[
https://issues.apache.org/jira/browse/MAPREDUCE-5156?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13648466#comment-13648466
]
Jason Lowe commented on MAPREDUCE-5156:
---------------------------------------
bq. Therefore, the changes is trading binary compatibility of Hadoop 2 with
that of Hadoop 1. IMHO, since Hadoop 2 is still in alpha, the impact of
incompatibility should be much smaller.
Please be aware that these kinds of changes break binary compatibility not only
with Hadoop 2 but also Hadoop 0.23.
In this particular case the API is LimitedPrivate, so it should only apply to
code within Hadoop. In addition, the user would have to run a previous
examples jar with HADOOP_USER_CLASSPATH_FIRST=true and HADOOP_CLASSPATH set to
their examples jar to see the incompatibility, otherwise it ends up picking up
the 2.x versions and works. Therefore I'm OK with this specific change since I
think it is unlikely to break our 0.23 users.
I just wanted to make sure we are all aware that breaking compatibility with
Hadoop 2 often means breaking compatibility with Hadoop 0.23 as well and should
not be done lightly.
> Hadoop-examples-1.x.x.jar cannot run on Yarn
> --------------------------------------------
>
> Key: MAPREDUCE-5156
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-5156
> Project: Hadoop Map/Reduce
> Issue Type: Sub-task
> Reporter: Zhijie Shen
> Assignee: Zhijie Shen
> Priority: Blocker
> Attachments: MAPREDUCE-5156.1.patch, MAPREDUCE-5156.2.patch
>
>
> M/R examples are run through ProgramDriver.driver. ProgramDriver.driver
> returns void in hadoop-1, while it returns int in hadoop-2. Therefore, the
> function signatures in the example jar and yarn are incompatible.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
