We are prefetching relationships. We prefer to do it that way because
sometimes we don't need to get them.
Quoting Marina Vatkina <Marina.Vatkina_at_Sun.COM>:
> Ellen,
>
> Are you adding a file category to a study before passing it back?
> Is 'fileCategories' marked as FetchType.EAGER when you pass a
> study to the client (ToMany relationship defaults to LAZY fetching),
> or do you prefetch the relationships? If not, does it make a difference
> if you do?
>
> thanks,
> -marina
>
> Ellen Kraffmiller wrote:
>> Thanks Marina, please see my comments
>> -Ellen
>>
>> Marina Vatkina wrote:
>>
>>> Ellen,
>>>
>>> Ellen Kraffmiller wrote:
>>>
>>>> Hi Gordon & Marina,
>>>> After doing more testing, the problem seems to be related to
>>>> using a detached entity. In our original example, the study
>>>> object we are merging was previously detached and sent to the web
>>>> tier. I tried the same logic with a Study entity that is
>>>> managed, and it works.
>>>>
>>>> So, to summarize, this works:
>>>> doUpdate(Long id) {
>>>> Study study = em.find(Study.class, id);
>>>> FileCategory c = new FileCategory();
>>>> c.setName( "test" );
>>>> c.setStudy(study);
>>>> study.getFileCategories().add(c);
>>>
>>>
>>>
>>> You do not need to merge a managed instance, assuming you have cascade
>>> PERSIST or ALL set:
>>>
>>>> em.merge(study);
>>>
>>>
>> Right, I forgot about that when I was doing the test.
>>
>>>>
>>>> }
>>>>
>>>> But the following throws an Opimistic Lock Exception:
>>>> (No modifications to the fileCategories collection were made
>>>> while the study was detached.)
>>>>
>>>> doUpdate(Study detachedStudy) {
>>>> FileCategory c = new FileCategory();
>>>> c.setName( "test" );
>>>> c.setStudy(study);
>>>
>>>
>>>
>>> Do you mean 'detachedStudy'?
>>
>>
>> Yes, that was a typo.
>>
>>>
>>>> study.getFileCategories().add(c);
>>>> em.merge(study);
>>>
>>>
>>>
>>> Does the opposite works (with appropriate cascade settings)?
>>>
>>> em.merge(study);
>>>
>>> FileCategory c = new FileCategory();
>>> c.setName( "test" );
>>> c.setStudy(study);
>>> study.getFileCategories().add(c);
>>>
>>>
>> This may work, but it doesn't really fit our requirements. Our
>> example is simplified to show what is happening in one code
>> snippet. Just to give you a background on what we are doing, we are
>> implementing a standard web application with forms for updating a
>> Study and all its related data. So, in this case, we are getting a
>> Study entitiy from the database, presenting a form to the user
>> where he can update the file categories for the Study (as well as
>> other dependent entities), then calling a session bean method that
>> attempts to merge the Study to the database. We can't add the
>> fileCategory object after merging the study, because the study
>> being passed to the session bean already has all the file category
>> data (and other dependent collections) set.
>>
>>>
>>> thanks,
>>> -marina
>>>
>>>> }
>>>>
>>>> Is something we need to be doing differently to merge the detached study?
>>>>
>>>> Thanks,
>>>> Ellen
>>>>
>>>> Marina Vatkina wrote:
>>>>
>>>>> Gordon,
>>>>>
>>>>> Is there a way to check which exactly PK causes this error?
>>>>> Can it be that there is another FileCategory instance that
>>>>> is the problem?
>>>>>
>>>>> thanks,
>>>>> -marina
>>>>>
>>>>> Gordon Yorke wrote:
>>>>>
>>>>>> Nothing jumps out as an issue, if you are not on the latest
>>>>>> version of TopLink Essentials I would recommend updating your
>>>>>> TopLink Essentials jar and if the problem persists file a
>>>>>> glassfish issue with testcase.
