[
https://issues.apache.org/jira/browse/HADOOP-11393?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14535161#comment-14535161
]
Roman Shaposhnik commented on HADOOP-11393:
-------------------------------------------
The patch is pretty straightforward and in general looks good to me. A couple
of comments still:
* the parts that patch documentation need to be rebased on the current state
of trunk
* the hadoop-functions.sh part needs to be rebased as well
* httpfs-env.sh httpfs-config.sh httpfs.sh mapred-config.sh rumen2sls.sh
still use HADOOP_PREFIX even after applying the patch
The biggest question I have is re: overriding HADOOP_HOME with HADOOP_PREFIX
unconditionally. Shouldn't we at least
start issuing a deprecation warning for use of HADOOP_PREFIX ?
[~aw] do you want to take care of the above?
> Revert HADOOP_PREFIX, go back to HADOOP_HOME
> --------------------------------------------
>
> Key: HADOOP-11393
> URL: https://issues.apache.org/jira/browse/HADOOP-11393
> Project: Hadoop Common
> Issue Type: Improvement
> Affects Versions: 3.0.0
> Reporter: Allen Wittenauer
> Labels: BB2015-05-TBR
> Attachments: HADOOP-11393-00.patch
>
>
> Today, Windows and parts of the Hadoop source code still use HADOOP_HOME.
> The switch to HADOOP_PREFIX back in 0.21 or so didn't really accomplish what
> it was intended to do and only helped confuse the situation.
> _HOME is a much more standard suffix and is, in fact, used for everything in
> Hadoop except for the top level project home. I think it would be beneficial
> to use HADOOP_HOME in the shell code as the Official(tm) variable, still
> honoring HADOOP_PREFIX if it is set.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
