We have a project where UI code will be developed by the same team but in a different language (Python/Django) from the services layer (REST/Java). The code for each layer exits in different code repositories and which can follow different release cycles. I'm trying to come up with a process that will prevent/reduce breaking changes in the services layer from the perspective of the UI layer.

I've thought to write integration tests at the UI layer level that we'll run whenever we build the UI or the services layer (we're using Jenkins as our CI tool to build the code which is in two Git repos) and if there are failures then something in the services layer broke and the commit is not accepted.

Would it also be a good idea (is it a best practice?) to have the developer of the services layer create and maintain a client library for the REST service that exists in the UI layer that they will update whenever there is a breaking change in their Service API? Conceivably, we would then have the advantage of a statically-typed API that the UI code builds against. If the client library API changes, then the UI code won't compile (so we'll know sooner that there was a breaking change). I'd also still run the integration tests upon building the UI or services layer to further validate that the integration between UI and the service(s) still works.