[hibernate-dev] Entity name in the context of JPA2 meta model (HHH-5709).

Emmanuel Bernard emmanuel at hibernate.org
Fri Nov 5 15:45:06 EDT 2010 

Sorry, I had a off-line week, catching up here.

I think we all agree the expected behavior (according to the spec) is that EntityType.getName() should return @Entity.name.
I'd say for the future major rev, we should make the change.
I also think that the metamodel use at this stage is small enough that such change would only impact a few users. And there is a clear and easy API change for people impacted.

I personally would do the move as soon as possible but I can see / respect the decision if people want to wait post 3.6.x (in practice that would most likely be Hibernate 4).

On 3 nov. 2010, at 16:55, Steve Ebersole wrote:

> I completely empathize since the JPA "metamodel" defines no other means to get 
> at this information.  I am however nervous especially nervous about 
> backporting this fix as it is a change in behavior that is far from clearly a 
> bug.  I am kind of new to annotations so it is hard for me to say how/if users 
> use this EntityType.getName() today.  Emmanuel?  Hardy? 
> 
> 
> On Wednesday, November 03, 2010, at 10:34 am, Tomasz Blachowicz wrote:
>> It's up to you. I put my arguments in the ticket. I agree that jsr317 spec
>> does not specify the correlation of the @Entity.name and
>> EntityType#getName(), but I do think it if reasonable to think that entity
>> name returned by the meta model and the entity name specified in the
>> annotation is the same thing. Furthermore the two other implementations of
>> jsp317, Apache OpenJPA and EclipseLink, went this very route.
>> 
>> On Wed, Nov 3, 2010 at 2:24 PM, Steve Ebersole <steve at hibernate.org> wrote:
>>> We were just discussing this on IRC.  Yes I think adding a new attribute
>>> to PersistentClass to capture the value of @Entity.name specifically
>>> (really we'd
>>> just populate it with EntityBinder.name) is best.
>>> 
>>> So we know *how* to do it.  But to be honest, I am still uncertain
>>> whether we
>>> *should* do it.
>>> 
>>> HHH-2597 is completely irrelevant here.  We are not discussing the
>>> Hibernate
>>> Criteria API here.
>>> 
>>> HHH-4375 is a bad idea.  As stated before, @Entity.name is not really in
>>> any
>>> way associated with Hibernate's notion of an entity name.  @Entity.name
>>> is what Hibernate calls a query import name.
>>> 
>>> On Wednesday, November 03, 2010, at 08:54 am, Tomasz Blachowicz wrote:
>>>> I think I understand what you mean. "Node name" might not be indeed the
>>>> best way to do that.
>>>> 
>>>> I think implementation of the EntityBinder with regards to the handling
>>> 
>>> of
>>> 
>>>> @Entity.name is correct. My understanding of the jsr-317 is that the
>>> 
>>> entity
>>> 
>>>> name used in queries should be @Entity.name if specified otherwise the
>>>> default value that is unqualified name of the entity class. This is how
>>>> EntityBinder works now I presume
>>>> (EntityBinder#bindEjb3Annotation(@Entity)). The think is that this
>>>> value is set to the name field of EntityBinder and then used to set
>>>> PresistentClass#setNodeName().
>>>> 
