Lightweight impact analysis using island grammars

Abstract

Impact analysis is needed for the planning and estimation of software maintenance projects. Traditional impact analysis techniques tend to be too expensive for this phase, so there is need for more lightweight approaches. We present a technique for the generation of lightweight impact analyzers from island grammars. We demonstrate this technique using a real-world case study in which we describe how island grammars can be used to find account numbers in the software portfolio of a large bank. We show how we have implemented this analysis and achieved lightweightness using a reusable generative framework for impact analyzers.