[ 
https://issues.apache.org/jira/browse/FLINK-21520?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17301521#comment-17301521
 ] 

Etienne Chauchot edited comment on FLINK-21520 at 3/15/21, 9:13 AM:
--------------------------------------------------------------------

True, that is more of a guess as I could not compare with a missing feature  :) 
but I'm benchmarking TPC-DS query3 using parquet and a DataSet based pipeline 
and I'm getting very good response times with fileter operator.  Pushed down 
filter should be even better but filter operator does not seem to slow the 
pipeline down too much.


was (Author: echauchot):
It is true that I could not compare with a missing feature  :) but I'm 
benchmarking TPC-DS query3 using parquet and a DataSet based pipeline and I'm 
getting very good response times with fileter operator.  Pushed down filter 
should be even better but filter operator does not seem to slow the pipeline 
down.

> ParquetInputFormat#setfilterPredicate() does not work
> -----------------------------------------------------
>
>                 Key: FLINK-21520
>                 URL: https://issues.apache.org/jira/browse/FLINK-21520
>             Project: Flink
>          Issue Type: Bug
>            Reporter: Etienne Chauchot
>            Priority: Major
>
> Simplified code:
> {code:java}
>  FilterPredicate filterPredicate = eq(intColumn("intField"), 10);
> parquetInputFormat.setFilterPredicate(filterPredicate);
> env.createInput(parquetInputFormat);
> {code}
>    
> produces no records whereas 
> {code:java}
> env.createInput(parquetInputFormat)
> .filter((FilterFunction) value ->  value.get("intField") == 10);
> {code}
> produces records.



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

Reply via email to