Author(s)

S.T. Holmes, B.F. Jones & D.E. Eyres

Abstract

An improved strategy for the automatic generation of test data S.T. Holmes", B.F. Jones\ D.E. Eyres" "Department of Computer Studies, University of Glamorgan, Ponfi/pHdd, Mzd GZamorgaTi, CF57 JDI, CW * Department of Electronics and Information Technology, University of Glamorgan, UK ABSTRACT Mutation analysis is used to check the efficacy of automatically generated test data. A program is mutated by making a single, syntactically correct change such as deleting a statement or changing the predicate in a selection. The mutation adequacy is defined to be the percentage of all mutants killed by the test data. Our strategy for automatically generating test data uses direct assignment and alternating variable heuristics to kill 79.2% of the mutants. Minor changes to predicates in if- statements are best revealed by generating test data which fall close to the domain boundary for that predicate using linear extrapolation of previous tests. Having found the domain boundary, a heurist

Keywords