Every suggestion looks good. 1. Java Client Is `JavaClient` related to Admin Interface? If you exclude `Admin Interface`, use Tinkerpop Interface as main Java Interface. In this case, however, it is likely to lose the advantage of Asynchronous Traverse, a feature of S2Graph. If need to implement it separately for the above reason, we should use `S2Edge, S2Vertex` from Tinkerpop implemention as POJO. and JavaClient should include a query builder. Given the user base, the provision of JavaClient is likely to be meaningful.
On Fri, Sep 8, 2017 at 9:29 PM DO YUNG YOON <[email protected]> wrote: > Hi folks. > > I want to discuss milestone for next release. > > Here is a my list that I wish include in next release. Please feel free to > correct me or add more issues I missed. > > 1. Java Client > > even though it is possible to use scala code in java, using scala code in > java project is frustrating. since there are more java users than scala > users, I think it is reasonable to provide both java and scala client. > > Regarding client, tinkerpop related codes are currently mixed with > s2graph's core scala code. it would be better if we can separate these two. > I think chainging tinkerpop related codes into java could make things easy > but not sure before try out. > > 2. Storage Backend > > I think it would be great if we can provide more storage backend. > personally, I really liked tinker graph which store data in memory, for > development and test. I suggest providing local storage data so users > without HBase can try out s2graph more easily. Also by trying to add new > storage backend, we can identify what need to be improved on our storage > interface. I think redis, cassandra, mysql, postgresql can be integrated > with s2graph as storage backend in theory, but have to try implementing > them to make sure they are possible. > > 3. Index Provider > > current implementation only use local lucene for global index. distributed > index provider, such as solr, elastic search seems more practical and it > would be great addition also. > > 4. Graph Compute > > it would be great if we can explore integrate with existing OLAP framework > from TinkerPop so our users not only run OLTP query but also run OLAP query > with data alread stored in s2graph. > > About schedule for next release, I would suggest to build next release > candidate at end of september. > > I don't expect all of above to be included on our next release scope, but I > think they are at least worth to discuss. > > What you guys think? >
