Hello all

Now that the cat^H^H^Hrelease is out of the bag, I'd like to take the 
opportunity to discuss a subject I've been thinking of for the past week.

You may have seen that we released something called "Release Candidate 1", 
which should mean that it's as good as a final release, but we aren't sure yet. 
And that is correct: our criteria for releasing is having all manual tests 
pass, we do an acceptance testing and we must have zero (or close to) 
remaining showstopper tasks.

A showstopper task is one with priority 0 or 1.

Now, what you may not know is that the package you're seeing is the seventh 
build of a release candidate. We've been creating RC packages for the past 3 
or 4 weeks and doing testing ourselves. One after the other, the packages were 
discarded because either there were remaining P1s or because the Black Team 
found something wrong.

I'm not questioning their effectiveness. In fact, they do a great job.

But what I'm thinking is: how should we do this in the Open?

Clearly, all the builds for the release candidate (or any other release, for 
that matter) must be public, so others are as able to do testing as our 
engineers are. Also, there must be a way for everyone to give feedback (which 
means they must have time to do that).

What should we call those packages? Should each of them be a Release 
Candidate? We'd be on RC7 by now, having released at least a dozen betas 
previously.

I was also thinking that release milestones should be a quality gate. That is, 
we periodically make packages, but the label changes according to the quality 
status.

Would this work? How do other projects do it?

-- 
Thiago Macieira - thiago.macieira (AT) nokia.com
  Senior Product Manager - Nokia, Qt Development Frameworks
     Sandakerveien 116, NO-0402 Oslo, Norway

Attachment: signature.asc
Description: This is a digitally signed message part.

_______________________________________________
Opengov mailing list
[email protected]
http://lists.qt-labs.org/listinfo/opengov

Reply via email to