I recently uncovered a somewhat subtle bug in the DB2Dialect. When running a limit query that does a distinct, it was applying the distinct incorrectly. The distinct was being placed at the beginning of the query. Unfortunately, this means that the distinct was being applied to the rownumber() as well. Since rownumber() always returns a unique value for each result of the query, the net result was the that the intended distinct was not having any affect. I fixed this in by adding yet another nested select, and applying the distinct to this. We spent a good bit of time trying to see if there was another solution that didn't involve yet another subselect, but there didn't seem to be. In any event, this modification did have desired effect of return correctly distinct results when a limit query is performed. I also created a testcase which I am attaching as well.
If it would be better, I could add this bug in JIRA,
but since I am submitting a fix I wasn't sure this was
necessary.
--Chris Nelson
__________________________________
Do you Yahoo!?
Yahoo! Mail - You care about security. So do we.
http://promotions.yahoo.com/new_mail
DB2Dialect.java
Description: DB2Dialect.java
DB2DialectTest.java
Description: DB2DialectTest.java
