Supporting impact analysis by program dependence graph based forward slicing

Abstract

Since software must evolve to meet the typically changing requirements, source code modifications can not be avoided. Impact analysis is one of the central and relatively demanding tasks of software maintenance. It is constantly needed while aiming at ensuring the correctness of the made modifications. Due to its importance and challenging nature automated support techniques are required. Theoretically, forward slicing is a very suitable technique for that purpose. Therefore, we have implemented a program dependence graph (PDG) based tool, called GRACE, for it. For example, due to the typical rewritings of Visual Basic programs there is a great need to support their impact analysis. However, there were neither earlier scientific studies on slicing Visual Basic nor reported slicers for it. In case of forward slicing there is a need to perform efficient static slicing revealing all the potential effects of considered source code modifications. Use of PDGs helps in achieving this goal. Therefore, this paper focuses on describing automated PDG-based forward slicing for impact analysis support of Visual Basic programs. GRACE contains a parser, a PDG-generator and all other necessary components to support forward slicing. Our experiences on the application of the PDGbased forward slicing has confirmed the feasibility of the approach in this context. GRACE is also compared to other forward slicing tools. © 2007 Springer.