I'm going to basically reply with my normal QA rant. 1) QA is important to the overall health of Gentoo. People will not use broken shit. 2) QA should be straightforward. If a developer need to do X to assure quality it should be fairly obvious why X is required. It should be clear where to go for help. 2a) A developer should not get chewed out for asking for assistance or questioning policies. 3) If a QA policy is not straightforward; developers will not follow it. 4) QA should not be a road-block to most developers. If you make development harder people will often stop working on development.
I think a number of developers understand why QA exists, why they should test packages, run repoman, and other policies that often get followed. Examining policies that are ignored will likely lead to a lack of understanding, documentation, or just bloat in policy. In general I hate talking about 'bad' QA versus 'good' QA because no one on the QA team ever talked about measurement. QA is 'bad' when some new person heads up the team and (s)he is going to 'clean up QA' by instituting these new policies. None of the policies have any kind of measurement attached so there is no real way to see if the new policies are effective. Perhaps this sort of thing is 'too corporate' and not possible in a volunteer project (I happen to think otherwise.) -A On Mon, Jan 31, 2011 at 12:14 AM, Petteri Räty <betelge...@gentoo.org> wrote: > On 01/31/2011 07:04 AM, Jeroen Roovers wrote: >> >>> 2. I don't think it makes sense for QA to discipline developers >>> permanently in these cases. They should suspend access pending Devrel >>> resolution of the issue. Devrel should of course strongly consider >>> the input of QA. >> >> That should be anyone's input, really. If a Gentoo Linux user finds a >> nasty `rm -rf /' timebomb, I suppose he could point that out to infra >> directly. And it's infra that suspends access, by the way. And devrel >> should be the intermediate between developers. And QA "aims to keep the >> portage tree in a consistent state"[1]. Wait, everyone is already in >> place? >> > > Actually recruiters can also suspend commit access and DevRel lead has > used that to safe guard the tree in the past. > > Regards, > Petteri > >
