[ 
https://issues.apache.org/jira/browse/LUCENE-4654?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Dawid Weiss resolved LUCENE-4654.
---------------------------------

    Resolution: Fixed

I've implemented this and committed it in along with an update to precomputed 
statistics.

This will *not* make your tests a whole lot faster because there is 
job-stealing by default anyway so unless you really hit a long test scheduled 
at the very end of a series of per-jvm tests it won't make a difference.

Obviously it won't make any difference if you're running with a single JVM 
either.

What may make a (small) difference is for people with a lot of cores or, 
paradoxically, a few cores on slower machines. I think it's an improvement over 
the existing version anyway so it's for the better.

                
> Test duration statistics from multiple test runs should be reused (locally).
> ----------------------------------------------------------------------------
>
>                 Key: LUCENE-4654
>                 URL: https://issues.apache.org/jira/browse/LUCENE-4654
>             Project: Lucene - Core
>          Issue Type: Bug
>            Reporter: Dawid Weiss
>            Assignee: Dawid Weiss
>            Priority: Trivial
>             Fix For: 4.2, 5.0
>
>
> This is trivial to accomplish: when somebody (or jenkins) runs tests multiple 
> times the execution statistics could be reused to improve load balancing on 
> the local machine (local hardware and settings) in favor of the precached 
> values currently version in the svn repo.
> At this moment we already do this, but keep the stats under build/ and every 
> ant clean effectively removes them. I could move those stats under an 
> svn-ignored folder elsewhere so that these stats are not lost and reused for 
> balancing.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to