Re: [JBoss-dev] TransactionRolledBackException thrown incorrectly

[Adrian B. <adr...@jb...>]

Yes, but as I said the exceptions are specializations (subclasses)
of the required types, and more informative.

I would never recommend coding an EJB that assumes its own transaction
demarcation or that of its callees. 
This is a deployment time decision, not a development
time decision, see Table 15
"Handling of Exceptions Thrown by a Business Method of a Bean with
Container-Managed Transaction Demarcation"

On Wed, 2005-05-25 at 19:01, Bill Burke wrote:
> Adrian Brock wrote:
> > Why are you reading the draft? 18.2.2 in ejb-2.1-fr-spec.pdf
> > refers to bean providers responsiblities.
> > 
> 
> I meant ejb-2.1-fr-spec.pdf.  This is I was reading.
> 
> > TransactionRolledBack IS a RemoteException and
> > TransactionRolledBackLocal IS an EJBException (RuntimeException)
> > 
> 
> Again, these exceptions are only mentioned in section 18.3.6 which 
> states the rules for exceptions thrown in callback methods, NOT EJB METHODS!
> 
> > In no circusmtance should an EJB throw a RuntimeException to the client,
> > it *should* be wrapped in either a RemoteException or EJBException.
> > Although this is more a philosophy thing in the EJB spec than having
> > any real reason.
> > 
> 
> It SHOULD, but what we do is wrap ANY System exception thrown anywhere 
> in TransactionRolledbackException if the tx was initiated by the caller. 
>   It seems what we should be doing is propagating the REmoteException 
> thrown, or stuffing the RuntimeException into an EJBException.
> 
> > Show me the TCK test that causes our implementation (which is more
> > informative) to fail? And explain the problem it causes since both
> > exceptions have a mechanism to reach the wrapped exception
> > (something that was standardize in j2se1.4).
> > 
> 
> The problem is that users have to dive into the cause/exception detail 
> when they should just be able to catch the RemoteException or extension 
> of EJBException.  To avoid try/catch/setRollbackOnly blocks, but still 
> be able to catch the exception and do something about it after TX rollsback.
> 
> Bill
> 
> 
> -------------------------------------------------------
> SF.Net email is sponsored by: GoToMeeting - the easiest way to collaborate
> online with coworkers and clients while avoiding the high cost of travel and
> communications. There is no equipment to buy and you can meet as often as
> you want. Try it free.http://ads.osdn.com/?ad_id=7402&alloc_id=16135&op=click
> _______________________________________________
> JBoss-Development mailing list
> JBo...@li...
> https://lists.sourceforge.net/lists/listinfo/jboss-development
-- 
xxxxxxxxxxxxxxxxxxxxxxxx 
Adrian Brock
Chief Scientist
JBoss Inc.
xxxxxxxxxxxxxxxxxxxxxxxx