[jira] [Commented] (PARQUET-428) Support INT96 and FIXED_LEN_BYTE_ARRAY types

["Ryan Blue (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/PARQUET-428?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15102254#comment-15102254 ] 

Ryan Blue commented on PARQUET-428:
-----------------------------------

[~mdeepak], thanks for taking this on! You're right about the Hive mappings so you'll need this to support DECIMAL and the current TIMESTAMP. A couple of things to keep in mind on that eventual goal: first, while Hive maps decimal to an underlying fixed, there are several primitive types allowed by the spec that need to be supported when it comes time. I'm happy to help when it comes to ensuring compatibility so just open up a JIRA and ping me. Second, the int96 timestamp type has a lot of problems that I'm trying to fix in various processing engines. Again, when it comes time to implement we can help you through the compatibility story.

> Support INT96 and FIXED_LEN_BYTE_ARRAY types
> --------------------------------------------
>
>                 Key: PARQUET-428
>                 URL: https://issues.apache.org/jira/browse/PARQUET-428
>             Project: Parquet
>          Issue Type: Improvement
>          Components: parquet-cpp
>            Reporter: Deepak Majeti
>
> I would like to add support for INT96 and FIXED_LEN_BYTE_ARRAY parquet types.
> Hive data types DATE and TIMESTAMP get mapped to INT96 parquet type.
> Hive DECIMAL gets mapped to parquet FIXED_LEN_BYTE_ARRAY type.



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