You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-issues@hadoop.apache.org by "Allen Wittenauer (JIRA)" <ji...@apache.org> on 2017/03/20 18:30:42 UTC
[jira] [Comment Edited] (HADOOP-14162) Improve release scripts to
automate missing steps
[ https://issues.apache.org/jira/browse/HADOOP-14162?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15933248#comment-15933248 ]
Allen Wittenauer edited comment on HADOOP-14162 at 3/20/17 6:30 PM:
--------------------------------------------------------------------
I don't think adding Yet Another Build Tool is a good idea. We should be able to do everything we can do in ant with maven. It's fine to steal ideas though.
That said.
The only relevant thing I'm seeing that really needs to change in the core create-release script is being able to specify a different location for PGP keys file verification. Everything else belongs elsewhere.
Let me clarify:
My modifications to create-release were based on the idea that it does one thing: it creates a release artifact. That's it. Taking bits on disk and turning them into verifiable tar balls is already complicated (and time consuming!) enough. A lot of the stuff being talked about in this JIRA issue should really be getting done in other components before and after create-release executes. This allows for those components to be easily replaced by local ones (if need be) as well as cuts down on the amount of program logic required to "do everything". It greatly simplifies the testing. If we want a "one step", then we just need a driver to run those different components.
was (Author: aw):
I don't think adding Yet Another Build Tool is a good idea. We should be able to do everything we can do in ant with maven. It's fine to steal ideas though.
That said.
The only relevant thing I'm seeing that really needs to change in the core create-release script is being able to specify a different location for PGP keys file verification. Everything else belongs elsewhere.
Let me clarify:
My modifications to create-release were based on the idea that it does one thing: it creates a release artifact. That's it. Taking bits on disk and turning them into verifiable tar balls is already complicated (and time consume!) enough. A lot of the stuff being talked about in this JIRA issue should really be getting done in other components before and after create-release executes. This allows for those components to be easily replaced by local ones (if need be) as well as cuts down on the amount of program logic required to "do everything". It greatly simplifies the testing. If we want a "one step", then we just need a driver to run those different components.
> Improve release scripts to automate missing steps
> -------------------------------------------------
>
> Key: HADOOP-14162
> URL: https://issues.apache.org/jira/browse/HADOOP-14162
> Project: Hadoop Common
> Issue Type: Improvement
> Components: build
> Reporter: Elek, Marton
> Assignee: Elek, Marton
>
> According to the conversation on the dev mailing list one pain point of the release making is that even with the latest create-release script a lot of steps are not automated.
> This Jira is about creating a script which guides the release manager throw the proces:
> Goals:
> * It would work even without the apache infrastructure: with custom configuration (forked repositories/alternative nexus), it would be possible to test the scripts even by a non-commiter.
> * every step which could be automated should be scripted (create git branches, build,...). if something could be not automated there an explanation could be printed out, and wait for confirmation
> * Before dangerous steps (eg. bulk jira update) we can ask for confirmation and explain the
> * The run should be idempontent (and there should be an option to continue the release from any steps).
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: common-issues-help@hadoop.apache.org