[
https://issues.apache.org/jira/browse/CALCITE-7741?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ASF GitHub Bot updated CALCITE-7741:
------------------------------------
Labels: pull-request-available (was: )
> Outer join register incorrect types for their input collections
> ---------------------------------------------------------------
>
> Key: CALCITE-7741
> URL: https://issues.apache.org/jira/browse/CALCITE-7741
> Project: Calcite
> Issue Type: Bug
> Components: core
> Affects Versions: 1.42.0
> Reporter: Mihai Budiu
> Assignee: Mihai Budiu
> Priority: Minor
> Labels: pull-request-available
>
> A LEFT join will produce nullable output fields for all fields that come from
> the RIGHT input, even if the right input does not have nullable fields. This
> is correct.
> However, a LEFT join in Calcite will register a nullable types for the fields
> of the right INPUT in the Rel tree. That is wrong: the output fields of the
> join are nullable, but the input field types should not change just because
> they are part of a join.
> As a concrete example, consider the following query:
> {code:java}
> select e.ename, d.name from emp as e
> left join dept as d on trim(d.name) = 'x {code}
> If you check the type of the TRIM function call in the Rel tree, it is
> nullable, although its argument is not nullable. Type inference for TRIM
> produces a non-nullable type.
> This is the plan after Sql2Rel containing the TRIM expression:
> {code:java}
> LogicalProject(ENAME=[$1], NAME=[$10])
> LogicalJoin(condition=[=(TRIM(FLAG(BOTH), ' ', $10), 'x')], joinType=[left])
> LogicalTableScan(table=[[CATALOG, SALES, EMP]])
> LogicalTableScan(table=[[CATALOG, SALES, DEPT]]) {code}
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
