I have a protocol buffer
file:
syntax = "proto3";
package v1api;
option java_multiple_files = true;
option java_package = "myApp.v1";
option java_outer_classname = "V1";
service API {
rpc Login(LoginRequest) returns (LoginResponse)
}
message LoginRequest {
int pin = 1
}
message LoginResponse {
string token = 1
}
I have my server written in Go (a language that can return multiple values), and my client being an Android application.
When I use this protoBuf to generate Go code for server, it is something like:
...
func (c *aPIClient) Login(ctx context.Context, in *LoginRequest, opts ...grpc.CallOption) (*LoginResponse, error) {
out := new(LoginResponse)
err := grpc.Invoke(ctx, "/v1api.API/Login", in, out, c.cc, opts...)
if err != nil {
return nil, err
}
return out, nil
}
...
Notice that there are two return values, (*LoginResponse, error)
.
Now, when I use this protoBuf to generate Java code for my Android side, I get something like:
public myApp.v1.LoginResponse login(myApp.v1.LoginRequest request) {
return blockingUnaryCall(getChannel(), METHOD_LOGIN, getCallOptions(), request);
}
Notice that, here, there is just one return value like Java allows, myApp.v1.LoginResponse
.
My question is, in case of an error returned by the server (something like: return nil, err
), how will I get that second return value in my Android side.
In Go, the error returned by your Login
method will get converted into the wire format status code
and message
used by gRPC. In your case, if you return a status.Status, the Code and message will be sent and show up on the Java side as a StatusRuntimeException. If you return a non Status
error, gRPC will convert it to the UNKNOWN
code, but preserve the message.
This is the reason there are two return values for Go, vs. just one for Java. Java can "return" a secondary value by throwing an exception in its place.
Collected from the Internet
Please contact [email protected] to delete if infringement.
Comments