Le 29/09/2019 à 19:59, Jacques Nadeau a écrit : > > It seems like you're saying: "flatbuffers is too complex an encoding, let's > create a new encoding".
Most of the spec is a plain C-level struct in the native ABI, so it avoids any kind of encoding issue. And, yes, flatbuffers must be dealt with and it's not something most people are used to. > You're talking about three separate things: > 1) How do I communicate schema > 2) How do I communicate encoded batches with zero copy > 3) How do I define rules around ownership semantics [...] > I agree for #1 but don't agree for #2 What exactly don't you agree with? > and think we > should do our best to avoid an alternative for #2. We don't have something > that defines #3 so that is clearly a gap. The definition should be done > with all languages in mind. Well, is there anything that doesn't work well with Java here? It would be good to know, because you don't seem to be attacking any specific aspect of the spec. The spec is designed to work well with any language where a decent C FFI is available. > To introduce this, I also think we need to have reference > tools/implementation/integration tests for both C++ and Java. I will definitely do the C++ part. I'm afraid I'm not qualified to do the Java part, though. Regards Antoine.
