views:

1297

answers:

2

Is it necessary that composite-id should be mapped to class ??

can it be like this ?

<composite-id>
  <key-property=..../>
  <key-property=..../>
</composite-id>

or should be

<composite-id class=....>
  <key-property=..../>
  <key-property=..../>
</composite-id>

should that necessary that if we have composite key then that class should implement equals() and override() method ?

+2  A: 

Hibernate needs to be able to compare and serialize identifiers. So the identifier class must be serializable, and override hashCode() and equals() consistently with the database's notion of composite key equality.

If you have a composite id mapped as properties of the entity, the entity itself is the identifier.

A second approach is called a mapped composite identifier, where the identifier properties named inside the <composite-id> element are duplicated on both the persistent class and a separate identifier class

Finally, a composite-id may be a component class. In this case the component class is the identifier class.

Note that it is strongly recommended to have the ID a separate class. Otherwise you will have only very awkward ways to lookup your object using session.get() or session.load().

Relevant sections of the Reference Documentation:

In this example, a composite-id is mapped as properties of the entity.

<composite-id>
    <key-property name="EmployeeNumber"/>
    <key-property name="Dependent"/>
</composite-id>

class EmployeeAssignment implements Serializable
{
    string getEmployeeNumber()
    void setEmployeeNumber( string value )
    string getDepartment()
    void setDepartment( string value )
    boolean equals( Object obj )
    int hashCode()
}

A mapped composite-id:

<composite-id class="EmployeeAssignmentId" mapped="true">
    <key-property name="EmployeeNumber"/>
    <key-property name="Dependent"/>
</composite-id>

class EmployeeAssignment
{
    string getEmployeeNumber()
    void setEmployeeNumber( string value )
    string getDepartment()
    void setDepartment( string value )
}

class EmployeeAssignmentId implements Serializable
{
    string getEmployeeNumber()
    void setEmployeeNumber( string value )
    string getDepartment()
    void setDepartment( string value )
    boolean equals( Object obj )
    int hashCode()
}

A component as a composite-id:

<composite-id name="Id" class="EmployeeAssignmentId">
    <key-property name="EmployeeNumber"/>
    <key-property name="Dependent"/>
</composite-id>

class EmployeeAssignment
{
    EmployeeAssignmentId getId()
    void setId( EmployeeAssignmentId value )
}

class EmployeeAssignmentId implements Serializable
{
    string getEmployeeNumber()
    void setEmployeeNumber( string value )
    string getDepartment()
    void setDepartment( string value )
    boolean equals( Object obj )
    int hashCode()
}
Lachlan Roche
A: 

Both are possible. If you use

<composite-id>
  <key-property=..../>
  <key-property=..../>
</composite-id>

Then no separate class is required to represent the key. The ID values are taken from the properties of the entity itself.

If you use

Then the specified class will be a used as a holder for the key properties. However, the entity class must also have these properties - the values are stored both in the entity class and the composite ID class. The entity class has no knowledge of the key class. Not very nice, in my opinion.

There is a nicer 3rd approach, described in the docs here:

<composite-id name="id" class="OrderLineId">
    <key-property name="lineId"/>
    <key-property name="orderId"/>
    <key-property name="customerId"/>
</composite-id>

Here, the composite key is represented by the class OrderLineId, an instance of which is stored under the field id in the entity class. This keeps the separation between entity and key much cleaner.

skaffman