[Qgis-developer] Realease and blockers

[Vincent Picavet]

Hello,

Le vendredi 6 juin 2014 12:30:23, Nyall Dawson a écrit :
> On 06/06/2014 8:20 pm, "Vincent Picavet" <vincent.ml at oslandia.com> wrote:
> > Could we make a more visible RC release on the website, encouraging
> testing ?
> 
> > I think that our user are not used yet to our new release cycle. We
> > really should educate them, and lower the barrier to RC testing and bug
> reporting. As for now, even installing the RCs is complicated, so is bug
> reporting.
> 
> I don't think the issue is just bug reporting for a RC though. As
> mentioned, we've already got a very long blocker list, which is still
> growing.
> I think a bigger issue is development resources and developer's time
> availability. At the current rate of bug closing we're very likely to still
> have a large number of these issues outstanding at the end of the month.

More bug reports, bug comments and better bug reports is already a lot of 
developer's time saved.
Having more testers (and regular testers) is as important as having developers 
to fix bugs.

My proposal was also more oriented towards getting users involved in all ways, 
and especially funding at Feature Freeze time.
Getting them to download, install and test more thoroughly the RCs, and using 
this media to incent them to fund the software do contribute to more 
developers time.

> I'm in favor of a longer freeze period for next release.

I am not sure that more non-paid developer time is a sustainable model.

Or it is a matter of going from a FF development model, to a more formal peer-
review commitfest model ( like http://rhaas.blogspot.fr/2011/03/commitfests-
and-meritocracy.html ).

There are many other ways to reduce bugs even before reaching feature freeze, 
some of them already discussed thoroughly here, I am just trying to find some 
easy, reachable and efficient means, which can of course be complementary to 
other means.

Vincent