]
Steven Hawkins updated TEIID-949:
---------------------------------
Fix Version/s: 7.0
Affects Version/s: 6.0.0
(was: 7.0)
Assignee: Steven Hawkins (was: Larry O'Leary)
Criteria on constant column with aggregate function doesn't push
criteria down correctly
----------------------------------------------------------------------------------------
Key: TEIID-949
URL:
https://jira.jboss.org/jira/browse/TEIID-949
Project: Teiid
Issue Type: Bug
Components: Query Engine
Affects Versions: 6.0.0
Environment: MM5.5.4
Reporter: Larry O'Leary
Assignee: Steven Hawkins
Fix For: 7.0
When a column value is a constant and used along with an aggregate, criteria on the
constant column is being ignored. Here is an example query:
SELECT * FROM (
SELECT b.count, enterprise_id FROM (
SELECT COUNT(*), 35 AS enterprise_id FROM (
SELECT "A Name" AS Name, 37 AS enterprise_id
) c
) b
) a WHERE enterprise_id = 1 OPTION DEBUG
The expected result is an empty result set. Instead, the result set comes back as:
COUNT, ENTERPRISE_ID
0, 35
The resulting plan looks like:
OPTIMIZATION COMPLETE:
PLAN TREE:
(type=Relational, command=SELECT * FROM (SELECT b."count", enterprise_id FROM
(SELECT COUNT(*), 35 AS enterprise_id FROM (SELECT 'A Name' AS Name, 37 AS
enterprise_id) AS c) AS b) AS a WHERE enterprise_id = 1 OPTION DEBUG,
props={0=PlanHints})
canonical plan:
Project(groups=[b], props={90=[b."count", enterprise_id],
30=[b."count", enterprise_id], 114=0.0, 91=[A."COUNT",
A.ENTERPRISE_ID]})
Source(groups=[b], props={90=[b."count", enterprise_id], 64=SELECT COUNT(*),
35 AS enterprise_id FROM (SELECT 'A Name' AS Name, 37 AS enterprise_id) AS c,
114=0.0, 60={b."count"=COUNT(*), b.enterprise_id=35}})
Project(groups=[], props={90=[b."count", enterprise_id], 30=[COUNT(*), 35
AS enterprise_id], 114=0.0, 60={b."count"=COUNT(*), b.enterprise_id=35},
91=[COUNT(*), 35 AS enterprise_id]})
Group(groups=[c], props={90=[COUNT(*)], 114=0.0, 81=[COUNT(*)]})
Null(groups=[c], props={90=[], 114=0.0})
PROCESSOR PLAN:
ProjectNode(1) output=[b."count", enterprise_id] [b."count",
enterprise_id]
ProjectNode(2) output=[b."count", enterprise_id] [COUNT(*), 35 AS
enterprise_id]
GroupingNode(3) output=[COUNT(*)] null
NullNode(4) output=[]
If the query is changed as follows, the expected result is received:
SELECT * FROM (
SELECT b.count, enterprise_id FROM (
SELECT COUNT(*), 35 AS enterprise_id FROM (
SELECT "A Name" AS Name, 37 AS enterprise_id
) c GROUP BY enterprise_id
) b
) a WHERE enterprise_id = 1 OPTION DEBUG
Which yields a very similar plan:
PLAN TREE:
(type=Relational, command=SELECT * FROM (SELECT b."count", enterprise_id FROM
(SELECT COUNT(*), 35 AS enterprise_id FROM (SELECT 'A Name' AS Name, 37 AS
enterprise_id) AS c GROUP BY enterprise_id) AS b) AS a WHERE enterprise_id = 1 OPTION
DEBUG, props={0=PlanHints})
canonical plan:
Project(groups=[b], props={90=[b."count", enterprise_id],
30=[b."count", enterprise_id], 114=0.0, 91=[A."COUNT",
A.ENTERPRISE_ID]})
Source(groups=[b], props={90=[b."count", enterprise_id], 64=SELECT COUNT(*),
35 AS enterprise_id FROM (SELECT 'A Name' AS Name, 37 AS enterprise_id) AS c GROUP
BY enterprise_id, 114=0.0, 60={b."count"=COUNT(*), b.enterprise_id=35}})
Project(groups=[], props={90=[b."count", enterprise_id], 30=[COUNT(*), 35
AS enterprise_id], 114=0.0, 60={b."count"=COUNT(*), b.enterprise_id=35},
91=[COUNT(*), 35 AS enterprise_id]})
Group(groups=[c], props={90=[COUNT(*)], 114=0.0, 80=[enterprise_id],
81=[COUNT(*)]})
Null(groups=[c], props={90=[enterprise_id], 114=0.0})
PROCESSOR PLAN:
ProjectNode(1) output=[b."count", enterprise_id] [b."count",
enterprise_id]
ProjectNode(2) output=[b."count", enterprise_id] [COUNT(*), 35 AS
enterprise_id]
GroupingNode(3) output=[COUNT(*)] [enterprise_id]
NullNode(4) output=[enterprise_id]
Except the NullNode node contains enterprise_id as its target column vs an empty value.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: