Re: PEAR a way forward

On 2006 10 05 12:57, Alexey Borzov wrote:

Hi, Scott Mattocks wrote:

We already had a couple groups with cool sounding names who did almost nothing. What exactly makes you think that it will be different this time?

The fact that the groups get regularly renewed with people that want to do the job. The current groups are stagnant and most of them have no real interest in doing what the group was designed to do.

QA team already had rotation mechanisms in place. And it was supposedly comprised of people who wanted to do the job (since these people volunteered for QA team in the first place). So, Scott, please enlighten me on two points: 1) Why did the previous QA team fail? 2) Why will the QA team you propose succeed?

Plus, I want to provide specific points of accountability for the groups. If you don't do your job, you get the boot and someone who will do it takes over.

Who exactly will be giving this "boot"?

I'm not sure why QA isn't the responsibility of the developer, and why passing QA isn't an automated prerequisite before a release is deemed stable. There's no question that building a set of automated regression tests is a royal pain in the a**, but in my experience it's always been worth the effort. If there is a QA role, wouldn't that role consist of vetting and adding test cases to a QA suite? Anyone should be able to prepare a test case, but there has to be a gatekeeper (preferably other than the package developer) since "bugs" can be spurious and/or misinformed. Filtering and promoting test cases is something that can be a widely distributed responsibility; a capability granted by karma. If there's a backlog, a notice to this mailing list should be enough to get people to go look through it.