Interesting posting. But i don't think, that this is the right model for refactoring, where a lot of 'depreciated classes' (reason: copyright, restructuring, replacement, creating 'orthogonality', whatever) cause much collateral damages in hundreds of packages, like it is in pharo at the moment. IMHO, there are three models necessary with three different cycles going round robin, with different *emphasis*: 1. advancing cycle in development as described in the article 2. refactoring cycle with strong emphasis on sunit and testing 3. package update and compatibility cycle #3 is similar to the 'maintainer model' in Debian development process. Smalltalk *users* (meant: app developers) should overtake, because it's rather in their interest. #2 'Beauty is, where we refactor it': User and core developer interest. #1 core developer only. 'Users' rather destroy, because things get too complicated. (debugger, event handler) regards, Guido Stepken Am 10.02.2012 21:11 schrieb "Stéphane Ducasse" <stephane.ducasse@inria.fr>:

http://nvie.com/posts/a-successful-git-branching-model/ tx dale for the pointer stef