[
https://issues.apache.org/jira/browse/PARQUET-201?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14482079#comment-14482079
]
Ryan Blue commented on PARQUET-201:
-----------------------------------
This isn't a regression because the filter2 API and INT_* types were introduced
since 1.5.0. But it is a known bug with the new code and should be an easy fix.
I don't mind not including this since it is so close to RC.
> Column with OriginalType INT_8 failed at filtering
> --------------------------------------------------
>
> Key: PARQUET-201
> URL: https://issues.apache.org/jira/browse/PARQUET-201
> Project: Parquet
> Issue Type: Bug
> Components: parquet-mr
> Reporter: Low Chin Wei
> Fix For: 1.6.0
>
>
> I hit the following error when apply a filter predicate on a column with
> PrimitiveType = INT32 & OriginalType equal to INT_8.
> java.lang.IllegalArgumentException: FilterPredicate column: c's declared type
> (java.lang.Integer) does not match the schema found in file metadata. Column
> c is of type: FullTypeDescriptor(PrimitiveType: INT32, OriginalType: INT_8)
> Valid types for this column are: null
> at
> parquet.filter2.predicate.ValidTypeMap.assertTypeValid(ValidTypeMap.java:114)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
