Re: [hibernate-dev] Does <one-to-one ... constrained="false"> make sense for unidirectional association?
If <one-to-one ... constrained="false"/> does make sense for a
unidirectional association, and the foreign key really should be
FROM_PARENT, then, AFAICT, the only way the unit tests could pass is if
Hibernate assumes such associations would have an implicit
ignoreNotFound="true".
As an example, take a look at OptionalOneToOneTest [1] with the mapping
[2]. The reason why OptionalOneToOneTest passes is because, currently,
OneToOneType#isNullable()
returns foreignKeyType==ForeignKeyDirection.TO_PARENT. For this
association, OneToOneType#isNullable returns true because its foreign key
direction is (incorrectly?) ForeignKeyDirection.TO_PARENT.
[1]
https://github.com/hibernate/hibernate-orm/blob/master/hibernate-core/src...
[2]
https://github.com/hibernate/hibernate-orm/blob/master/hibernate-core/src...
On Mon, Oct 29, 2018 at 9:24 PM Gail Badner <gbadner(a)redhat.com> wrote:
The fix for HHH-12436 involves correcting the foreign key direction
for
some one-to-one assocations. In the process of making changes I see that we
have a unit test that has a unidirectional one-to-one association mapped
with constrained="false". [1]
ModelBinder assumes that <one-to-one ... constrained = "true"> means
that
the foreign key direction is FROM_PARENT; if constrained = "false", it
assumes that the foreign key direction is TO_PARENT. [2]
It seems odd to me that the foreign key direction for a unidirectional
association would be TO_PARENT.
Should the unit test have constrained = "true"?
If not, should ModelBinder check to see if the association is
bidirectional and determine which side owns the association? This would
probably require adding a SecondPass.
Please let me know your thoughts.
Regards,
Gail
[1]
https://github.com/hibernate/hibernate-orm/blob/master/hibernate-core/src...
[2]
https://github.com/hibernate/hibernate-orm/blob/master/hibernate-core/src...