GNIREENIGNE: Reverse Engineering or Backwards Engineering?

Reverse engineering involves extracting high-level information from existing code. Reverse engineering is a crucial part of software maintenance since a maintainer needs to understand the code before attempting any modiication. Many eeorts have been made to automatically extract higher-level design information from the code to facilitate various maintenance activities. But many of them are limited to extracting and providing only static information like call graphs, and normally involve techniques like parsing. Some of them also try to \understand" the program in a limited toy domain so that it can generate a high-level description of the system in terms of the common programming concepts. Also, most of the systems do not have any knowledge of the domain. Only recently people have realized the limitations of current approaches and have begun advocating systems that will provide descriptions in terms of human-oriented domain concepts. This paper proposes a new approach to reverse engineering large software systems using granularity-based program recognition as the main recognition mechanism. The approach presented here is robust, exible and scalable. The granularity formalism can represent human oriented strategic knowledge better than other formalisms. The central message of the paper is that we need to take the human into account when designing any software reverse engineering tools. Some of the implications of the proposed approach for future research in the eld of reverse engineering are discussed. Also included is a discussion of how software reverse engineering provides an excellent domain where one can test diierent ideas about granularity research.