Early detection of ripple propagation in evolving software systems

Abstract

Ripple effect analysis is the analysis of the consequential knock on effects of a change to a software system. In the first part of this study, ripple effect analysis methods are classified into several categories based on the types of information the methods analyse and produce. A comparative and analytical study of methods from these categories was performed in an attempt to assist maintainers in the selection of ripple effect analysis methods for use in different phases of the software maintenance process. It was observed that existing methods are most usable in the later stages of the software maintenance process and not at an early stage when strategic decisions concerning project scheduling are made. The second part of the work, addresses itself to the problem of tracing the ripple effect of a change, at a stage earlier in the maintenance process than existing ripple effect analysis methods allow. Particular emphasis is placed upon the development of ripple effect analysis methods for analysing system documentation. The ripple effect analysis methods described in this thesis involve manipulating a novel graph theory model called a Ripple Propagation Graph. The model is based on the thematic structure of documentation, previous release information and expert judgement concerning potential ripple effects. In the third part of the study the Ripple Propagation Graph model and the analysis methods are applied and evaluated, using examples of documentation structure and a major case study.