Yes, if use CTR mechanism, the rollback or little polishes maybe will happen very often, this new branching model can cover this situation. BTW, as Xin and Roman mentioned, develop on `master` branch doesn’t have problems, so does develop on `develop` branch. Regards, yukon On Fri, Mar 10, 2017 at 3:51 PM, Von Gosling <vongosl...@apache.org> wrote: The master branch is a stable baseline. The develop branch is a CI baseline.
IMO, the develop and master branch model are easy to scalable, especially for CTR mechanism and more and more guys contributions. :-) > 在 2017年3月10日,15:39,Xin Wang <data.xinw...@gmail.com> 写道: > > Agree with Roman, I don't think developing on master is a problem. Just > like many bigdata projects: Storm,Spark,Flink.. > > - Xin > > 2017-03-10 15:34 GMT+08:00 yukon <yu...@apache.org>: > >> Hi, >> >> We also use tags for releases, but make master branch in a stable state may >> be better. And this branching model is very classical and used widely. >> >> Regards >> >> On Fri, Mar 10, 2017 at 3:07 PM, Roman Shtykh <rsht...@yahoo.com.invalid> >> wrote: >> >>> Guys, what is the reason for not developing on master branch and using >>> tags for releases? It works for many other projects. >>> Roman >>> >>> >>> On Friday, March 10, 2017 3:14 PM, yukon <yu...@apache.org> wrote: >>> >>> >>> Hi all, >>> >>> Our new branching model has been put into effect, more details please >> refer >>> to [1]. >>> >>> So, please kindly send pull requests to develop branch instead of master >>> branch. Thanks. >>> >>> [1]. http://rocketmq.incubator.apache.org/docs/branching-model >>> >>> Regards, >>> yukon >>> >>> >>> >>> >>