Comments, with reply, on "Axiomatizing software test data adequacy" by E.J. Weyuker

E.J. Weyuker (ibid., vol.SE-12, p.1128-38, Dec. 1986) recently proposed a set of properties which should be satisfied by any reasonable criterion used to claim that a computer program has been adequately tested. The author called these properties 'axioms'. She also evaluated several well-known testing strategies with respect to these properties, and concluded that some of the commonly used strategies failed to satisfy several of the properties. The commenters question both the fundamental nature of the properties and the precision with which they are presented, and illustrate how a number of ideas in E.J. Weyuker's paper can be simplified and clarified through greater precision and a more consistent set of definitions. They also reanalyze the testing strategies after accounting for these inconsistencies. The strategies tend to work better as a result of this reanalysis. The author rebuts the commenter's arguments.<
>