Using query cache for this was not TheRightWay imho. HHH-2879 is about
rectifying that and doing it what I (and others) think is TheRightWay.
In other words, implementing HHH-2879 will mean removing the way this
stuff is currently stored in the query cache; it will need to get stored
in the L2 cache in some fashion (probably mapping natural-id to id
values).
-
Steve Ebersole
Project Lead
http://hibernate.org
steve(a)hibernate.org
Principal Software Engineer
JBoss, a division of Red Hat
http://jboss.com
http://redhat.com
steve.ebersole(a)jboss.com
steve.ebersole(a)redhat.com
On Wed, 2008-09-17 at 21:43 -0700, Elias Ross wrote:
On Wed, Sep 17, 2008 at 9:12 PM, Steve Ebersole
<steve(a)hibernate.org> wrote:
>
http://opensource.atlassian.com/projects/hibernate/browse/HHH-2879 ???
>
If you use Hibernate EntityManager, get the Hibernate Session and use
Criteria to query by natural ID and the entity was deleted, you end up
with an EntityNotFoundException. I reported the issue as: HHH-3478
I have a patch which fixes this problem, but it's a hack. From the issue:
In StandardQueryCache, there is a "catch" which handles
"UnresolvableObjectException" but not "EntityNotFoundException"
catch ( UnresolvableObjectException uoe ) {
if ( isNaturalKeyLookup ) {
//TODO: not really completely correct, since
// the uoe could occur while resolving
log.debug( "could not reassemble cached result set" );
cacheRegion.evict( key );
return null;
}
I'm thinking that by keeping an association mapped in the query cache
-- but I'm thinking that's not the right spot --
EntityManager/Session.remove() could trigger a removal of any natural
ID criteria query as well. And so could any update, if the natural IDs
were mutable. Anyway, you'd get rid of this hack, possibly.