Dependence cluster visualization

  • Authors:

  • Syed S. Islam;Jens Krinke;David Binkley

  • Affiliations:

  • King's College London, London, United Kingdom;King's College London, London, United Kingdom;Loyola University Maryland, Baltimore, MD, USA

  • Venue:
  • Proceedings of the 5th international symposium on Software visualization
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Large clusters of mutual dependence have long been regarded as a problem impeding comprehension, testing, maintenance, and reverse engineering. An effective visualization can aid an engineer in addressing the presence of large clusters. Such a visualization is presented. It allows a program's dependence clusters to be considered from an abstract high level down thru a concrete source-level. At the highest level of abstraction, the visualization uses a heat-map (a color scheme) to efficiently overview the clusters found in an entire system. Other levels include three source code views that allow a user to "zoom" in on the clusters starting from the high-level system view, down through a file view, and then onto the actual source code where each cluster can be studied in detail. Also presented are two case studies, the first is the open-source calculator bc and the second is the industrial program copia, which performs signal processing. The studies consider qualitative evaluations of the visualization. From the results, it is seen that the visualization reveals high-level structure of programs and interactions between its components. The results also show that the visualization highlights potential candidates (functions/files) for re-factoring in bc and finds dependence pollution in copia.