[
https://issues.apache.org/jira/browse/HIVE-533?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12715708#action_12715708
]
Prasad Chakka commented on HIVE-533:
------------------------------------
we didn't push rand() but we pushed all other predicates in the clause. so
rand() is getting applied after rows are filtered. the current behavior seems
to be rand() gets applied before rows are filtered. not sure if there is a ANSI
semantics, but this fix keeps the current behavior of Hive even when PPD is
applied.
> predicate clauses with rand() should not be pushed down
> -------------------------------------------------------
>
> Key: HIVE-533
> URL: https://issues.apache.org/jira/browse/HIVE-533
> Project: Hadoop Hive
> Issue Type: Bug
> Components: Query Processor
> Affects Versions: 0.4.0
> Reporter: Prasad Chakka
> Assignee: Prasad Chakka
>
> in the following query, all predicates except the rand() is pushed down where
> in rand() is applied after filter which should not be the case.
> select a.* from srcpart a where rand(1) < 0.1 and a.ds = '2008-04-08' and
> not(key > 50 or key < 10) and a.hr like '%2';
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
