[jira] [Commented] (TEZ-2593) Script to publish tez jars built against different versions of hadoop

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

    [ https://issues.apache.org/jira/browse/TEZ-2593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15047275#comment-15047275 ] 

Hitesh Shah commented on TEZ-2593:
----------------------------------

Some more comments in the code will help. 

Deploy does not work without signing with gpg keys. Please check the release guide for more details. 

"-Dhadoop.verison=${hadoop_version}" has been known to not work correctly in some cases. It might be better to replace the hadoop.version value in the pom itself. Also, based on the hadoop version, certain profiles need to be enabled/disabled. 

One safety check could be to make sure that this command is run from within the git repo so we can leverage commands like git clean, git reset etc before starting the main execution. Or if the command is run on a dirty workspace, the command should error out. 





> Script to publish tez jars built against different versions of hadoop
> ---------------------------------------------------------------------
>
>                 Key: TEZ-2593
>                 URL: https://issues.apache.org/jira/browse/TEZ-2593
>             Project: Apache Tez
>          Issue Type: Sub-task
>            Reporter: Hitesh Shah
>         Attachments: TEZ-2593-1.patch
>
>
> One ask from the Flink devs was that tez automatically pulls in hadoop-2.6.0 as a dependency. They use hadoop-2.2 and would prefer to use a version of tez compiled against hadoop-2.2.
> In some sense, this can be addressed by using dependency exclusions but having a script to build and deploy tez compiled against different hadoop versions would be useful. 
> Basic ask: 
>    - update tez version to say 0.8.0-hadoop220 when compiling against hadoop version 2.2.0 
>    - publish artifacts as needed



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)