Program slicing based on sentence executability

Abstract

We present point slicing, a new slicing technique for imperative programs that gives an answer to the question Which sentences can be executed if sentence p is executed?, very common in program testing, debugging, and understanding tasks and, as far as we know, not directly addressed by other slicing techniques. Point slicing uses a program point as criterion and computes slices by deleting sentences that are proved to be not reachables by executions including the criterion point. We also show how to extend point slicing criterion to a set of program points and how the new technique can be also used to answer to a more precise question: Which sentences are possibly executed if sentence p is executed in a program state satisfying condition Á? Because, minimal point slices are, in general, not computable, we provide definitions of safe approximations for each type of point slice.