users@jpa-spec.java.net

[jpa-spec users] [jsr338-experts] Re: Current standalone TCK drop

From: Steve Ebersole <steve.ebersole_at_redhat.com>
Date: Wed, 13 Mar 2013 11:22:47 -0500

Well I did follow all documented instructions... ;)

On Wed 13 Mar 2013 11:16:01 AM CDT, Linda DeMichiel wrote:
>
>
> On 3/13/2013 9:03 AM, Steve Ebersole wrote:
>> Any idea on the second issue? Its by far the more obstructive of the
>> 2. Hard to test a provider if you cant tell the TCK
>> to use said provider :)
>>
>
> This sounds to me like a configuration issue. I'll let our TCK folks
> try to help....
>
>
>> On 03/13/2013 10:43 AM, Linda DeMichiel wrote:
>>>
>>>
>>> On 3/13/2013 8:16 AM, Steve Ebersole wrote:
>>>> As I understand it the current TCK drop is not supposed to be
>>>> final, but I have found serious problems with it. To the
>>>> point where it is essentially unusable.
>>>>
>>>> First, the signature tests are wrong. Running against the JPA API
>>>> jar I created for Hibernate testing using the latest
>>>> Proposed Final Draft I get errors that
>>>> "javax.persistence.criteria.CommonAbstractQuery" is missing from my
>>>> jar. But
>>>> javax.persistence.criteria.CommonAbstractQuery was removed from the
>>>> spec way back in Draft 6.
>>>>
>>>
>>> This is a bug in our APIs. It looks like I neglected to remove it
>>> after our having gone back and forth
>>> on the factorization so many times.
>>>
>>>
>>>> Second, I am completely unable to run any of the actual persistence
>>>> tests. Every single one of them fails in perfoming
>>>> set up. The error is always the same:
>>>>
>>>> 03-13-2013 01:22:19: ERROR: Map returned from emf.getProperties()
>>>> was null,this should never occur
>>>>
>>>> However, it is impossible for the Hibernate EMF to return a null
>>>> Map there. So, since I can't actually debug the tests
>>>> (or please I'd love to have details on how to actually accomplish
>>>> that) I fell back to caveman debugging and put
>>>> throwing exceptions as the first thing I do inside (1) the
>>>> Hibernate PersistenceProvider#createEntityManagerFactory call
>>>> and the constructor to the Hibernate EMF impl. Neither gets
>>>> triggered. So that tells me that the TCK is not even
>>>> attempting to call out to the configured "alternate provider".
>>>>
>>>>
>>