[
https://issues.apache.org/jira/browse/OPENJPA-977?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Pinaki Poddar reassigned OPENJPA-977:
-------------------------------------
Assignee: Pinaki Poddar
> Improve interface based domain model support
> --------------------------------------------
>
> Key: OPENJPA-977
> URL: https://issues.apache.org/jira/browse/OPENJPA-977
> Project: OpenJPA
> Issue Type: New Feature
> Components: kernel
> Reporter: Pinaki Poddar
> Assignee: Pinaki Poddar
>
> OpenJPA supports interfaces both via @ManagedInterface and @Type.
> However, my recent tests are showing several short comings in this area as
> well as I notice lack of good example or documentation of this support.
> One obvious issue seen in the test is queries do not work on interface. If
> InterfaceA is an interface implemented by X and Y, it will be natural to
> expect that a simple query 'select a from InterfaceA a' should return the
> combined extent of X and Y. But unfortunately the query fails to parse.
> The other not so obvious problem I notice is loading of relations that are
> declared to be interfaces. OpenJPA traditionally stored such references in
> database as encoded string with class name and identity value. And, on load,
> it will decode the string to determine the exact type of the instance and its
> primary key. I remember using it but currently it does not seem to. It seems
> that other changes in enhancer, or constructing identity object from a String
> have changed to cause such regression. Unfortunately, the test coverage is
> somewhat shallow for this support to catch these regressions.
> Suggested action:
> 1. Document the nature and limitation of interface support.
> 2. Create a comprehensive test suite to evaluate the capability and
> limitation of this support.
> 3. Find out the cause of regression of @Type and why the stringed identity
> decoding is broken.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
