Roadmap

[kafkiansky]

• Error handling
  We must enable the server to transmit a detailed error using the following message schema in the grpc-status-details-bin trailer (see the protocol for details), and the client to expose it as an exception.

• Service discovery
  We must implement the ability to discover services by their name using DNS or any other service discovery mechanism.

• Server reflection
  We need to enable the server to inform the client about which RPCs it supports, according to this proposal. This task will require modifications on the protoc-plugin side to generate classes that store the proto schema description in a binary format, compressed with gzip, for example. On the gRPC side, it will be necessary to generate reflection services and support retrieving the schema from the generated code.

  • Implement Descriptor Pool generation protoc-plugin#3
  • Implement Descriptor Pool protobuf#2
  • Optimize descriptor pool's memory consumption protoc-plugin#4

• Health
  We must generate health services to allow other systems, such as k8s, to monitor application status using the gRPC protocol.

• Service config
  See this document for details.

• Load balancing
  The client must be able to select the appropriate server according to various load balancing algorithms.

• Unknown fields
  Add the ability for the user to access unknown fields after deserializing a protobuf message. Consider using a WeakMap tied to the message, which should be cleaned up by gc after request processing.