openjpa-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Marc Prud'hommeaux <mprud...@apache.org>
Subject Re: @IdClass annotation for id field of type byte[]
Date Wed, 03 Jan 2007 00:30:50 GMT

Interesting ... sounds like a legit bug, then (although it bears  
noting that byte[] primary keys aren't actually allowed by the JPA  
spec, as per section 2.1.4 ... support for them is an OpenJPA  
extension).

My guess is that this only affects Oracle, due to our special  
handling of blobs. It'd be interesting to see if any other databases  
that support byte[] primary keys exhibit this problem.



On Jan 2, 2007, at 7:23 PM, Igor Fedorenko wrote:

> You can use use RAW(16) to store GUIDs in Oracle. This datatype is  
> allowed in primary keys.
>
> --
> Regards,
> Igor
>
> Dain Sundstrom wrote:
>> Can you have java field of type byte[] that maps to a NUMERIC (or  
>> heck a varchar) in he db?  I'm guessing that Kevin's guid is a  
>> fixed 128 bit number.  If it is and he can map it to a non-blob  
>> type, it should be possible to join with any database system.
>> -dain
>> On Jan 2, 2007, at 3:09 PM, Marc Prud'hommeaux wrote:
>>> Kevin-
>>>
>>>> Also, this exception is supposedly only being produced with  
>>>> Oracle, not
>>>> DB2.  (I have not been able to verify that yet.)  This would  
>>>> seem to
>>>> indicate that it's dictionary-specific, but I'm not seeing  
>>>> anything there
>>>> yet...
>>>
>>> Does Oracle even support blob primary keys? My recollection is  
>>> that it didn't...
>>>
>>> I suspect that the problem might be that since Oracle has a  
>>> number of problems with in-line blobs in statements, we  
>>> frequently issue a separate statement to load and store blobs  
>>> from and to rows, but if it is the primary key, then we might be  
>>> conflicting with that. Can you post the complete stack trace?
>>>
>>>
>>>
>>>
>>> On Jan 2, 2007, at 6:03 PM, Kevin Sutter wrote:
>>>
>>>> Hi,
>>>> Some experimenting with the @IdClass support is producing a strange
>>>> exception message when attempting to map an id field of type byte 
>>>> [].
>>>> According to the OpenJPA documentation, we need to use an  
>>>> Identity Class to
>>>> use byte[] as the id field type.  Something like this:
>>>>
>>>> @Entity
>>>> @IdClass (jpa.classes.Guid.class)
>>>> @Table(name="AGENT", schema="CDB")
>>>> public class Agent {
>>>>
>>>>    @Id
>>>>    @Column(name="ME_GUID")
>>>>    private byte[] guid;
>>>> ...
>>>>
>>>> The Guid class has also been created with a single instance  
>>>> variable of type
>>>> byte[]:
>>>>
>>>> public class Guid implements Serializable {
>>>>    private byte[] guid;
>>>> ...
>>>>
>>>> But, during the loading of the database, I am getting the  
>>>> following error...
>>>>
>>>> org.apache.openjpa.util.MetaDataException: You cannot join on  
>>>> column "
>>>> AGENT.ME_GUID".  It is not managed by a mapping that supports joins
>>>>
>>>> First off, the exception is confusing since I don't believe I am  
>>>> attempting
>>>> to do a join.  The guid column is in the same table as the Agent.
>>>>
>>>> Also, this exception is supposedly only being produced with  
>>>> Oracle, not
>>>> DB2.  (I have not been able to verify that yet.)  This would  
>>>> seem to
>>>> indicate that it's dictionary-specific, but I'm not seeing  
>>>> anything there
>>>> yet...
>>>>
>>>> I am in the process of validating the problem, but I thought I  
>>>> would drop a
>>>> line to the team to see if it rings any bells...
>>>>
>>>> Thanks,
>>>> Kevin
>


Mime
View raw message