[jira] [Commented] (HIVE-13287) Add logic to estimate stats for IN operator

["Ashutosh Chauhan (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-13287?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15197555#comment-15197555 ] 

Ashutosh Chauhan commented on HIVE-13287:
-----------------------------------------

Unnesting IN into ORs for stats estimation may be unnecessary CPU cycles. If possible, we should avoid conversion of IN into OR and try to do stats estimation on IN itself.

> Add logic to estimate stats for IN operator
> -------------------------------------------
>
>                 Key: HIVE-13287
>                 URL: https://issues.apache.org/jira/browse/HIVE-13287
>             Project: Hive
>          Issue Type: Bug
>          Components: Statistics
>    Affects Versions: 2.1.0
>            Reporter: Jesus Camacho Rodriguez
>            Assignee: Jesus Camacho Rodriguez
>         Attachments: HIVE-13287.patch
>
>
> Currently, IN operator is considered in the default case: reduces the input rows to the half. This may lead to wrong estimates for the number of rows produced by Filter operators.



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