Model‐based design of reverse engineering tools

Abstract

SUMMARY Tools built in an ad hoc way and without proper models often display problems for both tool users and designers. Firstly, without systematic analysis and good understanding of the underlying software process model, we have little chance to design a tool that will adequately address user’s needs. Next, because one tool is often used in many different situations and by people who have different working habits, tools should be flexible and allow a user to customise tools functionalities. Ad hoc built tools usually are not flexible enough, as possible variations in tool functions had not been incorporated into the tool architecture to make future customisations possible. Finally, ad hoc design practice does not lead to accumulating the tool design know-how, makes it difficult to repeat successful solutions and slows down the process of understanding and improving tool design methods. We applied conceptual modelling in design of tools for software maintenance to alleviate some of the above problems. In this paper, we describe a model-based method to designing reverse engineering tools. The design starts by modelling low level source program design models, higher level design models to be recovered and heuristic rules a reverse engineering tool uses to recover higher level designs from lower level designs. On one hand, conceptual models lead to better understanding of tool requirements. On the other hand, model-based approach leads to the design of a generic design abstractor, a component of a reverse engineering tool that evaluates reverse engineering heuristics. A generic design abstractor adds flexibility to reverse engineering tools in two ways: (1) we can customise the generic design abstractor to meet requirements of a reverse engineering project in hand and (2) a programmer (an end-user of a reverse engineering tool) can define new reverse engineering heuristics and tune in recovered designs.