okay, i agree your arguments, but every time we made the next release candidate (like 2.3.0 rc1, 2.4.0 rcX, ...) the real development in cvs stops, if we have no alternative cvs tree.
I agree - and it makes it a hell lot more difficult to trace changes and the influence of these changes to existing code. When involving many deveopers spread around the world - CVS becomes very essential to other developers in the spirit of checking code with changes made by other developers.
CVS is actually working as a developer-community and more important as a documentation of all minor/major changes - making it a lot easier to backtrace introduced bugs and resolving those.
So my conclusion: when developers ain't sitting around in the same office or building - CVS is essential to successfull development and bug-tracing/solving and it's really worth the trouble of maintaing "branches" - making it possible for everyone to see changes. So it's a bad idea to skip CVS and everyones staring to develop new features / making code more efficient in their own versions - making it a hell lot more dificult to merge all those versions in the end...
So let's keep the development visible to all in CVS and let it help in keeping the development-team together and focused on the same goal :-)))