On Tue, 2 Feb 2010, Adam Miller wrote:
On Tue, Feb 2, 2010 at 2:17 PM, Bill Nottingham notting@redhat.com wrote:
<Snip> > Would that mean that users who don't start with one of these 'products' > get to magically try and choose which implementation of which they want? > Perhaps even mix and match, leaving QA and the developers to sort out > the results. > > Furthermore, you then leave 'downstream' higher-level packages and > applications having to, for example, code to PolicyKit0, PolicyKit1, or > consolehelper, depending on what each 'product' use case might use. Or, > having to build their python extensions simultaneously for python2.4, python2.6, > and python3.0. These sorts of things would be extremely painful for > developers, and would bloat the QA matrix excessively. > > Not to reduce the debate to too much of a soundbite, but it almost > seems like attempting to decide whether we want Fedora to be Debian, > or to be something useful for users of it. I'd always pick the latter... <Snip>
I think the responsibility of these things should be placed upon the SIG members who perform the functions from within these different groups. Why not have a QA person from each SIG work together with the larger QA efforts instead of potentially against them?
QA is a particular skill set, not every sig has a QA member and requiring it wouldn't work either. I feel it's like assuming that just because I've done turbogears apps that someone would ask me to do CSS as well. I don't think it's safe to assume that because someone can put a spin together that they have the tools and knowledge to do proper QA on it.
-Mike