Hi Dain,I haven't looked in detail at the life cycle of CMP beans in a couple of years, but in general you can't simply keep the state of the underlying Entities through the life cycle. CMP beans are pooled and reused in transaction contexts and you have to load the state at specific points in the life cycle.
Using the primary key stored in the CMP bean to do em.find at the appropriate time is the obvious way to take advantage of the em second level cache. To the extent that this is not efficient, we should fix it in the JPA layer not the CMP layer.
Craig On Apr 8, 2007, at 1:15 PM, Dain Sundstrom wrote:
Thanks for the great info Patrick.For now, I'm going to leave my implementation the way it is, without instance reuse, and wait until someone complains loudly.BTW in CMP there are a several places in the lifecycle of entities that people like to reuse beans...1) Beans are created and the ejb context is set into them 2) A primary key is assigned 3) Data is loaded into the beanPeople like to reuse instances at all three levels. The solution you suggest below addresses the first. Would it be possible to hook the cache to give the OpenJPA system an entity with data already loaded?Anyway, for not I'm just going to wait and see if anyone deeply cares about this feature.Thanks for help, -dain On Apr 8, 2007, at 8:56 AM, Patrick Linskey wrote:Hi, The easy way to do this is with extended persistence contexts. In anextended persistence context, your working set of objects lasts for theduration of the entity manager, not the transaction.In a JTA environment, the container is supposed to set things up so that EMs are transactional by default. To the best of my knowledge, between this rule and the persistence context propagation rule, this means thatthe container creates an EM proxy that gets injected / looked up, and that internally delegates through to the "right" EM internally. (I'm assuming that you're doing this for your EJB2 container work.)So, going with an extended PC is probably not sufficient for you, if you want to be spec-compliant. One thing that you could do, though, is make your EM proxy smart about holding onto a single EM for a long amount of time for use in transactions, but only use that EM during transactions.This would be closer to spec-compliant behavior, but wouldn't be 100% there.The 100% solution would be to change OpenJPA to delegate to a pluggablefactory for creation of new instances. This would require creating aninterface with the two pcNewInstance() signatures in PersistenceCapable,and changing the nine places in code that use that method directly to call the factory instead.Then, you could plug in whatever instance pooling logic you prefer, and you'd be in good shape. You'd need to figure out some way to know when to re-pool the instances, but in a container environment, that shouldn'tbe that hard. Note that you wouldn't want to use anSynchronization.afterCompletion() callback, since if it's detached, the instance might be used after completion in a rendering step or something of the sort, or it might be tucked away in session state for use lateron.Of course, if you've got a pluggable instance factory, you'd have someother alternates as well, such as creating clones or something of the sort. Out of curiosity, what is slow about creating your instances? -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: Dain Sundstrom [mailto:[EMAIL PROTECTED] Sent: Sunday, April 08, 2007 12:18 AM To: open-jpa-dev@incubator.apache.org Subject: Can I reuse instances? Is it possible to reuse instances from transaction to transaction? I would like to be able to create a bean in one transaction, detach it and reattach the same instance in a new transaction. My goal here is specifically to reuse instances across transactions because they have a very expensive creation cost, and no I can not redesign the system. I tried a quick test case using merge, but merge returned a new instance: beginTx(); Employee dain = entityManager.find(Employee.class, dainPk); assertEquals(dain.getFirstName(), "Dain"); commitTx(); beginTx(); assertSame(dain, entityManager.merge(dain)); assertEquals(david.getFirstName(), "Dain"); commitTx(); When I try to use the refresh method, OpenJPA complains that the entity "is not managed by this context". So is there anyway for me to reuse instances? -dainNotice: 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.
Craig Russell Architect, Sun Java Enterprise System http://java.sun.com/products/jdo 408 276-5638 mailto:[EMAIL PROTECTED] P.S. A good JDO? O, Gasp!
smime.p7s
Description: S/MIME cryptographic signature