On Fri, Aug 7, 2015 at 11:11 AM, Kevin Fenzi <kevin(a)scrye.com> wrote:
On Fri, 07 Aug 2015 10:03:34 -0700
Adam Williamson <adamwill(a)fedoraproject.org> wrote:
> This is exactly what I *don't* want to happen, and what rather annoys
> me about this whole business.
>
> The blocker process is not a tool for the rest of the project to use
> as a reminder system. It doesn't work very well for that. The
> expectation the blocker process is built around is that people should
> be working towards its requirements *all the time*, ideally well in
> advance, and the blocker process is there to catch the relatively
> *few* cases where things break unexpectedly. It's absolutely not
> supposed to work like this:
>
> * QA tests Thing and finds no-one ever even started working on it
> * QA files bug and marks it blocker
> * Team Thing starts working on Thing
>
> I realize that's an exaggerated example, but it's just to make the
> point clear. That's not a sane development process.
+1000.
I'd like to propose something somewhat shocking.
How about we get these things done in rawhide, so we are a release
ahead of things instead of scrambling to get them done at the last
minute all the time.
Cannot the design team work on F24 wallpaper _right_ now? get it in
rawhide and test that it's all there and then when we branch it's done,
they can work on f25 in rawhide and so on.
Just a thought...
For that matter, it still baffles me that there are late changes
(before and even after freeze) that get applied to Rawhide and Fedora
N concurrently. I thought such changes were supposed to go in Rawhide
first. That's what I think needs teeth, is ending this simultaneous
application of changes to Rawhide and current.
--
Chris Murphy