[
https://issues.apache.org/jira/browse/HAWQ-442?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
zhenglin tao updated HAWQ-442:
------------------------------
Summary: Wrong results in rollup which contains the same expr targetlist
(was: set the wrong resname of TargetEntry for group by clause)
> Wrong results in rollup which contains the same expr targetlist
> ---------------------------------------------------------------
>
> Key: HAWQ-442
> URL: https://issues.apache.org/jira/browse/HAWQ-442
> Project: Apache HAWQ
> Issue Type: Bug
> Components: Core
> Reporter: zhenglin tao
> Assignee: zhenglin tao
>
> Reproduce steps:
> {code}
> create table sale
> (
> cn int not null,
> vn int not null,
> pn int not null,
> dt date not null,
> qty int not null,
> prc float not null
> ) WITH (appendonly=true, orientation = parquet) distributed by (cn,vn,pn);
> insert into sale values
> ( 2, 40, 100, '1401-1-1', 1100, 2400);
> taoz=# explain SELECT sale.qty as newalias1,sale.qty as newalias2 FROM sale
> GROUP BY ROLLUP(newalias2,newalias1);
> QUERY PLAN
> ------------------------------------------------------------------------------------------------------------------------------------------------
> Gather Motion 8:1 (slice2; segments: 8) (cost=1.06..1.09 rows=3 width=8)
> -> HashAggregate (cost=1.06..1.09 rows=1 width=8)
> Group By: partial_aggregation.newalias1,
> partial_aggregation.newalias1, partial_aggregation."grouping",
> partial_aggregation."group_id"
> -> Subquery Scan partial_aggregation (cost=1.02..1.06 rows=0
> width=20)
> -> Redistribute Motion 8:8 (slice1; segments: 8)
> (cost=1.02..1.06 rows=0 width=20)
> Hash Key: "rollup".newalias1, "rollup".newalias1
> -> GroupAggregate (cost=1.02..1.06 rows=0 width=20)
> Group By: "rollup"."grouping", "rollup"."group_id"
> -> Subquery Scan "rollup" (cost=1.02..1.06
> rows=0 width=20)
> -> GroupAggregate (cost=1.02..1.06 rows=0
> width=20)
> Group By: "rollup".newalias1,
> "rollup"."grouping", "rollup"."group_id"
> -> Subquery Scan "rollup"
> (cost=1.02..1.05 rows=1 width=20)
> -> GroupAggregate
> (cost=1.02..1.04 rows=1 width=20)
> Group By: sale.qty,
> sale.qty
> -> Sort (cost=1.02..1.02
> rows=1 width=4)
> Sort Key: sale.qty,
> sale.qty
> -> Parquet table
> Scan on sale (cost=0.00..1.01 rows=1 width=4)
> Settings: optimizer=off
> Optimizer status: legacy query optimizer
> (19 rows)
> {code}
> Wrongly used the name newalias1 and newalias2
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
