You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@calcite.apache.org by "Maryann Xue (JIRA)" <ji...@apache.org> on 2015/09/21 21:18:04 UTC
[jira] [Created] (CALCITE-894) Do not generate redundant column
alias for the left relation when translating IN subquery
Maryann Xue created CALCITE-894:
-----------------------------------
Summary: Do not generate redundant column alias for the left relation when translating IN subquery
Key: CALCITE-894
URL: https://issues.apache.org/jira/browse/CALCITE-894
Project: Calcite
Issue Type: Bug
Affects Versions: 1.4.0-incubating
Reporter: Maryann Xue
Assignee: Maryann Xue
Priority: Minor
{code}
select * from "hr"."emps" where "deptno" in (
select "deptno" from "hr"."depts")
{code}
would be converted into
{code}
LogicalProject(empid=[$0], deptno=[$1], name=[$2], salary=[$3], commission=[$4])
LogicalJoin(condition=[=($5, $6)], joinType=[inner])
LogicalProject($f0=[$0], $f1=[$1], $f2=[$2], $f3=[$3], $f4=[$4], $f5=[$1])
EnumerableTableScan(table=[[hr, emps]])
LogicalAggregate(group=[{0}])
LogicalProject(deptno=[$0])
EnumerableTableScan(table=[[hr, depts]])
{code}
There is an additional "$f5=[$1]" in the LogicalProject, which might cause us trouble in the later optimization stage. For example if "emps" table had a collation trait on $1, the Project would have multiple collation traits and would be flattened.
Instead, we want the converted rel to be like:
{code}
LogicalProject(empid=[$0], deptno=[$1], name=[$2], salary=[$3], commission=[$4])
LogicalJoin(condition=[=($1, $5)], joinType=[inner])
EnumerableTableScan(table=[[hr, emps]])
LogicalAggregate(group=[{0}])
LogicalProject(deptno=[$0])
EnumerableTableScan(table=[[hr, depts]])
{code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)