[jira] [Commented] (HIVE-7082) Vectorized parquet reader should create assigners only for the columns it assigns, not for scratch columns

["Ashutosh Chauhan (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-7082?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14001399#comment-14001399 ] 

Ashutosh Chauhan commented on HIVE-7082:
----------------------------------------

LGTM +1 [~jnp] Would you like to take a look too?

> Vectorized parquet reader should create assigners only for the columns it assigns, not for scratch columns
> ----------------------------------------------------------------------------------------------------------
>
>                 Key: HIVE-7082
>                 URL: https://issues.apache.org/jira/browse/HIVE-7082
>             Project: Hive
>          Issue Type: Bug
>          Components: Query Processor
>    Affects Versions: 0.13.0, 0.14.0
>            Reporter: Remus Rusanu
>            Assignee: Remus Rusanu
>            Priority: Minor
>              Labels: parquet, vectorization
>         Attachments: HIVE-7082.1.patch
>
>
> The code in VectorizedParquetRecordReader.next and in VectorColumnAssignFactory.buildAssigners iterates for all columns in the vectorization context trying to build an assigner for it. Scratch columns do not require assigners. Fix is to simply use the writables.length instead of outputBatch.numCols.



--
This message was sent by Atlassian JIRA
(v6.2#6252)