Today, we generate sample tests when generating a library. These sample tests call each RPC with an empty request, but don't test any behavior.
These have a long history of being flaky and unreliable. Furthermore, we are essentially just smoke testing our client services (which they should be doing themselves). As a result, I think it is appropriate to cease generating these tests.
Today, we generate sample tests when generating a library. These sample tests call each RPC with an empty request, but don't test any behavior.
These have a long history of being flaky and unreliable. Furthermore, we are essentially just smoke testing our client services (which they should be doing themselves). As a result, I think it is appropriate to cease generating these tests.