You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@flink.apache.org by "fa zheng (Jira)" <ji...@apache.org> on 2019/09/20 08:16:00 UTC

[jira] [Commented] (FLINK-13933) Hive Generic UDTF can not be used in table API both stream and batch mode

    [ https://issues.apache.org/jira/browse/FLINK-13933?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16934188#comment-16934188 ] 

fa zheng commented on FLINK-13933:
----------------------------------

It is not a problem. If you want to use hive udtf in flink sql, you must use by lateral way, e.g.

SELECT a from table_name, LATERAL TABLE(func(s)) as T(a)

> Hive Generic UDTF can not be used in table API both stream and batch mode
> -------------------------------------------------------------------------
>
>                 Key: FLINK-13933
>                 URL: https://issues.apache.org/jira/browse/FLINK-13933
>             Project: Flink
>          Issue Type: Bug
>          Components: Table SQL / Planner
>    Affects Versions: 1.9.0
>            Reporter: fa zheng
>            Priority: Major
>             Fix For: 1.9.1
>
>   Original Estimate: 12h
>  Remaining Estimate: 12h
>
> When i use Hive Generic UDTF in table API,  something wrong with CalciteContextException: No match found for function signature . As mentioned in tutorials, Hive’s GenericUDTF is automatically translated into Flink’s TableFunction. However, it didn't give a property of TABLE_FUNCTION. In convertToSqlFunction, when functionDefinition is instanceof TableFunctionDefinition, it will further call category.isTableFunction() and lead a wrong answer.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)