Approximately 50% of software development cost is devoted to testing. One of the most difficult testing tasks is test selection - developing tests that have potential to uncover errors in a program. This task has been mostly relegated to a human tester. The use of formal specification languages allows the possibility of automatically selecting functional tests based on the specification. An automated approach can greatly expedite the testing process and can reduce the cost of testing. We are investigating ways to automatically select tests from specifications written in a formal specification language called ADL. We also propose to develop the appropriate tool support. We believe that an automated and systematic approach to test selection based on formal specifications can provide better and more consistent functional coverage and thus can improve the dependability of the software under test.
from Debra J. Richardson djr@ics.uci.edu Department of Information and Computer Science, University of California, Irvine CA 92717-3425