The documentation repeatedly mentions gRPC, although I don't see any existing implementations, so I'm guessing this is a hypothetical at this point, with most of the focus on openapi/openapi3; I want to highlight that gRPC may have some additional implications that may need consideration. In particular, when talking about gRPC, we're usually (although not exclusively) talking about protobuf, which is an ordinal, not nominal, data format. To reliably be able to discuss protobuf models (without them breaking whenever there's any change to the schema), a way is required to nail down the ordinals very robustly. Failure to consider this promptly may significantly hamper any attempts to shoehorn gRPC into cadl at a later date.
I'd be happy to help discuss any relevant topics here; let me know if I can assist
(for context; I'm MSFT FTE in APIM, with extensive protobuf and gRPC knowledge)
The documentation repeatedly mentions gRPC, although I don't see any existing implementations, so I'm guessing this is a hypothetical at this point, with most of the focus on openapi/openapi3; I want to highlight that gRPC may have some additional implications that may need consideration. In particular, when talking about gRPC, we're usually (although not exclusively) talking about protobuf, which is an ordinal, not nominal, data format. To reliably be able to discuss protobuf models (without them breaking whenever there's any change to the schema), a way is required to nail down the ordinals very robustly. Failure to consider this promptly may significantly hamper any attempts to shoehorn gRPC into cadl at a later date.
I'd be happy to help discuss any relevant topics here; let me know if I can assist
(for context; I'm MSFT FTE in APIM, with extensive protobuf and gRPC knowledge)