Github user okram commented on the issue:
https://github.com/apache/tinkerpop/pull/705
Yes. Exactly.
Here is the thing is -- I don't really care for TinkerPop3 anymore so it
can go as people want it too, I won't fight. But if you want to start aligning
concepts so TinkerPop4 is done "right," we need to start getting away from
methods on `Graph` as users will not have access to `Graph` only `Connections`.
Thus, every action/process is a `Traversal` and thus, all notions of cloning,
clearing, etc. should be via compilation/strategy_introspection of the
`Traversal`.
"But Marko, what about schema stuff like `createIndex`? How will that work
in TinkerPop4" ... That is a vendor specific schema DSL.
`schema.index("name")`, etc.
I think instead of just going "willy nilly" with TinkerPop3, we should
start to make decisions on its design that will align best with TinkerPop4 so
we start to get the right thinking in place now for the big effort to come.
---
