Using the following entity configuration:
{code:title=ParentEntity.java}
@Entity
@Audited
@Inheritance(strategy = InheritanceType.JOINED)
@DiscriminatorColumn(name = "type")
public abstract class ParentEntity {
@Id
@GeneratedValue
private Integer id;
private String type;
/** getter/setters **/
}
{code}
{code:title=ChildEntity.java}
@Entity
@Audited
@DiscriminatorValue("child")
public class ChildEntity extends ParentEntity {
private String name;
/** getter/setters **/
}
{code}
Hibernate will insert **child** as the discriminator value on the ORM side; however, the fully qualified class name of the audited entity will be used in the envers audit tables.
{noformat}
insert
into
ParentEntity_AUD
(REVTYPE, type, id, REV)
values
(?, 'org.hibernate.test.stackoverflow.StackoverflowTest$ChildEntity_AUD', ?, ?)
09:08:25,631 TRACE BasicBinder:64 - binding parameter [1] as [INTEGER] - [0]
09:08:25,632 TRACE BasicBinder:64 - binding parameter [2] as [BIGINT] - [1]
09:08:25,632 TRACE BasicBinder:64 - binding parameter [3] as [INTEGER] - [1]
{noformat}
In Hibernate 4.3.11.Final, this wasn't observed and the actual configured discriminator value was used instead. This also introduces a column width problem as column width is configured based on the expected configured values, which are often far shorter than the fully qualified name; causing insertion failures on the Envers side. |
|