[
https://issues.apache.org/jira/browse/HIVE-6143?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13862826#comment-13862826
]
Timothy Chen commented on HIVE-6143:
------------------------------------
I can refactor the hive code but would like to get some feedback if this can be
an accepted approach,
> Refactor Orc file format parsing logic to be shared
> ---------------------------------------------------
>
> Key: HIVE-6143
> URL: https://issues.apache.org/jira/browse/HIVE-6143
> Project: Hive
> Issue Type: Bug
> Reporter: Timothy Chen
>
> Currently the Orc file format parsing logic is hidden in private methods in
> reader and record reader classes, for example footer parsing, stream loading,
> etc.
> For the Orc file format to be a more reusable file format outside of Hive, I
> suggest refactor these generic logic into a shared class.
> The current interface of reading per serialized as objects is not suffice as
> for columnar execution engines such as Drill/Impala, it's much more efficient
> to load in columnar data into its own columnar in memory formats.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
