[
https://issues.apache.org/jira/browse/YETUS-281?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15135875#comment-15135875
]
Allen Wittenauer commented on YETUS-281:
----------------------------------------
I haven't had a chance yet to trace the code and figure out why it wasn't
working. But if you look at, e.g., YARN-4219, you can see that it never ran
the unit tests or javadoc in the pluginstorage module
(https://builds.apache.org/job/PreCommit-YARN-Build/10373/artifact/patchprocess/
for what it did run). The modules listing in the footer is what the generic
tooling things it should be and doesn't reflect what the personality does with
it, so in this case is very misleading.
For testing, I've been making fake patches that touch pom files as part of the
work for YETUS-284. When comparing the output, that's when I noticed that the
HMM code was doing bad things (dropping modules and in some other edge case
doubling them!).
> hadoop: use built-in dependency order
> -------------------------------------
>
> Key: YETUS-281
> URL: https://issues.apache.org/jira/browse/YETUS-281
> Project: Yetus
> Issue Type: Improvement
> Components: Test Patch
> Reporter: Allen Wittenauer
> Assignee: Allen Wittenauer
> Priority: Blocker
> Attachments: YETUS-281.00.patch
>
>
> With the commit of YETUS-280, hadoop should loosen the reins a bit and drop
> several hundred lines of code to use the built-in dependency creator.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
