hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Branky Shao (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-6565) OrcSerde should be added as NativeSerDe in SerDeUtils
Date Fri, 07 Mar 2014 02:56:45 GMT

    [ https://issues.apache.org/jira/browse/HIVE-6565?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13923468#comment-13923468
] 

Branky Shao commented on HIVE-6565:
-----------------------------------

Xuefu Zhang thanks for following up. I got few problems when I work on Cascading and Hive/HCatalog
integration(https://github.com/branky/cascading.hive). They can be fixed along with HIVE-6163
(was also reported by me) and HIVE-5728. I did patch on my own fork of Hive(https://github.com/branky/hive/commit/20c62b3ec73a55d84eeffca6eab4af066e04e22a)
to make the integration work. 

Regarding this issue, I found the fix may not be that simple. OrcSerde is included in ql module
which depend on serde module actually. Then it will cause cyclic dependency. I would cancel
this issue and try to find other workaround.  

> OrcSerde should be added as NativeSerDe in SerDeUtils
> -----------------------------------------------------
>
>                 Key: HIVE-6565
>                 URL: https://issues.apache.org/jira/browse/HIVE-6565
>             Project: Hive
>          Issue Type: Bug
>          Components: Serializers/Deserializers
>    Affects Versions: 0.12.0
>            Reporter: Branky Shao
>
> If the table defined as ORC format, the columns info can be fetched from StorageDescriptor,
no need to get from SerDe.
> And obviously, ORC is a Hive's native file format, therefore, OrcSerde should be added
as NativeSerDe in SerDeUtils.
> The fix is fairly simple, just add single line in SerDeUtils :
> nativeSerDeNames.add(org.apache.hadoop.hive.ql.io.orc.OrcSerde.class.getName()); 



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

Mime
View raw message