[
https://issues.apache.org/jira/browse/ORC-17?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14600401#comment-14600401
]
Demai Ni commented on ORC-17:
-----------------------------
it would be great to see the code will also support
[libhdfs|https://wiki.apache.org/hadoop/LibHDFS] from hadoop, and
[libhdfs3|https://github.com/PivotalRD/libhdfs3] from pivotal. By theory, as
long as they share the same api interface defined by hdfs.h. The proposed
plug-in module will support them, right?
many thanks for bringing up this jira.
> Support HDFS as a plug-in module
> --------------------------------
>
> Key: ORC-17
> URL: https://issues.apache.org/jira/browse/ORC-17
> Project: Orc
> Issue Type: Improvement
> Components: C++
> Reporter: Charles Pritchard
>
> Continuing from: https://github.com/hortonworks/orc/issues/149
> There are C/C++ modules for reading from and writing to HDFS. There (may be)
> additional use cases for HTTP-based protocols like WebHDFS and S3.
> As omalley states: "The C++ API has a pluggable interface for where to get
> the bytes. It already has a binding the local file system and it would be
> easy to extend it to other file systems or object stores.".
> It would be nice to see a build of C++ ORC which includes HDFS support, as a
> build flag or otherwise provided as a module/separate repository.
> I'd also like to see WebHDFS/S3 supported, once that baseline has been met.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
