[Qgis-developer] Failing tests consider blockers

[Vincent Picavet]

Hello,

Le mercredi 19 février 2014 12:32:39, Martin Dobias a écrit :
[...]
> Agreed. If we decided to do time-based releases, let's do them as
> planned and not discuss the decision two days before the planned
> release...
> 
> I think we should mainly reconsider which bugs should be marked as
> blockers. The current policy "every regression is a blocker" does not
> make sense to me. In my opinion a blocker is something that makes the
> software unusable for potentially large amount of users - in such case
> it makes sense to delay the release for few days (as an exception, not
> a rule).

+1 to all this

It will probably take some time for all of us devs to take good habits of 
fixing blockers as soon as possible. It will probably also encourage funders to 
give money to developers to fix them before the release.
At the end, it should be a better solution to keep strict fixed releases, and 
should enforce code quality.
One way to ensure this is to work seriously on tests as well.


Vincent