, as in the past, we
> usually implment an interface if we intertested in those callbacks. So I
> would think that the bean class implment the SessionSynchronization
> interface should have the high priority IMO.
>
>
>
>>
>> On the validation front, let's expand the CheckInvalidCallbacksTest test
>> to cover the new checks we do. That test could really be expanded in
>> general, so feel free to have as much fun with it as you want. You can
>> attach that one to this JIRA:
>>
>> https://issues.apache.org/jira/browse/OPENEJB-1295
>
>
> Will check it soon !
>
>
>>
>>
>>
>> On our getParameters() TODO looks like we can expand that if block to
>> check if the operation is AFTER_COMPLETION and not throw the
>> IllegalArgumentException.
>>
>
> Just not sure whether it will break current spec definition, will add it.
>
>>
>>
>> Thanks so much for such a great patch!
>>
>>
>> -David
>>
>>
>
>
> --
> Ivan
>
--
Ivan
