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 2015/07/01 21:41:05 UTC

[jira] [Comment Edited] (HADOOP-12111) [Umbrella] Split test-patch off into its own TLP

    [ https://issues.apache.org/jira/browse/HADOOP-12111?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14610892#comment-14610892 ] 

Allen Wittenauer edited comment on HADOOP-12111 at 7/1/15 7:41 PM:
-------------------------------------------------------------------

A super-complex test case:  https://issues.apache.org/jira/browse/HADOOP-11984?focusedCommentId=14566293&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-14566293

I don't know if I can list all the things that the patch actually tests because it hits so many subsystems.  Highlights are that it used multijdk, used parallel test execution,  and some extra native compile time flags that are specific to Hadoop.

Two interesting bugs and some other bits from the output highlighted:
* checkstyle output can be multi-line, which makes it tricky to handle when doing before/after comparisons like we want to do.  (cc: [~romani]) 
* findbugs convertXmlToText is only failing for one of the modules and we do a pretty bad job of giving any hints as to why.   Filed HADOOP-12166 to see if we can do better.
* I'm assuming the mvn install convergence errors are just typical maven stupidity w/multiple writers. [HADOOP-12146 should alleviate a lot of that.]
* Hadoop-specific: we might need to bump up the unit test heap for JDK8. :(


was (Author: aw):
A super-complex test case:  https://issues.apache.org/jira/browse/HADOOP-11984?focusedCommentId=14566293&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-14566293

I don't know if I can list all the things that the patch actually tests because it hits so many subsystems.  Highlights are that it used multijdk, uses parallel test execution,  and extra native compile time flags that are specific to Hadoop.

Two interesting bugs highlighted:
* checkstyle output can be multi-line, which makes it tricky to handle when doing before/after comparisons like we want to do.  (cc: [~romani]) 
* findbugs convertXmlToText is only failing for one of the modules and we do a pretty bad job of giving any hints as to why.   Filed HADOOP-12166 to see if we can do better.
* I'm assuming the mvn install convergence errors are just typical maven stupidity w/multiple writers. [HADOOP-12146 should alleviate a lot of that.]
* Hadoop-specific: we might need to bump up the unit test heap for JDK8. :(

> [Umbrella] Split test-patch off into its own TLP
> ------------------------------------------------
>
>                 Key: HADOOP-12111
>                 URL: https://issues.apache.org/jira/browse/HADOOP-12111
>             Project: Hadoop Common
>          Issue Type: New Feature
>          Components: yetus
>    Affects Versions: HADOOP-12111
>            Reporter: Allen Wittenauer
>
> Given test-patch's tendency to get forked into a variety of different projects, it makes a lot of sense to make an Apache TLP so that everyone can benefit from a common code base.



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