[jira] [Commented] (HIVE-7189) Hive does not store column names in ORC

["Eric Sun (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-7189?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15235823#comment-15235823 ] 

Eric Sun commented on HIVE-7189:
--------------------------------

A temporary solution is to use org.apache.hive.hcatalog.pig.HCatLoader() instead of OrcStorage(). But it will be great to store the real field names in ORC data files.

* the field names inside nested structures are preserved even if we generate the ORC via Hive

> Hive does not store column names in ORC
> ---------------------------------------
>
>                 Key: HIVE-7189
>                 URL: https://issues.apache.org/jira/browse/HIVE-7189
>             Project: Hive
>          Issue Type: Bug
>          Components: File Formats, ORC
>    Affects Versions: 0.12.0, 0.13.0
>            Reporter: Chris Drome
>            Assignee: Daniel Dai
>
> We uncovered the following discrepancy between writing ORC files through Pig and Hive:
> ORCFile header contains the name of the columns. Storing through Pig (ORCStorage or HCatStorer), the column names are stored fine. But when stored through hive they are stored as _col0, _col1,....,_col99 and hive uses the partition schema to map the column names. Reading the same file through Pig then has problems as user will have to manually map columns.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)