[
https://issues.apache.org/jira/browse/SPARK-42473?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
kevinshin updated SPARK-42473:
------------------------------
Description:
*when 'union all' and one select statement use* *Literal as column value , the
other* *select statement has computed expression at the same column , then the
whole statement will compile failed. A explicit cast will be needed.*
for example:
{color:#4c9aff}explain{color}
{color:#4c9aff}*INSERT* OVERWRITE *TABLE* test.spark33_decimal_orc{color}
{color:#4c9aff}*select* *null* *as* amt1, {*}cast{*}('256.99' *as*
{*}decimal{*}(20,8)) *as* amt2{color}
{color:#4c9aff}*union* *all*{color}
{color:#4c9aff}*select* {*}cast{*}('200.99' *as* {*}decimal{*}(20,8)){*}/{*}100
*as* amt1,{*}cast{*}('256.99' *as* {*}decimal{*}(20,8)) *as* amt2;{color}
*will got error :*
org.apache.spark.{*}sql{*}.catalyst.expressions.Literal cannot be *cast* *to*
org.apache.spark.{*}sql{*}.catalyst.expressions.AnsiCast
The SQL will need to change to :
{color:#4c9aff}explain{color}
{color:#4c9aff}*INSERT* OVERWRITE *TABLE* test.spark33_decimal_orc{color}
{color:#4c9aff}*select* *null* *as* amt1,{*}cast{*}('256.99' *as*
{*}decimal{*}(20,8)) *as* amt2{color}
{color:#4c9aff}*union* *all*{color}
{color:#4c9aff}*select* {color:#de350b}{*}cast{*}({color}{*}cast{*}('200.99'
*as* {*}decimal{*}(20,8)){*}/{*}100 *as*
{*}decimal{*}(20,8){color:#de350b}){color} *as* amt1,{*}cast{*}('256.99' *as*
{*}decimal{*}(20,8)) *as* amt2;{color}
*but this is not need in spark3.2.1 , is this a bug for spark3.3.1 ?*
was:
*when 'union all' and one select statement use* *Literal as column value , the
other* *select statement has computed expression at the same column , then the
whole statement will compile failed. A explicit cast will be needed.*
for example:
explain
*INSERT* OVERWRITE *TABLE* test.spark33_decimal_orc
*select* *null* *as* amt1, *cast*('256.99' *as* *decimal*(20,8)) *as* amt2
*union* *all*
*select* *cast*('200.99' *as* *decimal*(20,8)){*}/{*}100 *as*
amt1,*cast*('256.99' *as* *decimal*(20,8)) *as* amt2
*will got error :*
org.apache.spark.*sql*.catalyst.expressions.Literal cannot be *cast* *to*
org.apache.spark.*sql*.catalyst.expressions.AnsiCast
The SQL will need to change to :
explain
*INSERT* OVERWRITE *TABLE* test.spark33_decimal_orc
*select* *null* *as* amt1,*cast*('256.99' *as* *decimal*(20,8)) *as* amt2
*union* *all*
*select* {*}cast{*}({*}cast{*}('200.99' *as* {*}decimal{*}(20,8)){*}/{*}100
*as* {*}decimal{*}(20,8)) *as* amt1,*cast*('256.99' *as* *decimal*(20,8)) *as*
amt2
but this is not need in spark3.2.1 , is this a bug for spark3.3.1 ?
> An explicit cast will be needed when INSERT OVERWRITE SELECT UNION ALL
> ----------------------------------------------------------------------
>
> Key: SPARK-42473
> URL: https://issues.apache.org/jira/browse/SPARK-42473
> Project: Spark
> Issue Type: Bug
> Components: Optimizer
> Affects Versions: 3.3.1
> Environment: spark 3.3.1
> Reporter: kevinshin
> Priority: Major
>
> *when 'union all' and one select statement use* *Literal as column value ,
> the other* *select statement has computed expression at the same column ,
> then the whole statement will compile failed. A explicit cast will be needed.*
> for example:
> {color:#4c9aff}explain{color}
> {color:#4c9aff}*INSERT* OVERWRITE *TABLE* test.spark33_decimal_orc{color}
> {color:#4c9aff}*select* *null* *as* amt1, {*}cast{*}('256.99' *as*
> {*}decimal{*}(20,8)) *as* amt2{color}
> {color:#4c9aff}*union* *all*{color}
> {color:#4c9aff}*select* {*}cast{*}('200.99' *as*
> {*}decimal{*}(20,8)){*}/{*}100 *as* amt1,{*}cast{*}('256.99' *as*
> {*}decimal{*}(20,8)) *as* amt2;{color}
> *will got error :*
> org.apache.spark.{*}sql{*}.catalyst.expressions.Literal cannot be *cast* *to*
> org.apache.spark.{*}sql{*}.catalyst.expressions.AnsiCast
> The SQL will need to change to :
> {color:#4c9aff}explain{color}
> {color:#4c9aff}*INSERT* OVERWRITE *TABLE* test.spark33_decimal_orc{color}
> {color:#4c9aff}*select* *null* *as* amt1,{*}cast{*}('256.99' *as*
> {*}decimal{*}(20,8)) *as* amt2{color}
> {color:#4c9aff}*union* *all*{color}
> {color:#4c9aff}*select* {color:#de350b}{*}cast{*}({color}{*}cast{*}('200.99'
> *as* {*}decimal{*}(20,8)){*}/{*}100 *as*
> {*}decimal{*}(20,8){color:#de350b}){color} *as* amt1,{*}cast{*}('256.99' *as*
> {*}decimal{*}(20,8)) *as* amt2;{color}
>
> *but this is not need in spark3.2.1 , is this a bug for spark3.3.1 ?*
--
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
