System dependence graphs based on parse trees and their use in software maintenance

Abstract

The concept of static program slicing was first introduced by Weiser. Ottenstein et al. indicated that an intraprocedural slice can be found in linear time by traversing a suitable graph representation of the program referred to as the program dependence graph (PDG). Horwitz et al. introduced algorithms to construct interprocedural slices by extending the program dependence graph to a supergraph of the PDG referred to as the system dependence graph (SDG). This extension captures the calling context of procedures. In this paper, the SDG is further extended so that it handles a grammar with additional constructs. Algorithms are introduced that permit one to build the SDG so that it is based on a parse tree instead of statements. A new method is described to calculate the transitive dependences and therefore build the SDG. In addition, we demonstrate how the parse tree-based SDG can be used to provide us with “smaller” and, therefore, more precise slices and we show how slices of variables that are used in a statement can be obtained. Finally, the versatility of the SDG as an internal program representation is illustrated by briefly presenting a tool that we have developed that permits slicing, dicing, and ripple analyzing in addition to other software engineering activities to be performed on programs written in a subset of ANSI C.