[jira] [Commented] (HIVE-11839) Vectorization wrong results with filter of (CAST AS CHAR)

["Sergey Shelukhin (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-11839?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14790886#comment-14790886 ] 

Sergey Shelukhin commented on HIVE-11839:
-----------------------------------------

+1. Does it also affect 1.2, 1.3 etc.? It should be backported accordingly

> Vectorization wrong results with filter of (CAST AS CHAR)
> ---------------------------------------------------------
>
>                 Key: HIVE-11839
>                 URL: https://issues.apache.org/jira/browse/HIVE-11839
>             Project: Hive
>          Issue Type: Bug
>          Components: Hive
>            Reporter: Matt McCline
>            Assignee: Matt McCline
>            Priority: Critical
>         Attachments: HIVE-11839.01.patch
>
>
> PROBLEM:
> For query such as
> select count(1) from table where CAST (id as CHAR(4))='1000';
> gives wrong results 0 than expected results.
> STEPS TO REPRODUCE:
> create table s1(id smallint) stored as orc;
> insert into table s1 values (1000),(1001),(1002),(1003),(1000);
> set hive.vectorized.execution.enabled=true;
> select count(1) from s1 where cast(id as char(4))='1000';
> – this gives 0
> set hive.vectorized.execution.enabled=false;
> select count(1) from s1 where cast(id as char(4))='1000';
> – this gives 2



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)