krpc
0.4.0indexedGenerates and consumes gRPC services using interfaces and annotations. Supports various RPC call types, code generation with KSP, and relies on the grpc-kotlin-stub runtime.
6
Stars
—
Used by
dependents
—
Health
/ 100
Kotlin RPC
A Kotlin Multiplatform library to generate and consume gRPC services directly from Kotlin. Check out the Sample project to see a full list of available features, or see the Quick Start guide on how to add kRPC to your project.
Note: At the moment, only Kotlin/JVM is supported as a target platform for code generation.
Overview
Instead of using Protoc and .proto files to define services, kRPC uses Kotlin's interfaces and annotations. Any
interface annotated with @Service will be recognized as a definition by the kRPC compiler:
@Service
interface GpsService {
@UnaryCall
suspend fun locationForVehicle(vehicle: Vehicle): Location
}
The methods inside the service definitions should have the proper annotation to reflect the type of RPC call:
@UnaryCalldefines a single-request, single-response rpc method, suspending until the server replies.@ServerStreamtakes a single request and returns a responseFlow, which you can later use to consume the responses streamed by the server. Server-stream methods don't suspend, instead the returned flow collector will suspend until the first value is received.
Method parameters and return types must be serializable.
The compiler will then generate two implementations of this interface: an abstract service provider, and a client. Additionally, a helper class is generated: the service's Descriptor, which contains all the generated method descriptors (on JVM) and can be used to create your own custom client/server implementations.
Implementation
The kRPC compiler uses Google's KSP to process the annotations and then generates code with KotlinPoet. On JVM, the runtime and generated stubs are implemented on top of the grpc-kotlin-stub runtime.
Resources
If you find the project useful, have an idea on how to improve it or if you find a bug, consider making a contribution or opening a new issue. Pull requests are welcome.