You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hawq.apache.org by "Lili Ma (JIRA)" <ji...@apache.org> on 2016/06/17 03:41:05 UTC
[jira] [Commented] (HAWQ-831) Re-implementation for some data types
of Parquet in HAWQ
[ https://issues.apache.org/jira/browse/HAWQ-831?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15335342#comment-15335342 ]
Lili Ma commented on HAWQ-831:
------------------------------
If we re-implement some data types in HAWQ, the Hive generated parquet files for these data types can be directly registered into HAWQ.
> Re-implementation for some data types of Parquet in HAWQ
> --------------------------------------------------------
>
> Key: HAWQ-831
> URL: https://issues.apache.org/jira/browse/HAWQ-831
> Project: Apache HAWQ
> Issue Type: Task
> Components: Storage
> Reporter: Lili Ma
> Assignee: Lei Chang
>
> Re-implement some HAWQ data type mapping to Parquet, so that HAWQ can be compatible with Hive.
> 1. Currently HAWQ converts data type decimal to byte array, we can modify it to fixed length byte array, so that HAWQ can be compatible with Hive.
> 2. HAWQ itself can convert the data type char to int32 instead of byte array, so that the space for char storage can be saved.
> 3. To be compatible with Hive, HAWQ can change its type mapping for decimal to int96, but this need discussion.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)