[
https://issues.apache.org/jira/browse/HIVE-18531?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16339046#comment-16339046
]
Teddy Choi commented on HIVE-18531:
-----------------------------------
+1 with pending tests.
I tried this patch with ptf.* q files and all of them passed.
> Vectorization: Vectorized PTF operator should not set the initial type infos
> ----------------------------------------------------------------------------
>
> Key: HIVE-18531
> URL: https://issues.apache.org/jira/browse/HIVE-18531
> Project: Hive
> Issue Type: Bug
> Components: Hive
> Affects Versions: 3.0.0
> Reporter: Matt McCline
> Priority: Critical
> Attachments: HIVE-18531.01.patch
>
>
> The Vectorized PTF operator is mistakenly setting the initial type infos for
> its output VectorizationContext. It should not. It is only creating a
> projection of the initial column names/type infos from ReduceSink (i.e. keys,
> values) plus scratch columns for output columns for which the type infos are
> already known.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
