[
https://issues.apache.org/jira/browse/HIVE-11428?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14653260#comment-14653260
]
Gopal V commented on HIVE-11428:
--------------------------------
[~hsubramaniyan]: this patch took the query from ~2100 secs to 98s.
The only log spew is from the constant folding impl
{code}
2015-08-04 04:17:01,434 ERROR [main]: optimizer.ConstantPropagateProcFactory
(ConstantPropagateProcFactory.java:evaluateFunction(768)) - Unable to evaluate
org.apache.hadoop.hive.ql.u
df.generic.GenericUDFStruct@777e2e27. Return value unrecoginizable.
{code}
> Performance: Struct IN() clauses are extremely slow (~10x slower)
> ------------------------------------------------------------------
>
> Key: HIVE-11428
> URL: https://issues.apache.org/jira/browse/HIVE-11428
> Project: Hive
> Issue Type: Bug
> Components: Query Processor
> Affects Versions: 1.3.0, 2.0.0
> Reporter: Gopal V
> Assignee: Hari Sankar Sivarama Subramaniyan
> Attachments: HIVE-11428.1.patch
>
>
> Hive today does not support tuple IN() clauses today, but provides a way to
> rewrite (a,b) IN (...) using complex types.
> select * from table where STRUCT(a,b) IN (STRUCT(1,2), STRUCT(2,3) ...);
> This would be fine, except it is massively slower due to ObjectConvertors and
> Struct constructor not being constant folded.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
