Combining Metrics and Graphs for Object Oriented Reverse Engineering

A software system may become very large during its evolution, getting less maintainable while its complexity rises. Since replacing the system with a new one is often out of question because of economic considerations, reengineering techniques are being developed to change the system into a form which makes it easier to maintain and to further develop. However, before a system can be reengineered, it has to be reverse engineered in order to understand its nature and inner logic. This work deals with a lightweight approach to software reverse engineering combining simple graphs with simple object oriented metrics. Our goal is to obtain a simple and scalable graphical display of a system and its parts through which we succeed to visually extract information, which is useful to the understanding of the system and the detection of its design problems. The primary goal of this work is to put up a repository of combinations of graphs and metrics which are useful to reverse engineer an object oriented system. To validate our approach we implemented a tool called CodeCrawler, which can graphically display source code while providing a layer of interactivity to the user: we use the term navigating the code. We ran CodeCrawler on two Smalltalk case studies and one large industrial case study written in C++. The positive experiences and reactions which we obtained are a proof of the usefulness of our idea.