On Feb 6, 2007, at 11:27 AM, Patrick Linskey wrote:
Out-of-band, Abe pointed out to me that the text about when RollbackExceptions are thrown is pretty clear. 3.7 says: "The RollbackException is thrown by the persistence provider when EntityTransaction.commit fails." So, it would seem that in the case where an EM.find() or some other non-commit() call triggers a callback that throws an exception, we should not wrap the exception in a RollbackException. Should we just throw the raw exception, or should we wrap in some other PersistenceException?
I think if a user throws an exception in a callback outside of a commit operation, we should simply rethrow it to the user after we clean up our internal state. Presumably, the specific runtime exception has meaning to the user's code, and we don't add much value in wrapping the exception.
I might think differently if the spec allowed/required us to continue after catching the first exception thrown from a callback, but it explicitly says that the first exception aborts the operation. From 3.5.6, "No further lifecycle callback methods will be invoked after a runtime exception is thrown."
From that same paragraph, "A runtime exception thrown by a callback method that executes within a transaction causes that transaction to be rolled back." But I think it's ok to set RollbackOnly to satisfy this requirement. And as a matter of design (separation of concerns) this really is best practice.
We should ask that the spec be clarified so applications can be more portable.
Craig
-Patrick -- Patrick Linskey BEA Systems, Inc.______________________________________________________________________ _ Notice: This email message, together with any attachments, may contain information of BEA Systems, Inc., its subsidiaries and affiliated entities, that may be confidential, proprietary, copyrighted and/or legally privileged, and is intended solely for the use of the individual or entity named in this message. If you are not the intended recipient, and have received this message in error, please immediately return thisby email and then delete it.-----Original Message----- From: Patrick Linskey [mailto:[EMAIL PROTECTED] Sent: Monday, February 05, 2007 7:14 PM To: open-jpa-dev@incubator.apache.org Subject: RE: Exceptions thrown from callbacksI don't recall the details about why we made thatdecision, but Iprefer wrapping in a RollbackException for consistency. It sucksto haveto do:I agree that that sucks. Of course, if they want their application to be portable, then they'll need to put in that logic anyway :)Agreed, and we may need to change our behavior later once the spec team clears this up. I'll be lobbying for the spec team to move to the always-wrapped model, though.Do you have any opinion about the rollback vs. mark-for-rollback distinction?Well, section 3.5 suggests that it should be rolled back immediately, but then if we start wrapping in a PersistenceException, then 3.7 says that it should just be marked for rollback. Tricky. I guess I favor immediate rollback, just since section 3.5 is so explicit in saying that it should happen.I think that rolling back directly is pretty evil, and as I recall, the spec team agreed at one point. I think that this one just slipped through the cracks. In any event, this is covered in the area of the CTS that was excluded, for just this reason presumably. -- Patrick Linskey BEA Systems, Inc. ______________________________________________________________ _________ Notice: This email message, together with any attachments, may contain information of BEA Systems, Inc., its subsidiaries and affiliated entities, that may be confidential, proprietary, copyrighted and/or legally privileged, and is intended solely for the use of the individual or entity named in this message. If you are not the intended recipient, and have received this message in error, please immediately return this by email and then delete it.-----Original Message----- From: Marc Prud'hommeaux [mailto:[EMAIL PROTECTED] On Behalf Of Marc Prud'hommeaux Sent: Monday, February 05, 2007 6:49 PM To: open-jpa-dev@incubator.apache.org Subject: Re: Exceptions thrown from callbacksI don't recall the details about why we made thatdecision, but Iprefer wrapping in a RollbackException for consistency. It sucksto haveto do:I agree that that sucks. Of course, if they want their application to be portable, then they'll need to put in that logic anyway :)Do you have any opinion about the rollback vs. mark-for-rollback distinction?Well, section 3.5 suggests that it should be rolled back immediately, but then if we start wrapping in a PersistenceException, then 3.7 says that it should just be marked for rollback. Tricky. I guess I favor immediate rollback, just since section 3.5 is so explicit in saying that it should happen. On Feb 5, 2007, at 6:27 PM, Patrick Linskey wrote:As for whether to wrap the exceptions duringnon-flush/commit times(e.g., during a find() call), I'm a little less keen onit, but notreally opposed. The reason is that the clause "Lifecycle callback methods may throw runtime exceptions" suggests to me (and, apparently, to the CTS authors) that they restricted theexceptiontype to be runtime exceptions because they expect the unmodified exception will be thrown straight up the application. Summary: +0I don't recall the details about why we made thatdecision, but Iprefer wrapping in a RollbackException for consistency. It sucksto haveto do: try { em.find(...); em.commit(); } catch (RollbackException re) { if (re.getCause() instanceof MySpecialException) { // custom logic } else { throw re; } } catch (MySpecialException mse) { // same custom logic as above } Do you have any opinion about the rollback vs. mark-for-rollback distinction? -Patrick -- Patrick Linskey BEA Systems, Inc.