Hi all,

I am a newcomer to hibernate. It would be great if someone could comment over the following query that i have:

Say i have a parent class and each parent has multiple children. So the mapping file of parent class would be something like:

parent.hbm.xml

<hibernate-mapping >
<class name="org.demo.parent" table="parent" lazy="true">
<cache usage="read-write" region="org.demo.parent"/>
<id name="id" column="id" type="integer" length="10">
<generator class="native">
</generator>
</id>
<property name="name" column="name" type="string" length="50"/>

<set name="children" lazy="true">
<cache usage="read-write" region="org.demo.parent.children" />
<key column="parent_id"/>
<one-to-many class="org.demo.children"/>
</set>

</class>
</hibernate-mapping>


*************
children.hbm.xml

<hibernate-mapping >
<class name="org.demo.children" table="children" lazy="true">
<cache usage="read-write" region="org.demo.children"/>
<id name="id" column="id" type="integer" length="10">
<generator class="native">
</generator>
</id>

<property name="name" column="name" type="string" length="50"/>

<many-to-one name="parent_id" column="parent_id" type="integer" length="10" not-null="true"/>

</class>
</hibernate-mapping>


So for the set children, should we specify the region org.demo.parent.children where it should cache the association or should we use the cache region of org.demo.children where the children would be getting cached.

I am using EHCache as the 2nd level cache provider. I tried to search for the answer to this question but couldnt find any answer in this direction. It makes more sense to use org.demo.children but I dont know in which scenarios one should use a separate cache region for associations/sets/collections as in the above case. Kindly provide your inputs also let me know if I am not clear in my question.

Also on another note, wouldnt having 2 cache regions for children cause inconsistency. If one of the regions gets updated then I guess that change wont be reflected in the other cache region? Is my understanding correct?? So what should be the correct way to configure the caches as in the above scenario?

Thanks all.

Yes.

In practice it is indifferent in which regions you store entities or collections,
but there's one particular think to pay attention:
you should configure the regions in a way that entities do never expire before the collections which refer to this entities.
Example:
Scenario 1:
- you call a.getAllB();
- the association collection and the target B entities are all present in 2L cache
This is the perfect scenario, no database hit.
Scenario 2:
- you call a.getAllB();
- the association collection is present in 2L cache but some of the B entities which are referred by the collection
are no more present in 2l cache (already expired).
In this case Hibernate wlll issue single sql-statements in order to load the missing entities again into the caches.
Now Instead to have 0 or 1 database-hit loading the association, you can have up to <n> database hits! (where n is the size of the collection)