hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Hari Sankar Sivarama Subramaniyan (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-12666) PCRExprProcFactory.GenericFuncExprProcessor.process() aggressively removes dynamic partition pruner generated synthetic join predicates.
Date Tue, 15 Dec 2015 00:36:46 GMT

    [ https://issues.apache.org/jira/browse/HIVE-12666?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15057061#comment-15057061
] 

Hari Sankar Sivarama Subramaniyan commented on HIVE-12666:
----------------------------------------------------------

This includes :
1. Revert HIVE-12462 (change in 
2. Correct fix made in PcrExprProcFactory.
3. Merged diff files 

cc-ing [~hagleitn], [~sershe], [~gopalv]

[~jpullokkaran] Can you please review this patch

Thanks
Hari

> PCRExprProcFactory.GenericFuncExprProcessor.process() aggressively removes dynamic partition
pruner generated synthetic join predicates.
> ----------------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: HIVE-12666
>                 URL: https://issues.apache.org/jira/browse/HIVE-12666
>             Project: Hive
>          Issue Type: Bug
>            Reporter: Hari Sankar Sivarama Subramaniyan
>            Assignee: Hari Sankar Sivarama Subramaniyan
>            Priority: Blocker
>         Attachments: HIVE-12666.1.patch
>
>
> Introduced by HIVE-11634. The original idea in HIVE-11634 was to remove the IN partition
conditions from the predicate list since the static dynamic partitioning would kick in and
push these predicates down to metastore. However, the check is too aggressive and removes
events such as below :
> {code}
> -                    Select Operator
> -                      expressions: UDFToDouble(UDFToInteger((hr / 2))) (type: double)
> -                      outputColumnNames: _col0
> -                      Statistics: Num rows: 1 Data size: 7 Basic stats: COMPLETE Column
stats: NONE
> -                      Group By Operator
> -                        keys: _col0 (type: double)
> -                        mode: hash
> -                        outputColumnNames: _col0
> -                        Statistics: Num rows: 1 Data size: 7 Basic stats: COMPLETE Column
stats: NONE
> -                        Dynamic Partitioning Event Operator
> -                          Target Input: srcpart
> -                          Partition key expr: UDFToDouble(hr)
> -                          Statistics: Num rows: 1 Data size: 7 Basic stats: COMPLETE
Column stats: NONE
> -                          Target column: hr
> -                          Target Vertex: Map 1
> {code}



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

Mime
View raw message