lidavidm commented on pull request #8962: URL: https://github.com/apache/arrow/pull/8962#issuecomment-748285044
> > Sorry, I was talking about the Java implementation - unilaterally filling in an empty buffer on deserialization when not provided there broke other code which assumed there would be _no_ buffer instead of an empty one. (So what I meant is that while the original code would be fine with any Protobuf implementation that doesn't write out empty fields, it would break with one that did write them out.) > > Yes, I meant that because the C++ client is currently a Protobuf implementation that is writing out an empty field (I think), wouldn't the archery integration tests with the Java server and C++ client on the generated_null test currently be failing? Ah sorry - C++ is fine because it only writes out those bytes conditionally, for message types that require a body (e.g. a RecordBatch). But if an implementation wrote out an (empty) body for a Schema, the Java client/server would break. https://github.com/apache/arrow/blob/25c736d48dc289f457e74d15d05db65f6d539447/cpp/src/arrow/flight/serialization_internal.cc#L274 ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
