[jira] [Commented] (TEZ-1031) Support Multiple Inputs/Outputs in MapProcessor / ReduceProcessor

["Hitesh Shah (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/TEZ-1031?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13964366#comment-13964366 ] 

Hitesh Shah commented on TEZ-1031:
----------------------------------

[~miloveme] Would you mind first starting a thread on user@tez.incubator.apache.org explaining what your use case is? 

MapProcessor and ReduceProcessor are for MR jobs. However, the Mapper and Reducer APIs do not support multiple inputs/outputs. Hence, it probably does not make sense to add such support into the processors. Let's start a discussion on the mailing list where we can guide you to a better solution. 

For now, marking this as Wont-fix.

> Support Multiple Inputs/Outputs in MapProcessor / ReduceProcessor 
> ------------------------------------------------------------------
>
>                 Key: TEZ-1031
>                 URL: https://issues.apache.org/jira/browse/TEZ-1031
>             Project: Apache Tez
>          Issue Type: New Feature
>    Affects Versions: 0.4.0, 0.5.0
>            Reporter: JunHo Cho
>              Labels: TEZ-0.3.0, TEZ-0.4.0
>
> Map/ReduceProcessor doesn't support multiple input / output.
> In some case DAG, this feature is necessary. 
> For examples, one reduce send output data to hdfs and another vertex to analyze it.



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