On Tue, Aug 16, 2011 at 6:06 PM, Laird
Nelson
<ljnelson@xxxxxxxxx>
wrote:
What is the expected behavior in the following case?
I have a mapped superclass annotated with
@Access(AccessType.FIELD).
It defines several fields and column mappings.
It features a private long id field, but it is entirely
unannotated:
private long id;
Then there is this:
@Access(AccessType.PROPERTY)
@Id
@GeneratedValue
@Column(name = "id")
public long getId() {
return this.id;
}
public void setId(final long id) {
this.id = id;
}
Next, I have an entity that extends this mapped superclass
which is as unannotated as possible.
I should point out here that section 2.3.2 of the spec says:
"This explicit access type specification does not affect the
access type of other entity classes or mapped superclasses in
the entity hierarchy."
Then it says:
"Persistent state inherited from superclasses is accessed in
accordance with the access types of those superclasses."
So that implies that my subclass should get the default access
type, which is FIELD.
In Hibernate (not EclipseLink) I'm seeing the subclass get
"switched over" to PROPERTY access. This seems like a
violation to me. EclipseLink and OpenJPA show no such
problems. I am at a loss to know what the proper behavior is
here.
Obviously it features an @Entity tag,
but most notably it does not declare an id field or override
(yet) the getId() method.
It's my understanding that only for the id property should the
getter/setter pair be used; every other field defined in
either the mapped superclass or its entity child should use
FIELD access.
...mainly because the subclass @Entity does not declare ANY
@Access annotations anywhere, and so should pick up the
default algorithm as described in section 2.3.1. Again, I
think EclipseLink gets this right, but I frankly would like
the EclipseLink developers to weigh in here, as I trust their
track record at interpreting the specification more than that
of the Hibernate crew. I've never seen a serious EclipseLink
specification violation, so I'm hoping that this is (yet
another) case where Hibernate gets it wrong, but I'm not sure.
Thanks. The related Hibernate "user forums" post--I use the
term guardedly--is over here:
https://forum.hibernate.org/viewtopic.php?f=1&t=1012254
Best,
Laird
