White Paper on Reverse Engineering

This paper motivates and describes a research program in the area of reverse engineering being w conducted at the Georgia Institute of Technology. Reverse engineering is an emerging interest area ithin the software engineering field. Software engineering itself is concerned with improving the pro-c ductivity of the software development process and the quality of the systems it produces. However, as urrently practiced, the majority of the software development effort is spent on maintaining existing o m systems rather than developing new ones. Estimates of the proportion of resources and time devoted t aintenance range from 50% to 80%[6]. The greatest part of the software maintenance process is devoted to understanding the system t a being maintained. Fjeldstad and Hamlen report that 47% and 62% of time spent on actual enhancemen nd correction tasks, respectively, are devoted to comprehension activities. These involve reading the documentation, scanning the source code, and understanding the changes to be made[12]. The implications are that if we want to improve software development, we should look at mainte-i nance, and if we want to improve maintenance, we should facilitate the process of comprehending exist ng programs. Reverse engineering provides a direct attack on the program comprehension problem. The process of understanding a program involves reverse engineering the source code. Chikofsky s and Cross[8] give the following definition. "Reverse enginering is the process of analyzing a subject ystem to identify the system's components and their interrelationships and create representations of the u system in another form or at a higher level of abstraction." The purpose of reverse engineering is to nderstand a software system in order to facilitate enhancement, correction, documentation, redesign, or 1 reprogramming in a different programming language. .2. Difficulties Reverse engineering is difficult. It is difficult because it must bridge different worlds. Of particular importance are bridges over the following five gaps. The gap between a problem from some application domain and a solution in some programming language. The gap between the concrete world of physical machines and computer programs and the abstract world of high level descriptions. The gap between the desired coherent and highly structured description of the system and the actual system whose structure may have disintegrated over time.-2-n The gap between the hierarchical world of programs and the associational nature of huma cognition. The gap between the bottom-up analysis of the source code and the top-down synthesis of the 1 description …