[
https://issues.apache.org/jira/browse/HIVE-11973?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14969231#comment-14969231
]
Yongzhi Chen commented on HIVE-11973:
-------------------------------------
[~xuefuz], the equal one has this stacktrace, it uses
getCommonClassForComparison if need convert:
FunctionRegistry.implicitConvertible(PrimitiveObjectInspector$PrimitiveCategory,
PrimitiveObjectInspector$PrimitiveCategory) line: 890
FunctionRegistry.getCommonClassForComparison(TypeInfo, TypeInfo) line: 763
GenericUDFOPEqual(GenericUDFBaseCompare).initialize(ObjectInspector[]) line:
138
GenericUDFOPEqual(GenericUDF).initializeAndFoldConstants(ObjectInspector[])
line: 139
ExprNodeGenericFuncDesc.newInstance(GenericUDF, String, List<ExprNodeDesc>)
line: 234
TypeCheckProcFactory$DefaultExprProcessor.getXpathOrFuncExprNodeDesc(ASTNode,
boolean, ArrayList<ExprNodeDesc>, TypeCheckCtx) line: 1082
TypeCheckProcFactory$DefaultExprProcessor.process(Node, Stack<Node>,
NodeProcessorCtx, Object...) line: 1301
DefaultRuleDispatcher.dispatch(Node, Stack<Node>, Object...) line: 90
The logic of getCommonClassForComparison very similar to
getCommonClassForUnionAll.
What I am trying to say in the previous comment is: When only consider type
conversion, x will need compare each value, that means it can unionall with
each value too. The ReturnObjectInspectorResolver.update is just to find out
which type will be used for the conversion that x can compare with each value
in the set.
> IN operator fails when the column type is DATE
> -----------------------------------------------
>
> Key: HIVE-11973
> URL: https://issues.apache.org/jira/browse/HIVE-11973
> Project: Hive
> Issue Type: Bug
> Components: Query Processor
> Affects Versions: 1.0.0
> Reporter: sanjiv singh
> Assignee: Yongzhi Chen
> Attachments: HIVE-11973.1.patch, HIVE-11973.2.patch
>
>
> Test DLL :
> {code}
> CREATE TABLE `date_dim`(
> `d_date_sk` int,
> `d_date_id` string,
> `d_date` date,
> `d_current_week` string,
> `d_current_month` string,
> `d_current_quarter` string,
> `d_current_year` string) ;
> {code}
> Hive query :
> {code}
> SELECT *
> FROM date_dim
> WHERE d_date IN ('2000-03-22','2001-03-22') ;
> {code}
> In 1.0.0 , the above query fails with:
> {code}
> FAILED: SemanticException [Error 10014]: Line 1:180 Wrong arguments
> ''2001-03-22'': The arguments for IN should be the same type! Types are:
> {date IN (string, string)}
> {code}
> I changed the query as given to pass the error :
> {code}
> SELECT *
> FROM date_dim
> WHERE d_date IN (CAST('2000-03-22' AS DATE) , CAST('2001-03-22' AS DATE)
> ) ;
> {code}
> But it works without casting :
> {code}
> SELECT *
> FROM date_dim
> WHERE d_date = '2000-03-22' ;
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
