[
https://issues.apache.org/jira/browse/SPARK-23549?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16391282#comment-16391282
]
zhoukang commented on SPARK-23549:
----------------------------------
I think this is a bug.Which may caused by rule below:
{code:java}
case p @ BinaryComparison(left, right)
if findCommonTypeForBinaryComparison(left.dataType,
right.dataType).isDefined =>
val commonType = findCommonTypeForBinaryComparison(left.dataType,
right.dataType).get
p.makeCopy(Array(castExpr(left, commonType), castExpr(right,
commonType)))
{code}
findCommonTypeForBinaryComparison will return StringType:
{code:java}
case (DateType, TimestampType) => Some(StringType)
{code}
May be we can add a new rule for this case?
> Spark SQL unexpected behavior when comparing timestamp to date
> --------------------------------------------------------------
>
> Key: SPARK-23549
> URL: https://issues.apache.org/jira/browse/SPARK-23549
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Affects Versions: 2.2.1
> Reporter: Dong Jiang
> Priority: Major
>
> {code:java}
> scala> spark.version
> res1: String = 2.2.1
> scala> spark.sql("select cast('2017-03-01 00:00:00' as timestamp) between
> cast('2017-02-28' as date) and cast('2017-03-01' as date)").show
> +---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
> |((CAST(CAST(2017-03-01 00:00:00 AS TIMESTAMP) AS STRING) >=
> CAST(CAST(2017-02-28 AS DATE) AS STRING)) AND (CAST(CAST(2017-03-01 00:00:00
> AS TIMESTAMP) AS STRING) <= CAST(CAST(2017-03-01 AS DATE) AS STRING)))|
> +---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
> |
>
> false|
> +---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+{code}
> As shown above, when a timestamp is compared to date in SparkSQL, both
> timestamp and date are downcast to string, and leading to unexpected result.
> If run the same SQL in presto/Athena, I got the expected result
> {code:java}
> select cast('2017-03-01 00:00:00' as timestamp) between cast('2017-02-28' as
> date) and cast('2017-03-01' as date)
> _col0
> 1 true
> {code}
> Is this a bug for Spark or a feature?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
