<hibernate-mapping default-lazy="false" package="edu.mayo.gensupp.hmm.businessobject">
<class name="InventoryItem"
table="hmmtb_inventory_item">
<id name="id" type="int" column="inv_item_id">
<generator class="org.hibernate.id.MultipleHiLoPerTableGenerator">
<param name="table">hmmtb_next_id</param>
<param name="primary_key_column">nxt_pk_col</param>
<param name="value_column">nxt_pk_value</param>
<param name="primary_key_value">inv_item_id</param>
<param name="primary_key_length">30</param>
<param name="max_lo">1</param>
</generator>
</id>
<many-to-one name="inventory" class="Inventory" column="inv_id" />
<property name="itemType" type="string" column="item_type" insert="false" update="false" />
<any name="item" id-type="int" meta-type="string">
<meta-value value="P" class="Product"/>
<meta-value value="U" class="UnreviewedItem"/>
<column name="item_type"/>
<column name="item_id"/>
</any>
<!-- unit of measure may not be known -->
<many-to-one name="unitOfMeasure" class="DropdownValue" column="inv_item_uom_id" not-found="ignore" />
<property name="qty" type="float" column="inv_item_qty" />
<property name="effectiveDate" type="calendar" column="inv_item_eff_date" />
<property name="expirationDate" type="calendar" column="inv_item_exp_date" />
<property name="notes" type="string" column="inv_item_notes" />
</class>
This InventoryItem class contains an object that can either be an instance of Product or UnreviewedItem, with the type determined by the item_type field. I almost shed tears of joy when I saw how painless Hibernate made this to implement.
Unfortunately the performance is terrible. Hibernate is issuing a select statement for each InventoryItem object to get its corresponding item. I tried setting the batch-size attribute directly in the any mapping and in the Product and UnreviewedItem classes and it was ignored in all places. I tried fetching the items via an inner or outer join and got null pointer exceptions when Hibernate tried to parse the query.
My questions here are a) why is the batch-size attribute ignored for this mapping type, and b) is this really the most efficient way to get the mapped objects? Why can't Hibernate do a batch fetch of each mapped type and union the results together (i.e. fetch all items of type 'P' union fetch all items of type 'U')?
Login
FAQ
Search