Quanlong Huang created IMPALA-15746:
---------------------------------------
Summary: Missing operand index for intermediate tuples
Key: IMPALA-15746
URL: https://issues.apache.org/jira/browse/IMPALA-15746
Project: IMPALA
Issue Type: Bug
Components: Frontend
Reporter: Quanlong Huang
Assignee: Quanlong Huang
The following two queries has the same HBO key string in the final
AggregationNode.
Query A:
{code:sql}
SELECT c1, count(*) FROM (
SELECT STRAIGHT_JOIN sum(a.id) c1, sum(b.id) c2
FROM alltypestiny a JOIN alltypestiny b ON a.id < b.id
GROUP BY a.month
) v GROUP BY c1;{code}
Query B:
{code:sql}
SELECT c2, count(*) FROM (
SELECT STRAIGHT_JOIN sum(a.id) c1, sum(b.id) c2
FROM alltypestiny a JOIN alltypestiny b ON a.id < b.id
GROUP BY a.month
) v GROUP BY c2;{code}
The HBO key string is
{noformat}
CARDINALITY:AggregationNode:FIRST|Preagg:false|GroupingSet:false
|AggClasses:[0:Group:sum(id)]
|Child:[
CARDINALITY:AggregationNode:FIRST|Preagg:false|GroupingSet:false
|AggClasses:[0:Group:op0.`month`]
|Child:[
CARDINALITY:JoinNode:INNER|Operands:
[CARDINALITY:ScanNode:functional.alltypestiny,
CARDINALITY:ScanNode:functional.alltypestiny]
|Predicates:[op0.id < op1.id]]]{noformat}
In "AggClasses:[0:Group:sum(id)]", the "id" column should be qualified by the
operand index.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
