How to make Hibernate only query the default mapping?

Question

I have an entity class:

Cabbage.java

package womble;

public class Cabbage {
    private int id;
    // getters, setters, etc.
}

That I've got mapped to two separate database tables as follows:

Cabbage.hbm.xml

<hibernate-mapping>
  <class name="womble.Cabbage"> <!-- Using the default entity and table name. -->
    <id name="id" unsaved-value="null">
      <generator class="native"/>
    </id>
  </class>
</hibernate-mapping>

CabbageBackup.hbm.xml

<hibernate-mapping>
  <class name="womble.Cabbage" entity-name="CabbageBackup" table="cabbage_backup">
    <id name="id">
      <generator class="assigned"/>
    </id>
  </class>
</hibernate-mapping>

I'm trying to retrieve the last saved Cabbage instance like this:

CabbageDao.java

package womble;

public class CabbageDao {
    Session session; // Hibernate session
    String GET_MAX_ID = "select max(id) from Cabbage";

    public getLastCabbage() {
        Object lastId = session.createQuery(GET_MAX_ID).uniqueResult();
        return session.load(Cabbage.class, lastId);
    }
}

This blows up with a NonUniqueResultException: "query did not return a unique result: 2"

Looking through the logs, Hibernate apparently queries both the tables and then joins the results together.

Is there a way to make Hibernate only query the default mapping for an entity? (That is, to specifically prevent it from magically querying both mappings when I use the simple class name.)

I tried using the above, using the criteria API passing Cabbage.class to #createCriteria(), as well as the string "Cabbage", both select from both tables. I'm trying to avoid specifying an entity name to the "main" mapping, because there's multiple classes like that in my code and I'd have to fix up all the relationships between them as well.

Answer 1

Try using an entity-name on both mappings, and specify which you'd like. I believe that is the contract of that attribute.

Answer 2

Well, an easy answer (not necessarily the best one) would be to use .list() instead of .uniqueResult() bringing back a Collection of the max ids. You could then toss the results into a new HashSet and let Java do the work of unique-ing them. Not elegant, but it would work.

public getLastCabbage() {
    Set lastIds = new HashSet(session.createQuery(GET_MAX_ID).list());
    // ignore the possibility of null; full steam ahead
    return session.load(Cabbage.class, lastIds.iterator().next());
}

Answer 3

Firstly, Hibernate has no notion here of your "default" - both mappings are on an equal footing as far as it's concerned.

Do you ever need to access old cabbages (not a phrase I've ever used before) from the backup table, or are you just using that table for archiving old data?

If the latter, then there are probably better ways of achieving the same result, such as using a database trigger.

Answer 4

You might consider one of the extensions to hibernate which do this table revisions for you. I've never used envers personally, but the documentation looked interesting.