Mainly for Stack as he's working on RPC stuff. I'm just curious what you have in mind for the user-facing client interfaces. Do you plan to keep the Result object the same for now and populate it with KeyValues? Or will client's Result object be changing in conjunction with the protocol buffer stuff?
Longer term, we probably want to let the client access the CellScanner directly so we don't have to explode the encoded bytes into KeyValues. But short term I don't think there's a rush to do that. Maybe 0.96.1 or 0.96.2 could give clients access to both interfaces, and we deprecate the Result object and remove later. Or maybe result object is a good thing to keep forever... I'm not really making a recommendation, just seeing if anyone has thought about it.
