RSpec "ideological" question: how to separate between test description and implementation

Read a lot of materials regarding the behavior-driven testing. Indeed, as it is stated (for example) here, the specs' code should be extremely readable, short and straightforward.

Now I have a question: suppose there are two kinds of team players: one is 开发者_C百科the developer, the other is a tester (QA). What would be the best test plan if not describe and it stubs, where all the its are marked as pending? So, the QA might want to create such a test plan and pass it to the developer for the proper implementation.

After a few days, the tester (or product manager) wants to have a look on the tests. But... there is a hell of a code in there.

Hence, the question: is there a way to separate between the behavior specification and implementation in RSpec? I can fabricate a plenty of methods, but before inventing the wheel, I would rather prefer to stick to some common, proven and well-known practice.

as according to The RSpec Book RSpec is what developers manage for them internally, but Cucumber is what production manager and business can see. Think about RSpec as like internal tool to code coverage to see the objects internal behaviour: models, controllers, views. Other team members can see overall pictures that Cucumbers covers as well