[teiid-issues] [JBoss JIRA] (TEIID-1991) GROUP BY pushdown into subselect of UNIONs
[
https://issues.jboss.org/browse/TEIID-1991?page=com.atlassian.jira.plugin...
]
Steven Hawkins commented on TEIID-1991:
---------------------------------------
The simplified plan does not show any grouping. Also I'm not quite sure what you mean
by changing a to 1 since you still need a projected. From what I can infer from the
larger plan you have the form:
select x from (select x, y, z ... union select x, y, z ... ) as VW group by x
which is equivalent to:
select distinct x from (select x, y, z ... union select x, y, z ... ) as VW
The logic sees that all columns are needed in the inline view since it is a union and not
a union all. Since there are no aggregates against the extra columns they can be safely
left off of the union and for that matter the union is really the same as a union all. Is
that what you are trying to convey?
If you issue the query as a union all, then Teiid will correctly omit unnecessary
columns.
GROUP BY pushdown into subselect of UNIONs
------------------------------------------
Key: TEIID-1991
URL: https://issues.jboss.org/browse/TEIID-1991
Project: Teiid
Issue Type: Bug
Components: Query Engine
Affects Versions: 7.7
Reporter: Mark Addleman
Assignee: Steven Hawkins
Attachments: plan.txt, plan.txt, simplified_plan.txt
If have a query of the form
SELECT a FROM (
SELECT a FROM (
SELECT a FROM ds.t1
) u
) t GROUP BY a
Teiid correctly optimizes the inner SELECT to SELECT 1 FROM ds.t1. However, if the inner
subselect is a union:
SELECT a FROM (
SELECT a FROM (
SELECT a FROM ds.t1 UNION
SELECT a FROM ds.t2
) u
) t GROUP BY a
Teiid does not optimize to
SELECT 1 FROM ds.t1 UNION
SELECT 1 FROM ds.t2
as I would expect.
See attached plan
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.jboss.org/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira