Abstract Program Slicing: From Theory towards an Implementation

Program Slicing: From Theory towards an Implementation Isabella Mastroeni and Ðurica Nikolić Dipartimento di Informatica, Università di Verona, Italy [email protected], [email protected] Abstract. In this paper we extend the formal framework proposed by Binkley et al. for representing and comparing forms of program slicing. This framework describes many well-known forms of slicing in a unique formal structure based on (abstract) projections of state trajectories. We use this formal framework for defining a new technique of slicing, called abstract slicing, which aims to slice programs with respect to properties of variables. In this way we are able to extend the original work with three forms of abstract slicing, static, dynamic and conditioned, we show that all existing forms are instantiations of their corresponding abstract forms and we enrich the existing slicing technique hierarchy by inserting these abstract forms of slicing. Furthermore, we provide an algorithmic approach for extracting abstract slices. The algorithm is split into two modules: the simple approach, used for abstract static slicing, and the extended approach, composed of several applications of the simple one, which is used for abstract conditioned slicing. In this paper we extend the formal framework proposed by Binkley et al. for representing and comparing forms of program slicing. This framework describes many well-known forms of slicing in a unique formal structure based on (abstract) projections of state trajectories. We use this formal framework for defining a new technique of slicing, called abstract slicing, which aims to slice programs with respect to properties of variables. In this way we are able to extend the original work with three forms of abstract slicing, static, dynamic and conditioned, we show that all existing forms are instantiations of their corresponding abstract forms and we enrich the existing slicing technique hierarchy by inserting these abstract forms of slicing. Furthermore, we provide an algorithmic approach for extracting abstract slices. The algorithm is split into two modules: the simple approach, used for abstract static slicing, and the extended approach, composed of several applications of the simple one, which is used for abstract conditioned slicing.