Strongly typed integration - compile-time check

Developer Experience

Take Google mapping API as an example.

• User wants to talk to Google mapping API, LB asks for swagger file
• If they provide, LB can do the introspection to generate all the artifacts.
• If they don't, LB can go with the traditional REST connector approach, asking them to specify the template. The purpose of the template is to describe the REST api.
• If they want to re-expose the API, then LB can guide them to generate the controller.

What's involved

• tooling to:
  • introspect the service spec
  • generate the corresponding LB artifacts, e.g. TS interface definition and model definition.
  • create service / service proxy. Similar to LB3 lb soap command

Going back to the Google mapping api, it will end up with google mapping service -> Google Map service interface / class. Then the controller can delegate its calls to its strongly typed methods.

If users want to re-expose the service API, LB needs to have some tooling / convention to help them.

Questions to answer

Spike: handwrite to see how the generated artifacts work together with the controller.
If we can handwrite the generated artifacts, then we can automate it.

• Starting point: write sample app which has controller that tries to talk to REST / SOAP services.
  Inside the controller, you can import the type information of the service. Make method call which will be translated to the juggler / connector which will invoke the backend services.
• tooling: if you receive datasource configuration (URL/path to the OpenAPI spec/template for REST API/ URL/path to WSDL for SOAP), then you should have enough info to introspect the service spec, and generate the corresponding LB artifacts.
  • scope of tooling for the spike: wireframe/mockup of how the tooling can look like, i.e. what to prompt users.

Note:

• task for strongly typed and weakly typed can go in parallel.
• strongly typed would be a stretch goal of GA
• cross-posting from Type safety for services #1268
  • Introduce CLI tooling to discover service and introspect the service specifications and generate the corresponding TypeScript artifacts
  • Make this mapping pluggable and extensible to support different api spec flavours (swagger, wsdl, openapi, grpc, graphql, etc)