[
https://issues.apache.org/jira/browse/HIVE-13562?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15250906#comment-15250906
]
Ashutosh Chauhan commented on HIVE-13562:
-----------------------------------------
I think it will be good idea because now udfs may get inserted into operator
pipeline even when user has not specified any in her query text. e.g. 1) In
security enabled deployment, we may insert udfs for column masking. 2) When
column stats autogather is on, stats collector udf will show up in pipeline.
So, I think its better to always have vectorization on.
> Enable vector bridge for all non-vectorized udfs
> ------------------------------------------------
>
> Key: HIVE-13562
> URL: https://issues.apache.org/jira/browse/HIVE-13562
> Project: Hive
> Issue Type: Improvement
> Components: Vectorization
> Reporter: Ashutosh Chauhan
>
> Mechanism already exists for this via {{VectorUDFAdaptor}} but we have
> arbitrarily hand picked few udfs to go through it. I think we should enable
> this by default for all udfs.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
