| Software must evolve over time or it becomes useless. Much of software production today is involved not in creating wholly new code from scratch but in maintaining and building upon existing code. Much of this code resides in old legacy software systems.; Unfortunately, these systems are often poorly documented. Typically, they become more complex and difficult to understand over time. Thus, there is a need to better understand existing software systems. An approach toward this problem would be a first step toward easing changes and extending the continuous evolution of these systems.; This dissertation addresses the problem by enabling continuous software that are carried forward during evolution. The proposed approach seeks to redocument existing software structure, capture the analysis decisions made, and support personal, customizable, and live perspectives of the software in an online journal called the Reverse Engineering Notebook.; The premise that software reverse engineering be applied continuously through-out the lifetime of the software has major tool design implications. Thus, tool integration, process, and adoption are key issues for the Notebook. In particular, data integration requirements, control integration via pervasive scripting, presentation integration through the management of views, user roles, methodology, end user needs, and goal-directed framework for the Notebook are described.; A major theme of the dissertation is learning from the successes and failures of studies involving tool integration and reverse engineering technologies. Case studies and user experiments helped to evaluate various aspects of the Notebook approach and provide feedback into software understanding tool requirements. |
