Re: [eclipselink-users] join fetch fails with "DotNode cannot be cast to

[Tom Ware <tom.ware@xxxxxxxxxx>]

I read single_valued_embeddable_object_field as a field representing an embeddable.

-Tom

Cronemberger, Constantino wrote:
> Hi Tom,
>
> Thanks for your replies.
>
> In the spec (persistence-2_0-final-spec.pdf) we have:
>
> fetch_join ::= [ LEFT [OUTER] | INNER ] JOIN FETCH join_association_path_expression
> join_association_path_expression ::= join_collection_valued_path_expression | join_single_valued_path_expression
> join_single_valued_path_expression ::= identification_variable.{single_valued_embeddable_object_field.}*single_valued_object_field
>
> The funny thing is that I did not find the definition of "single_valued_embeddable_object_field" in the spec.
> Regards,
>    Constantino.
>
> ________________________________
>
> ------------------------------
>
> Message: 2
> Date: Mon, 06 Dec 2010 10:30:18 -0500
> From: Tom Ware <tom.ware@xxxxxxxxxx>
> To: EclipseLink User Discussions <eclipselink-users@xxxxxxxxxxx>
> Subject: Re: [eclipselink-users] join fetch fails with "DotNode cannot
>         be cast to VariableNode"
> Message-ID: <4CFD018A.40409@xxxxxxxxxx>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
> Hi Constantino,
>
>    My reading of the JPA specification is that a multi-level fetch is not
> allowed in JPQL.
>
>    I believe the bug here is that we are not throwing an exception that tells
> you this.
>
>    Our fetch join hint was added to cover this case.
>
> -Tom
>
> Cronemberger, Constantino wrote:
> > Hi,
> >
> > I have these mappings:
> >
> > In TestClassA:
> >
> >     @OneToOne(fetch = FetchType.LAZY)
> >     public TestClassB getObjectBLocalFk() {
> >         return objectBLocalFk;
> >     }
> >
> >
> >
> > In TestClassB:
> >
> >
> >
> >     @OneToOne(fetch = FetchType.LAZY)
> >     @JoinColumn(name="objectALocal_Fk")
> >     public TestClassA getObjectALocalFk() {
> >         return objectALocalFk;
> >     }
> >
> > In this example both relationships are one-to-one but with other types it also happens.
> >
> >
> >
> > I have tested the suggested workaround and it worked.
> >
> >
> >
> > Thanks,
> >
> >     Constantino.
> _______________________________________________
> eclipselink-users mailing list
> eclipselink-users@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/eclipselink-users