Ok, so I booked my little world trip today. After GUADEC I will be going to London, Singapore, Sydney, Perth, Johanesburg and then Winhoek in Namibia before flying home. Just need to mail and call a lot of people over the next few days before I get the travel agent to lock the travel dates in stone :)
We are having a discussion on ABI stability on the GStreamer list promted by Wheeler’s suggestion that KDE would demand around 3 years cycles for ABI stability for the media framework they choose. Many of the GStreamer hackers don’t feel we are at a point in time where offering 3 year cycles without ABI breakage and I am not really in a position where I have any reason to think they are wrong. I mean yes of course we can offer X years of ABI stability on a certain release series, we do that today with the 0.6 branch and will do that with the 0.8 branch. But what is harder to promise is the level of development will happen in that branch eventually. On the other side one could question how much work would be needed in that branch eventually. If all it is meant to do is to provide well working media playback the 0.8 branch could probably do that easily for 3 years without needing much work except being keept up to date with the underlaying libraries.
In GNOME we have kinda worked around the issue up to now by only being in the desktop release instead of the ABI stable development plattform release. But also here I guess there would be a wish to move GStreamer into the plattform release as more and more parts of GNOME starts to depend on GStreamer. As a sidenote we have not been really good at following the GNOME release process (although we where much better this time around that the previous round:) and Jeff suggested in a mail that maybe GStreamer should be moved out of GNOME and instead be considered a underlaying library like X or libpng. Not sure if it will solve anything except releasing us from the GNOME release process. On the other side trying to follow the GNOME release cycle has helped us discipline ourselves in regards to stablizing things and provided us with a focus point for getting stuff done.
I guess things are never easy :) And just so it is said I don’t think any of our potential multimedia framework competitors are anywhere near a level where keeping 3 year ABI stability in their main development branch is viable. Unless they feel that sucking is a nice state to be in :)