|
|
|
|
Release models make a huge difference in the properties of the software delivered. I believe there is no single "ideal" model; what to choose depends on the code base, the group/community creating the code, the users/customers, the technology available (languages, CI tools, others)... What works for Hudson does not work for GlassFish nor for Solaris.
|
MySQL is changing its release model to improve agility, quality, predictability and facilitate contributions. Giuseppe just posted an Overview; in a nutshell, the trunk tree is always in beta quality, new features are first developed in stage branches, then integrated into the trunk, which is then brought to RC quality and another cycle starts (see Diagram). Full details at the MySQL Forge and in Tomas's Presentation at MySQL University (slides; recording is NYA). |
The basic model seems feasible; now we need a few release cycles to adjust the model and we will see how it works in real-life. As Don Quijote said... the proof of the pudding is in the eating.
I sounds good. Indeed, we'll have to see how it works out.
It's important to realise that processes (development, or business) aren't created by a policy decision as such, and nor can they be changed that way.
Posted by Arjen Lentz on June 14, 2009 at 12:18 PM PDT #