Property Based Dynamic Slicing of Object Oriented Programs

Abstract

Slicing is used for program analysis. It the process of extracting the statements of a program that are relevant to a given computation. Static slicing generates slices for all possible execution of a program helping in program understanding, verification, maintenance and testing. Dynamic slices are smaller in size as they extract slices for a given execution of a program and helps in interactive applications like debugging and testing. With the wide spread use of object oriented software, there are many papers on Dynamic Slicing of object oriented programs but few papers only address in details about the most basic features of Object Oriented Programming that is class definition, Object creation, accessing object through reference, invoking methods of a class , polymorphism, inheritance etc. From last three decades many algorithms have been designed to slice a program with respect to the syntax of the program. The real world object oriented programs consists of thousands of lines of code. Traditional Syntax based slices for program variables used at many places in a program are generally large even for dynamic slices. Recently, some work has been done to get slices based on abstract/concrete properties of program variables. For smooth debugging and testing, the slice will be small if any particular property is being considered (semantics based). Most of the semantics based slicing algorithms have focused on finding static slices on the abstract properties by using SSA as intermediate representation and extract slices by storing an execution trace of a program. To the best of our knowledge generating dynamic slices based on abstract/Concrete properties of program variables is scarcely reported in literature. In this paper we present an algorithm for generating dynamic abstract slices of object oriented programs addressing all key object oriented features.