Maybe more of this work needs to be done in feature branches? You can make the argument that SVN (at least as far as I know) encourages an "everything goes in the central repository" work flow because it doesn't have the GIT distributed workflow. We certainly don't want to discourage new and adventurous development because that is the road to "fossilization" but, on the other hand, we definitely don't want every wild idea just going into upstream and turning it into a big mess. In the GIT workflow, these changes would be made in someone's fork and they would issue a pull request. Reasonable criticisms about style or architecture would be addressed and then the work would be pulled in as a whole when it reaches a certain level of quality.
This is an important topic. Its basically the same issue that provoked David to go create the Moqui framework. Evolving the platform while keeping architectural coherency and stability is not easy. I do think we need to pursue this "app store" concept and find a way for "plugins" to be a major part of how we add features. Adding these features in needs to be as easy as clearing the cache in Webtools so that implementors do not feel like they are a "second class citizen" just because their code isn't in the core upstream repository. If we figured that out properly we might want to jettison A LOT more stuff from core. ----- "Scott Gray" wrote: > One thing I'm starting to get tired of is contributors (and committers) > beginning major works without a thorough discussion about the suitability of > the work for OFBiz before starting. I find it frustrating that reviewers are > then forced to review under some sort of urgency because it is "ready to > commit" and also made to feel like the contributor's time has been wasted if > there are any major issues/disagreements with the design decisions made in > the work. > In regards to Jacques, I also find it frustrating that he encourages and > actively participates in this behavior without actually really performing > much in the way of design review other than a generic "does it seem like a > good feature?" test. Don't get me wrong, encouraging contributors to > contribute is a great thing and Jacques does an amazing job interacting with > the community as a whole but whenever a major work is undertaken without > prior discussion then the contributor is taking a big gamble and they should > be made well aware of that before starting. -- Ean Schuessler, CTO e...@brainfood.com 214-720-0700 x 315 Brainfood, Inc. http://www.brainfood.com
