[
https://issues.apache.org/jira/browse/SPARK-9147?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Josh Rosen resolved SPARK-9147.
-------------------------------
Resolution: Fixed
Fix Version/s: 1.5.0
Issue resolved by pull request 7194
[https://github.com/apache/spark/pull/7194]
> UnsafeRow should canonicalize NaN values
> ----------------------------------------
>
> Key: SPARK-9147
> URL: https://issues.apache.org/jira/browse/SPARK-9147
> Project: Spark
> Issue Type: Sub-task
> Components: SQL
> Reporter: Reynold Xin
> Assignee: Josh Rosen
> Fix For: 1.5.0
>
>
> NaN has many different representations in raw bytes.
> When we set a double/float value, we should check whether it is NaN, and a
> binary representation that is canonicalized, so we can do comparison on bytes
> directly.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
