On Tue, 05 Mar 2013 03:09:15 +1300, Stephan Aßmus <superstippi@xxxxxx> wrote:
Am 04.03.2013 um 02:47 schrieb Clemens <clemens.zeidler@xxxxxxxxx>:My opinion is that the R1 target is not appealing any more (except package fs which is unique and new!) and if there are people interesting in developing R2 they should be supported to fill the gap.I don't see the connection you make between the R1 target and the Haiku source tree, or the "main branch". Putting some feature, preflet, application or server into the Haiku source tree has never meant that it *will* be included in R1 or the "Haiku image" target. There is plenty of code that is not included. And plenty of private, unfinished API that is scheduled to (maybe) become public one day.No one forces you to only commit code to the main branch that is useful to reach the R1 milestone. Of course it is always good to discuss things with the other developers. If a discussion gets out of hand, or too distracting, then sometimes it is shut down with the "lets keep focussed on R1" argument. But that doesn't mean you are not allowed to work on stuff which is unlikely included in R1.
fully agree. However, I have the feeling it is harder do that in the main branch. At the moment it might be easier but when it comes to a release it will become hard again. A second branch would be helpful to try stuff out, i.e. things that not can be developed entirely in a branch. For example, the last window management discussion (maximizing, resizing, zooming...) One can implement such stuff in a branch but if nobody is using and trying it, its pointless. There must be some user feedback like: "this feature is good, the other one is useless/annoying..."
I see the point that much stuff can be done in the main branch, but some testing stuff just do not fit there. I'm also not fixed on a R2 branch but some platform to do research on new stuff would be good. A R2 branch would have the advantage that all R2 developers automatically become guinea pigs ;-) Furthermore, official nighties would make it easier for non-developers. Hope you got the idea...
Of course you should not break stuff. And working on new global, system level APIs or services likely breaks stuff until it is ready and polished. So you should work on that in your own branch anyway. This was even true in the SVN times, but it has become much easier now.Given all that, what exactly would an "R2" branch accomplish? Do you really think there will be a sudden surge of fresh developers?
not a sudden surge, but it would indicate to the outside that there is already something planed for R2. The project would look more agile. A roadmap with a set of preliminary targets (make stuff from glass elevator more concrete) could attract people for R2 and even motivate people to finish R1. IMHO the R1 target was great to accomplish what has been done, thus we already should set some new goals. This would allow a softer transition from R1 to R2 development.
How are actually other projects like KDE4, GNOME3 got their roadmap? They did some fundamental changes. That must have been a big discussion, or did just one guy had a good idea and everybody else followed? Haiku hasn't had this problem so far because Be did everything for us.
Regards,
Clemens