>>>> I wonder how I could implement the meta model change to make sure the
>>>> EntityType#getName() returns @Entity.name or unqualified class name in
>>> 
>>> case
>>> 
>>>> the @Entity.name has not been specified. How about adding new property
>>>> to the PersistentClass that would hot the value as computed in
>>>> EntityBinder and the I could use the value of this property to
>>>> populate EntityTypeImpl from MetamodelImpl#buildEntityType? Can you
>>>> comment that approach?
>>>> 
>>>> On Wed, Nov 3, 2010 at 1:22 PM, Steve Ebersole <steve at hibernate.org>
>>> 
>>> wrote:
>>>>> Ah, I misread.  I thought you meant the Hibernate notion of an
>>>>> "entity name".
>>>>> As far as I can tell now looking at it, there is no such annotation
>>> 
>>> equiv
>>> 
>>>>> for
>>>>> that.
>>>>> 
>>>>> @javax.persistence.Entity.name is really more akin to Hibernate's
>>> 
>>> notion
>>> 
>>>>> of an
>>>>> import name for queries.
>>>>> 
>>>>> I am not really understanding the benefit of returning the query
>>>>> import name
>>>>> for EntityType.getName().  Personally I think its just a case of
>>>>> badly matched
>>>>> attribute names and an inference being drawn there.
>>>>> 
>>>>> But we do not really use that internally, so i have no qualms with
>>>>> changing it
>>>>> I guess.  However, I do not want "node name" used.  That is a totally
>>>>> different beast, and if annotations ever support alternate entity
>>>>> modes this
>>>>> will become a problem in this code you suggest; its currently working
>>>>> just by
>>>>> a quirk where the default node name just happens to be the
>>>>> unqualified entity
>>>>> name, and I am assuming the annotation binding code is setting it if
>>>>> @Entity.name is specified as well.  To be honest, after looking at
>>>>> org.hibernate.cfg.annotations.EntityBinder#bindEntity my first
>>>>> thought was whether this code even handles @Entity.name properly in
>>>>> terms of making it an
>>>>> import name.  It would work as is for the default case because that
>>>>> is Hibernate's internal default behavior (to register the
>>>>> unqualified
>>> 
>>> entity
>>> 
>>>>> name
>>>>> as an import name).
>>>>> 
>>>>> On Wednesday, November 03, 2010, at 07:58 am, Steve Ebersole wrote:
>>>>>> If that is really the case, then that is a problem with the
>>> 
>>> annotation
>>> 
>>>>>> binding code.
>>>>>> 
>>>>>> On Wednesday, November 03, 2010, at 06:50 am, Tomasz Blachowicz
>>> 
>>> wrote:
>>>>>>> Hi Steve,
>>>>>>> 
>>>>>>> PersistentClass#getEntityName is not good, because it is exactly
>>> 
>>> the
>>> 
>>>>> same
>>>>> 
>>>>>>> value as PersistentClass#getClassName.
>>>>>>> PersistentClass#getClassName is populated in
>>>>>>> org.hibernate.cfg.annotations.EntityBinder#bindEntity with the
>>> 
>>> value
>>> 
>>>>> from
>>> 
>>> org.hibernate.annotations.common.reflection.java.JavaXClass#getName.
>>> 
>>>>>>> JavaXClass#getName is implemented as clazz.getName(), so it is
>>> 
>>> always
>>> 
>>>>>>> class name and the @Entity.name is simply ignored for the purpose
>>> 
>>> of
>>> 
>>>>> the
>>>>> 
>>>>>>> entity name. And this is the reason I used
>>>>>>> PersistentClass#getNodeName that returns @Entity.name or
>>>>>>> "unqualified" class name in case @Entity.name is unspecified.