>>>>>> --Gordon
>>>>>>
>>>>>> -----Original Message-----
>>>>>> From: Ellen Kraffmiller [mailto:ekraffmiller_at_hmdc.harvard.edu]
>>>>>> Sent: Thursday, September 28, 2006 11:24 AM
>>>>>> To: persistence_at_glassfish.dev.java.net
>>>>>> Subject: Re: Trting to add new child entity; getting
>>>>>> OptimisticLockException
>>>>>>
>>>>>>
>>>>>> The version is being left as null - we are never setting it.
>>>>>> Thanks,
>>>>>> Ellen
>>>>>>
>>>>>> Gordon Yorke wrote:
>>>>>>
>>>>>>
>>>>>>> Hello Ellen,
>>>>>>> In the constrctor of FileCategory is version being initialized
>>>>>>> to new Long(X) or being left as null?
>>>>>>> --Gordon
>>>>>>>
>>>>>>> -----Original Message-----
>>>>>>> From: Ellen Kraffmiller [mailto:ekraffmiller_at_hmdc.harvard.edu]
>>>>>>> Sent: Thursday, September 28, 2006 10:27 AM
>>>>>>> To: persistence_at_glassfish.dev.java.net
>>>>>>> Subject: Re: Trting to add new child entity; getting
>>>>>>> OptimisticLockException
>>>>>>>
>>>>>>>
>>>>>>> Hi Gordon,
>>>>>>> I am working with Gustavo on this problem so I thought I'd answer your
>>>>>>> questions. The version field is defined as:
>>>>>>>
>>>>>>> @Version
>>>>>>> private Long version;
>>>>>>> (with normal getters and setters)
>>>>>>>
>>>>>>> We are using SostgreSQL, and the database column is version, int8.
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Ellen
>>>>>>>
>>>>>>> Gordon Yorke wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>> Hello Gustavo,
>>>>>>>> What is the version field in FileCategory being initialized
>>>>>>>> to? What type is the version field?
>>>>>>>> --Gordon
>>>>>>>>
>>>>>>>> -----Original Message-----
>>>>>>>> From: gdurand_at_hmdc.harvard.edu [mailto:gdurand_at_hmdc.harvard.edu]
>>>>>>>> Sent: Wednesday, September 27, 2006 5:57 PM
>>>>>>>> To: persistence_at_glassfish.dev.java.net
>>>>>>>> Subject: Trting to add new child entity; getting
>>>>>>>> OptimisticLockException
>>>>>>>>
>>>>>>>>
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> I'm am trying a very simple example and am getting a problem
>>>>>>>> (I am using glassfish v2 milestone 1):
>>>>>>>>
>>>>>>>> Our DB has a Study entity which has a one to many relationship
>>>>>>>> with a FileCategory entity.
>>>>>>>>
>>>>>>>> When I try to add a FileCategory to an existing study:
>>>>>>>>
>>>>>>>> Collection categories = study.getFileCategories();
>>>>>>>>
>>>>>>>> FileCategory c = new FileCategory();
>>>>>>>> c.setName( "test" );
>>>>>>>>
>>>>>>>>
>>>>>>>> c.setStudy(study);
>>>>>>>> categories.add(c);
>>>>>>>>
>>>>>>>> em.merge(study);
>>>>>>>>
>>>>>>>> I am getting:
>>>>>>>>
>>>>>>>> Exception Details:
>>>>>>>> oracle.toplink.essentials.exceptions.OptimisticLockException
>>>>>>>> Exception Description: The object
>>>>>>>> [edu.harvard.hmdc.vdcnet.study.FileCategory_at_1f1c748] cannot be merged
>>>>>>>> because it has changed or been deleted since it was last read.
>>>>>>>> {3}Class> edu.harvard.hmdc.vdcnet.study.FileCategory
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Any ideas why this is happening, since the FileCategory doesn't exist
>>>>>>>> in the first place? (I have seen an example in the O'Reilly book that
>>>>>>>> pretty much does just this)?
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>> Gustavo
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>
>>>