You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hive.apache.org by "Owen O'Malley (JIRA)" <ji...@apache.org> on 2013/11/19 20:27:25 UTC
[jira] [Commented] (HIVE-5663) Refactor ORC RecordReader to operate
on direct & wrapped ByteBuffers
[ https://issues.apache.org/jira/browse/HIVE-5663?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13826839#comment-13826839 ]
Owen O'Malley commented on HIVE-5663:
-------------------------------------
This looks good, but a few comments:
* Please remove the code for OutStream.HEADER_SIZE > 3, since it is always false.
* Please remove the offsets array from the interface since we are switching to ByteBuffer.
Other than that, it looks great.
> Refactor ORC RecordReader to operate on direct & wrapped ByteBuffers
> --------------------------------------------------------------------
>
> Key: HIVE-5663
> URL: https://issues.apache.org/jira/browse/HIVE-5663
> Project: Hive
> Issue Type: Improvement
> Components: File Formats
> Affects Versions: 0.13.0
> Environment: Ubuntu LXC
> Reporter: Gopal V
> Assignee: Gopal V
> Labels: ORC
> Attachments: HIVE-5663.01.patch, HIVE-5663.02.patch
>
>
> The current ORC RecordReader implementation assumes array structures backing the ByteBuffers it passes around between RecordReaderImpl and Compressed/Uncompressed InStream objects.
> This patch attempts to refactor those assumptions out of both classes, allowing the future use of direct byte buffers within ORC (as might come from HDFS zero-copy readers).
--
This message was sent by Atlassian JIRA
(v6.1#6144)