Jon,
By global cache I meant, cache at EntityManagerFactory level which is
maintained across transactions.
Jon Miller wrote:
> Mitesh,
>
> You were right. I wasn't managing my relationships correctly. I found
> out what it was. That's interesting what you said about the cache
> being global (I was assuming there was a cache per EntityManager).
> Does that mean that if I call EntityManager.clear() on one
> EntityManager in one thread, it will clear the objects that another
> thread using another EntityManager has in it's cache?
No. That will not happen.
> The way I've been doing things so far is I have a single
> EntityManagerFactory which is shared between threads and each thread
> has their own EntityManager.
I think thats the correct pattern to use.
Regards,
Mitesh
>
> Jon
>
> ----- Original Message ----- From: "Mitesh Meswani"
> <Mitesh.Meswani_at_Sun.COM>
> To: <persistence_at_glassfish.dev.java.net>
> Sent: Wednesday, January 10, 2007 8:24 PM
> Subject: Re: java.lang.IllegalStateException: During synchronization a
> new object was found through a relationship that was not marked
> cascade PERSIST.
>
>
>> Hi Jon,
>>
>> The symptoms seem to indicate that somewhere there is a mismatch in
>> relationship management. Please check. Toplink maintains a global
>> object level cache (across transactions). Hence closing and opening
>> an em will not help.
>>
>> -Mitesh
>>
>> Jon Miller wrote:
>>> I don't know if this is a bug or not. I'm at a loss as to what might
>>> be the problem. I'm receiving the following exception. It works OK
>>> with Hibernate, but not with the version of TopLink that comes with
>>> NetBeans 5.5 or with V2 B30. I set my logging level to finest and I
>>> can see the objects that are being "registered" into the persistence
>>> context. It looks like it's calling toString() on my objects which
>>> prints the id values out. I can see that there are only two objects
>>> which have null values (i.e. haven't been persisted yet). One of the
>>> objects is the one that I'm calling merge() on and the other is
>>> linked to that object with a CascadeType.ALL. The weird thing is
>>> that if I call EntityManager.clear() first before I do my merge, it
>>> works. When it fails, it seems to corrupt things even after I close
>>> the EntityManager. i.e. I close the EntityManager and get a new one
>>> and receive the exception when attempting to do other things like
>>> perform unrelated queries.
>>>
>>> Jon
>>>
>>> ----- Original Message ----- From: "Jon Miller" <jemiller_at_uchicago.edu>
>>> To: "Glassfish Persistence List" <persistence_at_glassfish.dev.java.net>
>>> Sent: Wednesday, January 10, 2007 6:13 PM
>>> Subject: java.lang.IllegalStateException: During synchronization a
>>> new object was found through a relationship that was not marked
>>> cascade PERSIST.
>>>
>>>
>>>> Hi all,
>>>>
>>>> I'm running into the following exception. I'm wondering if anyone
>>>> has suggestions on how to find the offending object? I'm thinking
>>>> it might be nice if the error message did something like print the
>>>> toString() representation of the offending object.
>>>>
>>>> Caused by: javax.faces.el.EvaluationException:
>>>> javax.persistence.RollbackException:
>>>> java.lang.IllegalStateException: During synchronization a new
>>>> object was found through a relationship that was not marked cascade
>>>> PERSIST.
>>>>
>>>> Jon
>>>>
>>