Re: [eclipselink-users] Dyn. Weaving, Lazy Collection, Clone Problem

[Oliver Vesper <oliver.vesper@xxxxxx>]

Hi,

here some additional information which hopefully help. First of all the 
eclipselink version which I am using - it's the current 2.0 RC1. And I 
debugged a bit more into the EclipseLink code and here are my observations:
When the cloning is performed an instance of PersistenceEntityCopyPolicy 
is being used. Its buildWorkingCopyClone(Object,Session) method is 
called which in turn calls the weaved _persistence_shallow_clone() 
method of my Bar entity. Unfortunately I couldn't debug into that 
method, but it's obvious that after the clone operation both objects 
(original and clone) reference the *same* PCS instance - which is wrong 
to my mind.

So hopefully somebody can help me out of this situation?! Maybe there is 
some configuration option to influence the clone operation.

Thanks in advance,
Oliver

Oliver Vesper schrieb:
> Hi,
>
> I have two entities as follows:
>
> @Entity
> public class Foo {
>   @Id
>   private Long id;
>
>   @OneToMany(fetch = FetchType.LAZY, cascade = CascadeType.ALL)
>   private final Collection<Bar> bars = new ArrayList<Bar>();
> }
>
> @Entity
> public class Bar {
>   @Transient
>   private transient final PropertyChangeSupport pcs = new 
> PropertyChangeSupport(this);
>
>   @Id
>   private Long id;
>
>   @Basic(fetch = FetchType.LAZY)
>   @Lob
>   private byte[] data;
> }
>
> When calling Foo.getBars() I receive an IndirectList. When accessing 
> elements of that list, the lazy object (here it's only 1 element) is 
> being loaded and instantiated. However I can see during debug that there 
> is some kind of "cloning" going on (registerExistingObject() is being 
> called). In detail the Bar instance is being "cloned" and afterwards 
> original and clone both point at the same PCS instance - PCS has a 
> back-reference to the original Bar instance. But when accessing the list 
> retrieved by getBars() I have access to the clone whose PCS is messed up.
>
> Any ideas what is going wrong? When weaving is turned off everything 
> works fine.
>
> I hope the problem is clear enough (it's 2 am over here).
>
> Thanks in advance,
> Oliver