[
https://issues.apache.org/jira/browse/MAPREDUCE-2661?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ahmed Radwan updated MAPREDUCE-2661:
------------------------------------
Attachment: MAPREDUCE-2661_rev3.patch
Thanks Devaraj for your review!
> I feel it is not correct to give the getSplitsAsString() implementation
> details of MapTaskImpl.java and ReduceTaskImpl.java in TaskImpl.java.
Yes, and this is why the patch pushes the implementation detail to subclasses.
> TaskImpl.java can provide default implementation of getSplitsAsString()
> instead of making it as abstract and returning with empty string in
> ReduceTaskImpl.java.
Agree, it will be similar that way. I have updated the patch accordingly.
Thanks
> MR-279: Accessing MapTaskImpl from TaskImpl
> -------------------------------------------
>
> Key: MAPREDUCE-2661
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-2661
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mrv2
> Reporter: Ahmed Radwan
> Assignee: Ahmed Radwan
> Priority: Minor
> Fix For: 0.23.0
>
> Attachments: MAPREDUCE-2661.patch, MAPREDUCE-2661_rev2.patch,
> MAPREDUCE-2661_rev3.patch
>
>
> We are directly accessing MapTaskImpl in
> TaskImpl.InitialScheduleTransition.transition(..). It'll be better to
> reorganize the code so each subclass can provide its own behavior instead of
> explicitly checking for the subclass type.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
