hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jesus Camacho Rodriguez (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-15209) Set hive.strict.checks.cartesian.product to false by default
Date Tue, 15 Nov 2016 18:15:58 GMT

    [ https://issues.apache.org/jira/browse/HIVE-15209?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15667850#comment-15667850
] 

Jesus Camacho Rodriguez commented on HIVE-15209:
------------------------------------------------

[~xuefuz], OK, then it does not violate compliance... but I would argue it is better to disable
it by default and let the admin decide whether users should be able to execute cartesian products
or not. That would give us better tests coverage and the possibility to run randomly a cartesian
product if we want to.

Btw, cartesian product does not need to be specified explicitly; it might be produced by the
Calcite optimizer too e.g. if we can prune with a filter on a constant equality both inputs
of a join.

> Set hive.strict.checks.cartesian.product to false by default
> ------------------------------------------------------------
>
>                 Key: HIVE-15209
>                 URL: https://issues.apache.org/jira/browse/HIVE-15209
>             Project: Hive
>          Issue Type: Bug
>          Components: Parser
>    Affects Versions: 2.2.0
>            Reporter: Jesus Camacho Rodriguez
>            Assignee: Jesus Camacho Rodriguez
>         Attachments: HIVE-15209.patch
>
>
> If we aim to make Hive compliant with SQL, w should disable this property by default,
as expressing a cartesian product, though inefficient, is perfectly valid in SQL.
> Further, if we express complex predicates in the ON clause of a SQL query, we might not
be able to push these predicates to the join operator; however, we should still be able to
execute the query.



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

Mime
View raw message