I have the next problem. "A" library contains custom JSF components. "B" library contains Selenium tests for these custom components, and some others (Primefaces JSF components).
When I develop "A", I want to test any change using "B" in order to check if I've broken something. So "A" needs "B" dependency in test scope.
When I develop "B", I want to test any change using "A" for deploying a website on the fly at testing-time in order to check if I've broken something. So "B" needs "A" dependency in test scope.
So, How is the best way to avoid the cyclic dependency and get my goal?
you may have to choose between:
having B tests with "A app" like app but an app that is not A at all to avoid having A in B dep.
having a new C test module that have A and B as dependencies. Where C is testing B with A. But this case begin to be less easy to maintain...
In general you dont have the need to test your test tool except if your tool make some complex stuff. In this case I do some small test sets that are autonomous.