[jira] [Commented] (HIVE-28830) Fix HIVE-25043 Support custom UDF in Vectorized mode

["Ryu Kobayashi (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-28830?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17936732#comment-17936732 ] 

Ryu Kobayashi commented on HIVE-28830:
--------------------------------------

I created PR: [https://github.com/apache/hive/pull/5703]

> Fix HIVE-25043 Support custom UDF in Vectorized mode
> ----------------------------------------------------
>
>                 Key: HIVE-28830
>                 URL: https://issues.apache.org/jira/browse/HIVE-28830
>             Project: Hive
>          Issue Type: Improvement
>          Components: Query Processor
>            Reporter: Ryu Kobayashi
>            Assignee: Ryu Kobayashi
>            Priority: Major
>              Labels: pull-request-available
>         Attachments: explain1.txt, explain2.txt, explain3.txt, query.sql
>
>
> I previously created HIVE-25043, and it was merged, but it turns out this is not working.
>  * [https://github.com/apache/hive/blob/rel/release-4.0.1/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorizationContext.java#L1439-L1445]
>  * [https://github.com/apache/hive/blob/rel/release-4.0.1/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorizationContext.java#L1461-L1469]
> As mentioned in the comments, it will not be effective even if I register a custom UDF for vectorize here. I created a custom UDF called TIME_PARSE and ran the [^query.sql], and the results changed to the following:
>  * Default:
>  ** Enable vectorized: Map 1, Map 3, Map 8, Map 9 -> [^explain1.txt]
>  * Current implementation:
>  ** Enable vectorized: Map 1, Map 3, Map 8, Map 9 -> [^explain2.txt]
>  * New PR:
>  ** Enable vectorized: Map 1, Map 3, Map 5, Map 6, Map 8, Map 9, Map 11, Map 12 -> [^explain3.txt]



--
This message was sent by Atlassian Jira
(v8.20.10#820010)