I would see reason to create separate branch if we would be doing some new and possibly breaking feature while working on smaller features on the main line with possibility to cut release before the new big thing is done. As SQOOP-1367 seems important enough to finish prior any additional Sqoop 2 release nor other bigger features, I would be personally fine with destabilizing the sqoop2 branch and simply do all the drastic changes there.
Jarcec On Jul 10, 2014, at 8:03 PM, Gwen Shapira <gshap...@cloudera.com> wrote: > Good idea to work on drastic changes in a separate branch. > > Gwen > > On Jul 10, 2014 3:00 PM, "Abraham Elmahrek" <a...@apache.org> wrote: > >> Dev folks, >> >> SQOOP-1367 requires drastic changes to the Sqoop2 code base and will likely >> bring some level of instability for a short period of time. Could we create >> a separate branch for SQOOP-1367 development? Here are a few reasons why >> this seems like a good idea: >> >> - Since releases are built from the "sqoop2" branch, it seems like it >> should be stable. >> - We could disable and enable tests freely as this branches stability is >> not as important. >> - Chunking work up will become a lot easier since we would not care if >> this branch immediately works. This will make code reviews much easier. >> - SQOOP-1367 is a core change to Sqoop2. It will touch several files and >> reshape the shape of jobs. >> >> -Abe >>
