jboss-ql ORDER BY in 3.2.2 -- bug?

[jfairbairn]

Hello all,

I've got CMP bean aBean and bean bBean. I've got a query as follows:

select object(a) from aBean a order by a.bBean.name

The problem is that we get duplicate aBeans back in the returned collection; the number of duplicates is equal to the number of bBeans, so if there are 6 rows in bBean's table, we get 6 duplicate sets of aBeans back.

In JBoss 3.0.7 and 3.0.8, we got the expected behaviour, i.e. a collection of distinct aBeans ordered by the name field of their associated bBean.

Is this a bug, or am I misunderstanding the way jboss-ql works in 3.2.2?

At a guess, if this is a bug, the SQL query that's being built isn't setting a join condition (WHERE a_bean.b_bean_id=b_bean.b_bean_id), as we are getting what looks like a cartesian product of the contents of the 2 tables. But that's just a guess. Anyone give us a clue?

Thanks,

James

[galatti]

I´m having exacly the same problem. Have you found the solution? Thanks

[carloseugenio]

Have you tried to use the SELECT DISTINCT ...?

eugenio.

[invisiblemage]

Hi,
this seems to be a bug.. (at least in 3.2.2)
in the following query

select object(c) from sometable
where c.id=?1
order by c.othertable.id


a cross join between sometable and othertable is created.
In 3.2.2RC4 this worked fine, though.


Boris