Using Dependence Graphs for Slicing Functional Programs

One of the popular ways to perform automated analysis of programs is by manipulating slices — reduced executable programs derived from the originals by removing some steps in such a way that they replicate parts of the original behaviour [18]. Such slices are useful for debugging and comprehension, refactoring and restructuring, reverse engineering and maintenance, model checking, as well as other tasks related to similar problems. At the core of this family of methods are the slicing strategies (backward slicing [18], forward slicing [8], as well as other variations [19], including higher-order lazy functional slicing [13]) and the base data structures. In rare cases analysis, slicing and transformation are performed directly on abstract syntax trees enriched with annotations [3]. However, it is more productive to operate on a data structure specifically designed to facilitate these tasks. Many such structures were proposed in the last three decades: