Skip to main content

Re: "Sun appears commited to fixing any/all regressions prior to release"

Posted by robogeek on March 25, 2006 at 10:54 AM PST

There was an interesting discussion thread on javalobby last week: Poll: Delay Mustang in favor of more fixes? ... I think the discussion added a lot of value to the discussion around Java and especially the in-progress Mustang release.

I wanted to focus on one statement that really stands out to me as the project leader for the Mustang Regression contest: "Sun appears commited to fixing any/all regressions prior to release". That's exactly the idea we wanted to reinforce with the public.

The promise we're making with the contest is not new. For as long as I have been with the Java team (since the tail end of 1.2.2) we have treated regressions with high priority. The contest comes out of this policy. It is not a new policy we're following, we have for a very long time held regressions in high priority, and would hold shipping a release to fix all known regressions.

The Mustang Regression contest has less than 1 week to go.

If you find a Mustang regression after the contest closes, no worry. A regression is a regression is a regression, and we will treat it the same as if it were a contest entry. The contest entries are not receiving special treatment, except for the prizes and t-shirts.

The key thing is we can only fix regressions that we know about. We have a lot of testing we do ourselves and we find many regressions ourselves. However, I've written this before ... there's a very important set of tests we cannot do. We do not have your applications, we do not have your environment, we do not have your data, we do not have your workload, etc.

The critical test is done by you when you first run your application with Mustang. It's better for all of us if you tell us about regressions now, rather than waiting for a year past the Mustang release before telling us about regressions.

Related Topics >>