Time based releases?

[Leonardo Franchi]

On Friday 27 March 2009 09:16:39 Mark Kretschmann wrote:
> On Sat, Mar 14, 2009 at 8:12 AM, Mark Kretschmann <kretschmann at kde.org> 
wrote:
> > So we've recently been discussing something on IRC. The gist of it:
> > "Hey, everybody is moving to time based release schedules. Let's do it
> > too!"
> >
> > I've been proposing doing something like a "2/2" scheme: 2 months
> > features, 2 months bug fixing, which would leave us with an
> > approximately 4 months cycle. Some others on IRC agreed to this scheme
> > too, in general.
>
> If possible I'd like to revive this thread. I just had the following
> short conversation with Leinir on IRC, which reflects my feelings
> pretty well:
>
> 10:00 < markey> man we really need that beta1 release, I predict it
> will refresh our batteries, create a big hype, and be a Good Thing
> (TM) in general
> 10:01 < markey> right now my motivations are a bit on a low point,
> partly an effect of not releasing for such a long time
> 10:01 < markey> shorter releases cycles ftw!
> 10:01 < markey> pretty please
> 10:02 < markey> users keep posting about 2.0.x, which is ancient by
> our standards
> 10:02 < leinir> Shorter cycles do seem to work :)
> 10:03 < leinir> i rather suspect git will help us with that as well...
> 10:03 < markey> yep :)
> 10:03 < leinir> that whole branch thing - being able to develop new
> stuff without trashing the whole app for months at a time :)
> 10:03 < markey> I still think this "4 months scheme" has a lot of merit
> 10:03 < leinir> Indeed :)
> 10:03 < markey> cramming too many new features into one release makes
> it hard to get everything functional
> 10:04 < markey> we're seeing that right now
>
>
> For the love of Seigo, let's get into a shorter, predictable release
> cycle. It's obvious to see that the project would benefit from it.
>
> ...and ignore Ian on this topic; he managed to bring this whole debate
> into a stasis single-handedly, without providing any useful input :p

I think the idea of a concrete release cycle is good---it gives us focus and 
allows us to work hard on new features then switch to bugfix mode. But I don't 
see why we need to fix the release cycle for *all* our releases beforehand---
why not just decide the length of the next release cycle after each release? 
Most likely we'll have more information at hand, and can tailor it to our 
specific needs.

leo



-- 
-----
lfranchi at kde.org		Tufts  University 2010
leonardo.franchi at tufts.edu                The KDE Project