>>>>>>> 
>>>>>>> If I'd stick with PersistentClass#getEntityName the issues is not
>>>>>>> resolved and EntityType.getName() returns stil FQN of the class
>>>>>>> instead of @Entity.name or "unqualified" class name.
>>>>>>> 
>>>>>>> Regards,
>>>>>>> Tom
>>>>>>> 
>>>>>>> On Wed, Nov 3, 2010 at 11:12 AM, Steve Ebersole <
>>> 
>>> steve at hibernate.org>
>>> 
>>>>> wrote:
>>>>>>>> Uncertain yet of how these pull request comments work on
>>>>>>>> GitHub,
>>> 
>>> so
>>> 
>>>>>>>> I thought
>>>>>>>> I'll make sure and respond here.  The change should use
>>>>>>>> org.hibernate.mapping.PersistentClass#getEntityName instead of
>>>>>>>> org.hibernate.mapping.PersistentClass#getNodeName.
>>>>>>>> 
>>>>>>>> See my other email to the dev list about porting to 3.6
>>>>>>>> 
>>>>>>>> On Wednesday, November 03, 2010, at 05:48 am, Tomasz Blachowicz
>>>>> 
>>>>> wrote:
>>>>>>>>> I've submitted the pull request (
>>>>>>>>> https://github.com/hibernate/hibernate-core/pull/8) for
>>> 
>>> master. I
>>> 
>>>>>>>>> should have it also ready for 3.6 branch as soon as I manage
>>>>>>>>> to
>>>>> 
>>>>> merge
>>>>> 
>>>>>>>>> the change between branches in this git thng that is pretty
>>>>>>>>> new to
>>>>> 
>>>>> me
>>>>> 
>>>>>>>>> ;)
>>>>>>>>> 
>>>>>>>>> On Tue, Nov 2, 2010 at 1:18 PM, Emmanuel Bernard
>>>>>>>> 
>>>>>>>> <emmanuel at hibernate.org>wrote:
>>>>>>>>>> Cool :)
>>>>>>>>>> If you use GitHub's infrastructure and use a pull-request,
>>>>>>>>>> we can get that integrated quite fast.
>>>>>>>>>> 
>>>>>>>>>> On 2 nov. 2010, at 13:50, Tomasz Blachowicz wrote:
>>>>>>>>>> 
>>>>>>>>>> I've already created the ticket (
>>>>> 
>>>>> http://opensource.atlassian.com/projects/hibernate/browse/HHH-5709)
>>>>> 
>>>>>>>>>> . And I should be able to create the patch sometime this
>>>>>>>>>> evening.
>>>>>>>>>> 
>>>>>>>>>> Cheers,
>>>>>>>>>> Tom
>>>>>>>>>> 
>>>>>>>>>> On Tue, Nov 2, 2010 at 12:24 PM, Emmanuel Bernard
>>>>>>>> 
>>>>>>>> <emmanuel at hibernate.org>wrote:
>>>>>>>>>>> I'd say you're correct and the name should match in the
>>>>> 
>>>>> metamodel.
>>>>> 
>>>>>>>>>>> Can you open a JIRA issue and even better try and work out
>>>>>>>>>>> a patch?
>>>>>>>>>>> 
>>>>>>>>>>> PS: I've spilled hot chocolate on my laptop in the past:
>>>>>>>>>>> not
>>>>> 
>>>>> good,
>>>>> 
>>>>>>>>>>> the machine slowly but inevitably dies as corrosion wins
>>>>>>>>>>> it
>>>>> 
>>>>> over.
>>>>> 
>>>>>>>>>>> On 2 nov. 2010, at 12:35, Tomasz Blachowicz wrote:
>>>>>>>>>>>> Hi,
>>>>>>>>>>>> 
>>>>>>>>>>>> I had been working recently on some stuff that deals
>>>>>>>>>>>> with JPA2 meta
>>>>>>>>>>> 
>>>>>>>>>>> model.
>>>>>>>>>>> 
>>>>>>>>>>>> Hibernate is my JPA2 provider/engine. Everything works
>>> 
>>> fine
>>> 
>>>>>>>>>>>> except
>>>>>>>> 
>>>>>>>> one
>>>>>>>> 
>>>>>>>>>>>> little thing that is name of the Entity. I'd expect that
>>>>>>>>>>>> value returned
>>>>>>>>>>> 
>>>>>>>>>>> by
>>>>>>>>>>> 
>>>>>>>>>>>> EntityType.getName() would be the same as @Entity.name
>>>>>>>>>>>> but not the
>>>>>>>> 
>>>>>>>> FQN
>>>>>>>> 
>>>>>>>>>>> of
>>>>>>>>>>> 
>>>>>>>>>>>> the entity class. Although this is not explicitly stated
>>> 
>>> in
>>> 
>>>>> the
>>>>> 
>>>>>>>>>>>> JPA2 specification it is reasonable to expect that
