Query generation differences between EDS and JDV produce unexpected results
Issue
- Join sort order altered inappropriately
- Does DV no longer support using a comma separated list in conjunction with the equals operator?
- If a merge or other join configuration is chosen and there is a subquery affecting only a side of the join that is to be sorted such that the subquery can be rewritten as a semi/anti-semi join, then the subquery rewrite will alter the intended sort order of the original join. This can result in incorrect results.
An example query:
select pm2.g2.e1 from pm1.g1 inner join pm2.g2 on (pm1.g1.e1 = pm2.g2.e1) where pm2.g2.e2 in /*+ mj */ (select count(e2) FROM pm1.g2 group by e1 having e1 < pm2.g2.e3)
Environment
- Red Hat JBoss Data Virtualization (DV) 6
- JBoss Enterprise Data Services (EDS) 5
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.