[
https://issues.apache.org/jira/browse/HADOOP-4448?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12674257#action_12674257
]
Enis Soztutar commented on HADOOP-4448:
---------------------------------------
bq. In my example, the jar file that was submitted and the jar file that was
incorrectly run both had the same full class name.
In java, the classname is really the package name + class name, and with the
standard naming convention, the class names are unique. If any of two classes
have the same name, then there is something wrong with the naming in the first
place, you should have set the package name accordingly.
As I understand this, there is nothing can be done here. I will close this as
won't fix unless any one objects.
> "hadoop jar" runs wrong jar file in case of name clash
> -------------------------------------------------------
>
> Key: HADOOP-4448
> URL: https://issues.apache.org/jira/browse/HADOOP-4448
> Project: Hadoop Core
> Issue Type: Bug
> Components: util
> Environment: RHEL Linux
> Reporter: John DeTreville
> Fix For: 0.17.3
>
> Original Estimate: 1h
> Remaining Estimate: 1h
>
> Imagine I run a command as follows:
> hadoop jar test.jar
> Let's imagine that my main module in test.jar is named Sort.
> Let's also imagine that some jar file in my CLASSPATH contains a module also
> named Sort, which also defines main.
> Hadoop will run THAT jar file, not the one I specified.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
