Re: [eclipselink-users] Looking for hints to map inheritance relationshi

[Phillip Ross <phillip.w.g.ross@xxxxxxxxx>]

I guess I should also point out... the Automobile class has a
setCategory method which sets this.category but the line below has a
super.category assignment which is commented out.  When this is
uncommented, the creation of the entities works... but it seems like
this should not be necessary since the category property of the
Automobile class is mapped and referencing AutomobileCategory, not
ProductCategory.

Furthermore, if the super.category assignment is uncommented you can
see that querying all ProductCategory instances will return two... of
which one is a concrete Automobile category, but is contains no
Automobiles in it's collection.  Ideally it should... since querying
for all AutomobileCategories returns the same automobile category
instance which DOES have the collection populated with Automobile
instances.

I'm wondering if I'm hitting a bug or maybe just getting too fancy
with respect to mapping inheritance with JPA :)

On Fri, Jul 15, 2011 at 6:09 AM, Phillip Ross
<phillip.w.g.ross@xxxxxxxxx> wrote:
> Thanks for the reply Tom.  Here are the annotated property definitions
> from the classes:
>
> Product:
>    @JoinColumn(nullable=false)
>    @ManyToOne(cascade={CascadeType.PERSIST,CascadeType.MERGE,CascadeType.REFRESH})
>    protected ProductCategory category;
>
> ProductCategory:
>    @OneToMany(mappedBy = "category",
> cascade={CascadeType.PERSIST,CascadeType.MERGE,CascadeType.REFRESH})
>    private Set<Product> products;
>
> Automobile:
>    @JoinColumn(nullable=false)
>    @ManyToOne(cascade={CascadeType.PERSIST,CascadeType.MERGE,CascadeType.REFRESH})
>    private AutomobileCategory category;
>
> AutomobileCategory:
>    @OneToMany(mappedBy = "category",
> cascade={CascadeType.PERSIST,CascadeType.MERGE,CascadeType.REFRESH})
>    private Set<Automobile> automobiles;
>
>
> The following is the code that just persists some instances of the entities:
>        entityManager.getTransaction().begin();
>        AutomobileCategory automobileCategory = new
> AutomobileCategory("AC_" + UUID.randomUUID().toString(), "ENAME",
> "SNAME");
>        for (int i = 0; i < 5; i++) {
>            Automobile automobile = new Automobile("A_" +
> UUID.randomUUID().toString(), automobileCategory, "ENAME", "SNAME");
>            entityManager.persist(automobile);
>        }
>        entityManager.getTransaction().commit();
>
> And I'm attaching the log as it's probably too big to inline.
>
> I have a full testcase that probably illustrates more of what is going
> on.  I'll attach those source files too.
>
> Thanks again
> - Phillip
>
>
> On Thu, Jul 14, 2011 at 8:39 AM, Tom Ware <tom.ware@xxxxxxxxxx> wrote:
>> Hi Phillip,
>>
>>  Can you please post the following:
>>
>> - examples of how the category fields are mapped in both Product and
>> Automobile
>> - a code snippet the shows what you are doing
>> - logging of the SQL produced and the full exception trace (for logging, use
>> persistence unit property eclipselink.logging.level=FINEST)
>>
>> -Tom
>>
>> Phillip Ross wrote:
>>>
>>> Hi all, I'm having a problem with mapping inheritance and not sure if
>>> things are acting correctly or not.  Was hoping for some hints.
>>>
>>> The scenario I'm working with is that I have two related entity
>>> superclasses mapped with an inheritance strategy or JOINED, and there
>>> is a ManyToOne relationship between the two.  The subclass entities
>>> subclass the superclasses and override the relationship (or at least
>>> this is what is intended).  I had first assumed that when the tables
>>> were created, a field would be created in both the parents table and
>>> child table... but it appears that it was only created in the parent
>>> table.  I figured that this was OK since there would be records in the
>>> parent tables that would map related subclass instances, but when I
>>> attempt to persist the subclass with the ManyToOne member attribute...
>>> an exception is thrown caused by a not-null constraint violation
>>> corresponding to the parent table's FK field.  As an experiment, I
>>> changed the child class to populate the related property field in both
>>> the superclass and itself, but I'm assuming this won't really work out
>>> in the future as fetching the correct records may not work either.
>>>
>>> To put this in more concrete terms... say I have Product and
>>> ProductCategory as superclasses.  Product has a property called
>>> category which holds a ProductCategory instance.  Automobile
>>> subclasses Product and AutomobileCategory subclasses ProductCategory.
>>> The Automobile class defined a property named category and is defined
>>> to be AutomobileCategory.  Since persisting an Automobile and an
>>> AutomobileCategory would create records in all four tables... shouldnt
>>> eclipselink know that the category field in the Product table relating
>>> to ProductCategory be used to map the Automobile instances to the
>>> AutomobileCategory instances?
>>>
>>> I'm hoping someone knoww if I'm trying to map a scenario that's not
>>> supported or if there is a trick to doing this?  Perhaps an
>>> eclipselink specific trick that goes beyond the JPA spec?  If it is
>>> supported and sounds like a bug... I have a test case to submit :)
>>>
>>> Thanks!
>>> - Phillip
>>> _______________________________________________
>>> eclipselink-users mailing list
>>> eclipselink-users@xxxxxxxxxxx
>>> https://dev.eclipse.org/mailman/listinfo/eclipselink-users
>>
>> _______________________________________________
>> eclipselink-users mailing list
>> eclipselink-users@xxxxxxxxxxx
>> https://dev.eclipse.org/mailman/listinfo/eclipselink-users
>>
>