[
https://issues.apache.org/jira/browse/HIVE-18291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16810755#comment-16810755
]
Ivan Suller commented on HIVE-18291:
------------------------------------
I'd vote against this change as well. At least the default should be the
current behaviour. When I used Hive daily we depended on this behaviour very
often: if an input is not what we expect we simply want it to be ignored. Still
better to have one row incomplete than have a billion rows missing.
> An exception should be raised if the result is outside the range of decimal
> ---------------------------------------------------------------------------
>
> Key: HIVE-18291
> URL: https://issues.apache.org/jira/browse/HIVE-18291
> Project: Hive
> Issue Type: Sub-task
> Components: SQL
> Affects Versions: 2.3.0
> Reporter: Marco Gaido
> Priority: Major
>
> Citing SQL:2011 on page 27 available at
> http://standards.iso.org/ittf/PubliclyAvailableStandards/c053681_ISO_IEC_9075-1_2011.zip:
> {noformat}
> If the result cannot be represented exactly in the result type, then whether
> it is rounded
> or truncated is implementation-defined. An exception condition is raised if
> the result is
> outside the range of numeric values of the result type, or if the arithmetic
> operation
> is not defined for the operands.
> {noformat}
> Currently Hive is returning NULL instead of throwing an exception if the
> result is out of range, eg.:
> {code}
> > select 1000000000000000000.000001*1000000000000000000.000001;
> +-------+
> | _c0 |
> +-------+
> | NULL |
> +-------+
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