______________________________________________________________ _________ Notice: This email message, together with any attachments, may contain information of BEA Systems, Inc., its subsidiaries and affiliated entities, that may be confidential, proprietary, copyrighted and/or legally privileged, and is intended solely for the use of the individual or entity named in this message. If you are not the intended recipient, and have received this message in error, pleaseimmediately returnthis by email and then delete it.-----Original Message----- From: Marc Prud'hommeaux [mailto:[EMAIL PROTECTED] On Behalf Of Marc Prud'hommeaux Sent: Monday, February 05, 2007 6:20 PM To: open-jpa-dev@incubator.apache.org Subject: Re: Exceptions thrown from callbacks I don't have any strong opinions either way. Wrapping is useful, because we can usually provide the FailedObject so thatthe user canmore easily attempt some recovery. And I do agree that ifa callbackexception occurs during a commit()/flush() operationthen we shouldwrap it (and it might as well be in a RollbackException).Summary: +1As for whether to wrap the exceptions duringnon-flush/commit times(e.g., during a find() call), I'm a little less keen onit, but notreally opposed. The reason is that the clause "Lifecycle callback methods may throw runtime exceptions" suggests to me (and, apparently, to the CTS authors) that they restricted theexceptiontype to be runtime exceptions because they expect the unmodified exception will be thrown straight up the application. Summary: +0 On Feb 1, 2007, at 3:27 PM, Patrick Linskey wrote:Hi, There's a bit of ambiguity in the spec about what should happen when an exception is thrown from a callback. I propose that we change OpenJPA's behavior to always wrapexceptionsthrown from callbacks in a RollbackException. Additionally,I proposethat if a callback is thrown from a direct flush() call ora find() orother data load, we should mark the transaction for rollback instead of immediately rolling back the transaction. Details: Section 3.5 says: "Lifecycle callback methods may throw unchecked/runtimeexceptions. Aruntime exception thrown by a callback method thatexecutes within atransaction causes that transaction to be rolled back." 3.5.6: "Lifecycle callback methods may throw runtime exceptions.A runtimeexception thrown by a callback method that executes within a transaction causes that transaction to be rolled back. No further lifecycle callback methods will be invoked after a runtime exception is thrown." 3.7: "The PersistenceException is thrown by the persistenceprovider when aproblem occurs. [...] All instances of PersistenceExceptionexcept forinstances of NoResultException and NonUniqueResultExceptionwill causethe current transaction, if one is active, to be marked forrollback."... "The RollbackException is thrown by the persistenceprovider whenEntityTransaction.commit() fails. So.... in my opinion, this means that if a callback fails during commit(), the exception thrown by the callback clearly should be wrapped in a RollbackException. It is less clear to me what shouldhappen if acallback fails at some other time, such as during a find()call. In myopinion, we should be wrapping the user-thrown exceptions in RollbackExceptions all the time. Further, I think that 3.7 trumps 3.5.6, so if anexception is thrownfrom a callback during a find(), we should be marking thetransactionfor rollback, rather than actually rolling it back. I've got changes in place that implement the behavior I just described. The CTS tests surrounding this issue have been excluded,due to theambiguity in the spec. Thoughts? -Patrick -- Patrick Linskey BEA Systems, Inc.______________________________________________________________ _________ Notice: This email message, together with any attachments, may contain information of BEA Systems, Inc., its subsidiaries and affiliated entities, that may be confidential, proprietary, copyrighted and/or legally privileged, and is intended solely for the use of the individual or entity named in this message. If you are not the intended recipient, and have received this message in error, pleaseimmediately returnthis by email and then delete it.
Craig Russell DB PMC [EMAIL PROTECTED] http://db.apache.org/jdo
smime.p7s
Description: S/MIME cryptographic signature
