Reworking non-applicable tests in a customer´s test suite

Acceptance testing is a crucial stage of software development process. Results of acceptance testing define customer´s perception of the software product and, as such, overall success of the project. Due to importance of acceptance testing, some projects even use acceptance test suites for their internal system testing. Customers often share a common acceptance test suite between several similar projects. Each project may then use just a subset of the generic test suite. Choice of this subset is subject to negotiation. Eventually, some test cases are negotiated as not applicable. Some tests, although declared as not applicable, may still have input conditions which are valid for the system under test. System output for these test cases, although different to initial customer´s expectations, must still conform to some pre-defined rule. A good example is a Java method which, if unsupported, must throw some valid exception. The article describes an approach to reworking a large set of not applicable manual acceptance tests into a smaller and automated test suite. This approach allowed for drastic test cycle time improvement while keeping quality of testing at the same level