Abstract: Specification-based testing enables us to detect errors
in the implementation of functions defined in given specifications.
Its effectiveness in achieving high path coverage and efficiency in
generating test cases are always major concerns of testers. The automatic
test cases generation approach based on formal specifications
proposed by Liu and Nakajima is aimed at ensuring high effectiveness
and efficiency, but this approach has not been empirically assessed.
In this paper, we present an experiment for assessing Liu-s testing
approach. The result indicates that this testing approach may not be
effective in some circumstances. We discuss the result, analyse the
specific causes for the ineffectiveness, and describe some suggestions
for improvement.