I like the overall idea. One concern is that it might be hard to maintain backward compatibility with both 0.6 and 0.7.
On Mon, Nov 21, 2016 at 10:17 PM, Lei Xia <[email protected]> wrote: > Hi, All > > Helix 0.7.x branch has been there for a while, however, given it has > back-incompatible API changes, most of our exiting customers are reluctant > to move to 0.7. This forces us to maintain both branches, in addition, > most of recent new features and important fixes (task framework > improvements, new auto-rebalancer features) have only been pushed to > 0.6.x, which makes two branches diverged further apart. It is especially > harder to keep maintaining both branches now. > > I proposed to fork a new branch (helix-0.8.x) from 0.6.x, with a new > helix-api module containing all new API classes introduced in 0.7, but > still also keeps all old API classes (maybe marked as deprecated) in > helix-core. In this way, we could push existing customers move to 0.8.x > release without forcing them to remodel their codes. > Then we only need to maintain a single unified branch, and keep moving > forwards to new API with all new developments happening in this branch. > > I have cloned a 0.8.x-test branch ( > https://github.com/apache/helix/tree/helix-0.8.x-test) from 0.6.x, and we > (me and Junkai) are going to cherry-pick changes from 0.7.x and apply them > to this branch and continue testing it until we reach a point that we can > confidently release it :). > > Please let me know what you think about it, any suggestions or comments > are appreciated! Thanks > > > Best > Lei >
