[
https://issues.apache.org/jira/browse/OPENJPA-717?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12628492#action_12628492
]
Xiaoqin Feng commented on OPENJPA-717:
--------------------------------------
In JDBCStoreManager.selectBaseMappings(), it recursively call parent class
selectBaseMappings(). For one-to-one relationship in parent class
fms[i].selectEagerJoin(sel, sm, this, fetch.traverseJDBC(fms[i]), eager) is
called so join is added to join set.
When recursive call returns to subclass, sub-super join will add to join sets.
That is the result you see in the first SQL.
For one-to-many relation, it calls fms[i].selectEagerParallel(esel, sm, this,
fetch.traverseJDBC(fms[i]), eager);
In current code design, it uses SelectExecutor and start with new join set in
StoreCollectionFieldStrategy.selectEager().
So there is no way to have sub-super join info for one-to-many relation.
I will attach a test case later.
> OpenJPA doesn't generate SQL with efficient joins when joined inheritance
> strategy is used and there is one-many relation in superclass
> ---------------------------------------------------------------------------------------------------------------------------------------
>
> Key: OPENJPA-717
> URL: https://issues.apache.org/jira/browse/OPENJPA-717
> Project: OpenJPA
> Issue Type: Improvement
> Components: query
> Affects Versions: 1.0.0, 1.0.1, 1.0.2, 1.0.3, 1.1.0, 1.2.0
> Reporter: Xiaoqin Feng
>
> The test case has a subclass that extends an abstract superclass that
> basically contains a single additional field 'S1', a one-to-one relation
> 'superOne' and a one-to-many relation 'superMany'. The superclass has an
> inheritance strategy of 'JOINED'.
> When executing the query:
> SELECT s FROM Subclass s WHERE s.a1 = :a1
> three SQL statements get generated:
> SELECT t1.ID, t0.id, t1.S1, t2.ID, t2.S1, t0.A1, t3.ID, t3.A1
> FROM SUBCLASS t0, SUPERCLASS t1, SUPERCLASSONE t2, SUBCLASSONE t3
> WHERE (t0.A1 = ?) AND t0.SUBCLASSONE_ID = t3.ID(+) AND t0.id = t1.ID AND
> t1.SUPERCLASSONE_ID = t2.ID(+)
> SELECT t0.id, t1.ID, t1.A1
> FROM SUBCLASS t0, SUBCLASSMANY t1
> WHERE (t0.A1 = ?) AND t0.id = t1.SUBCLASS_ID ORDER BY t0.id ASC
> SELECT t1.ID, t2.ID, t2.S1
> FROM SUBCLASS t0, SUPERCLASS t1, SUPERCLASSMANY t2
> WHERE (t0.A1 = ?) AND t1.ID = t2.SUPERCLASS_ID ORDER BY t1.ID ASC
> The third one doesn't contain the condition t0.ID = t1.ID
> Although final returned result set is correct because of the logical union,
> the third SQL returns more rows so it has bad performance.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
