[ 
https://issues.apache.org/jira/browse/OPENJPA-152?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13063563#comment-13063563
 ] 

Michael Dick commented on OPENJPA-152:
--------------------------------------

I'd be in favor of throwing an exception for the reason Dain mentioned above. 
It's very easy to miss a warning like this one and not notice until you get 
back the wrong results. 

It would be interesting to run through the junit bucket with the change. The 
tests that use SimpleEntity2 and SimpleEntity will break, but it'd be good to 
know how many other entities have duplicate query names. 

If we do go forward with the change we'll want to make it configurable to give 
existing applications a migration path. A new compatibility option is probably 
the way to go here.  

> Warn or throw an exception when a persistence unit has multiple named queries 
> with the same name
> ------------------------------------------------------------------------------------------------
>
>                 Key: OPENJPA-152
>                 URL: https://issues.apache.org/jira/browse/OPENJPA-152
>             Project: OpenJPA
>          Issue Type: Improvement
>          Components: jpa, logging, query
>            Reporter: Dain Sundstrom
>            Priority: Minor
>
> The JPA spec makes it quite easy for uses to create multiple named queries 
> with the same name.  The problem stems from named queries being declared as 
> part of an entity either via annotations or nested xml, but the name space 
> for these queries is not scoped to the entity but scoped to the whole 
> persistence unit.  This problem is compounded due to the ease at which the 
> persistence unit is easily expanded with orm.xml files and with annotated 
> beans.
> I propose that we throw a deployment exception when the combined entity 
> mapping xml files and listed classes contain multiple named queries with the 
> same name.   If after deployment, we discover another annotate bean that 
> creates a conflict, we should log a ERROR and take extra cautions to not 
> override the existing in place query with the newly discovered one, as this 
> could drastically change the behavior of an application at runtime.
> Alternatively, we could just log warnings, but I would prefer a deployment 
> exception as it guarantees I'm not running with a randomly selected set or 
> queries.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Reply via email to