Bruce Robbins created SPARK-41991:
-------------------------------------
Summary: Interpreted mode subexpression elimination can throw
exception during insert
Key: SPARK-41991
URL: https://issues.apache.org/jira/browse/SPARK-41991
Project: Spark
Issue Type: Bug
Components: SQL
Affects Versions: 3.4.0
Reporter: Bruce Robbins
Example:
{noformat}
drop table if exists tbl1;
create table tbl1 (a int, b int) using parquet;
set spark.sql.codegen.wholeStage=false;
set spark.sql.codegen.factoryMode=NO_CODEGEN;
insert into tbl1
select id as a, id as b
from range(1, 5);
{noformat}
This results in the following exception:
{noformat}
java.lang.ClassCastException:
org.apache.spark.sql.catalyst.expressions.ExpressionProxy cannot be cast to
org.apache.spark.sql.catalyst.expressions.Cast
at
org.apache.spark.sql.catalyst.expressions.CheckOverflowInTableInsert.withNewChildInternal(Cast.scala:2514)
at
org.apache.spark.sql.catalyst.expressions.CheckOverflowInTableInsert.withNewChildInternal(Cast.scala:2512)
{noformat}
The query produces 2 bigint values, but the table's schema expects 2 int
values, so Spark wraps each output field with a {{Cast}}.
Later, in {{InterpretedUnsafeProjection}}, {{prepareExpressions}} tries to wrap
the two {{Cast}} expressions with an {{ExpressionProxy}}. However, the parent
expression of each {{Cast}} is a {{CheckOverflowInTableInsert}} expression,
which does not accept {{ExpressionProxy}} as a child.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
