Testing
Currently we try to apply a comparable test framework to each project. The major testing used are the System Level Tests described in the Documentation Guidelines, which correspond to Use Cases from the Use Case Model. We have made some additions to JUnit's testing mechanism to generate an xml result upon running our test suites. This xml result is parsed by a web page to provide an HTML page that looks like JUnit's test results from within TRAC (our currently preferred ticket trackign software). The idea is that every commit should have tests run, and thus each commit can be traced to test results so one can verify that tests passed upon commit (or verify which tests fail).
The System level tests are generated directly from Use Cases, with variations dictated Use Case Extensions (this can go both ways; the code informing the UCM is valid). At a high level they represent Dispatchers in the code, and more specifically the used Commands (which represent low level steps or groupings of steps from Use Cases).