>>>>> 
>>>>> @Entity.name
>>>>> 
>>>>>>>>>>>> or the
>>>>>>>>>>> 
>>>>>>>>>>> default
>>>>>>>>>>> 
>>>>>>>>>>>> value (shortened class name) is the name of the entity
>>> 
>>> used
>>> 
>>>>>>>>>>>> in
>>>>>>>> 
>>>>>>>> queries
>>>>>>>> 
>>>>>>>>>>> ans
>>>>>>>>>>> 
>>>>>>>>>>>> well as other places such as meta model. I know the
>>>>>>>>>>>> topic
>>> 
>>> of
>>> 
>>>>>>>>>>>> entity name
>>>>>>>>>>> 
>>>>>>>>>>> was
>>>>>>>>>>> 
>>>>>>>>>>>> discussed  many times in the past (HHH-2597, HHH-4375,
>>>>> 
>>>>> HHH-4465,
>>>>> 
>>>>>>>>>>> HHH-5709),
>>>>>>>>>>> 
>>>>>>>>>>>> and the rationale has been given for current
>>> 
>>> implementation
>>> 
>>>>>>>>>>>> of
>>>>>>>>>>> 
>>>>>>>>>>> Hibernate,
>>>>>>>>>>> 
>>>>>>>>>>>> however the topic hasn't been touched in the context of
>>> 
>>> meta
>>> 
>>>>>>>>>>>> model that
>>>>>>>>>>> 
>>>>>>>>>>> is
>>>>>>>>>>> 
>>>>>>>>>>>> new stuff added in JPA2.
>>>>>>>>>>>> 
>>>>>>>>>>>> I just wanted to know what is your thinking with regards
>>> 
>>> to
>>> 
>>>>> the
>>>>> 
>>>>>>>> matter
>>>>>>>> 
>>>>>>>>>>> at
>>>>>>>>>>> 
>>>>>>>>>>>> this stage and how likely is implementation of Hibernate
>>>>>>>>>>>> would
>>>>>>>> 
>>>>>>>> change
>>>>>>>> 
>>>>>>>>>>>> to match EntityType.getName with @Entity.name.
>>>>>>>>>>>> 
>>>>>>>>>>>> P.S.
>>>>>>>>>>>> I just have spilled the entire cup of fresh coffee on my
>>>>> 
>>>>> desktop
>>>>> 
>>>>>>>> while
>>>>>>>> 
>>>>>>>>>>>> writing this message. The mouse seems to be drowned :|
>>>>>>>>>>>> 
>>>>>>>>>>>> Cheers,
>>>>>>>>>>>> Tom
>>>>>>>>>>>> _______________________________________________
>>>>>>>>>>>> hibernate-dev mailing list
>>>>>>>>>>>> hibernate-dev at lists.jboss.org
>>>>>>>>>>>> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>>>>>>>>> 
>>>>>>>>> _______________________________________________
>>>>>>>>> hibernate-dev mailing list
>>>>>>>>> hibernate-dev at lists.jboss.org
>>>>>>>>> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>>>>>>>> 
>>>>>>>> ---
>>>>>>>> Steve Ebersole <steve at hibernate.org>
>>>>>>>> http://hibernate.org
>>>>>>>> _______________________________________________
>>>>>>>> hibernate-dev mailing list
>>>>>>>> hibernate-dev at lists.jboss.org
>>>>>>>> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>>>>>> 
>>>>>> ---
>>>>>> Steve Ebersole <steve at hibernate.org>
>>>>>> http://hibernate.org
>>>>> 
>>>>> ---
>>>>> Steve Ebersole <steve at hibernate.org>
>>>>> http://hibernate.org
>>>>> _______________________________________________
>>>>> hibernate-dev mailing list
>>>>> hibernate-dev at lists.jboss.org
>>>>> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>>> 
>>> ---
>>> Steve Ebersole <steve at hibernate.org>
>>> http://hibernate.org
>>> _______________________________________________
>>> hibernate-dev mailing list
>>> hibernate-dev at lists.jboss.org
>>> https://lists.jboss.org/mailman/listinfo/hibernate-dev
> 
> ---
> Steve Ebersole <steve at hibernate.org>
> http://hibernate.org
> _______________________________________________
> hibernate-dev mailing list
> hibernate-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hibernate-dev

More information about the hibernate-dev mailing list