Beta 2 tagging delay

[Boudewijn Rempt]

On Thursday 29 September 2011 Sep, Cyrille Berger Skott wrote:
> On Thursday 29 September 2011, Boudewijn Rempt wrote:
> > On Thursday 29 September 2011 Sep, Cyrille Berger Skott wrote:
> > > I am rather convince that our current model (or the one we aim at, ie
> > > master in a releasable state at all time), but it does require people to
> > > accept that their latest newest shiny super cool feature is less
> > > important than shipping a finished product to our users, and therefor
> > > merge things that are reasonnably ready, for which the developers don't
> > > know of any major bugs. Maybe we should have a long discussion on the
> > > development workflow at the next sprint.
> > 
> > It also means that we need to setup the system of branches we have
> > discussed before a couple of times already. We don't have that now, with
> > the result that we are still working as if in svn.

> Huh ? That is not true, we do work with branches. 

Yes, it is true. All branches related to the release process are missing. That we are using feature branches during development is not relevant, because those branches get pushed into master when we "think" they are ready. There's no big difference to svn here, where stuff got pushed into trunk because we "thought" they were ready as well

There's no staging branch where people can easily track all branches in one go, and that is the reason that the branches do not get sufficient testing before merging into master.

> There are three branches 
> that are missing from that system, "master-validated", "stable" and "stable-
> validated", the later two, mostly because we don't have any stable release... 
> And "master-validated" because, at this point, there isn't a single revision 
> of calligra that validate. And setting up a system where checking that there 
> is not an increase of test regressions is way more complicated and time 
> consuming than checking that the number of test failures == 0.
> 
> > And there is still the problem that branches don't get enough user testing
> > which is why we thought that Dmitry's strokes branch was pretty ok.
> > Several people have tested it, but not thoroughly enough.

> So krita isn't that broken or is it ? Do we have bug numbers that we can refer 
> to, to make an evaluation ?

Yes we do have bug numbers. We have been discussing them all week. And yes, Krita is that broken. I do know what I'm talking about. For your reference:

KDE bug 282737 in krita (general) "Crash while painting" [Critical,Unconfirmed] http://bugs.kde.org/282737

KDE bug 282754 in krita (general) "Crash while painting" [Crash,Unconfirmed] http://bugs.kde.org/282754

Currently, painting for more than a few minutes is impossible. And we did not realize that because nobody tested that branch while making a real painting, so we got the go-ahead from our test users.

-- 
Boudewijn Rempt
http://www.valdyas.org, http://www.krita.org, http://www.boudewijnrempt.nl