The benefits of impact analysis in the maintenance and evolution of software systems are well known, and many forms of impact analysis, over different software life cycle objects, have been proposed. However, one form of impact from software change has yet to be explored by the research community: these are the impacts of changes to software on data. In particular, when the business rules enforced by a system change, it may be necessary to perform some cleanup or transformations on persistent data, in order to bring it into line with the new system functionality. Alternatively, the proposed new rule may itself have to be modified, if the data impacts are too costly to address. In this paper, we show how such impacts can arise, and propose an approach to assisting in their identification through user-driven exploration of hypothetical change-impact scenarios.
Suzanne M. Embury, David Willmor, Lei Dang