Locating distinguishing features using diff sets

  • Authors:
  • Julia Rubin;Marsha Chechik

  • Affiliations:
  • University of Toronto, Canada / IBM Research, Israel;University of Toronto, Canada

  • Venue:
  • Proceedings of the 27th IEEE/ACM International Conference on Automated Software Engineering
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, we focus on the problem of feature location for families of related software products realized via code cloning. Locating code that corresponds to features in such families is an important task in many software development activities, such as support for sharing features between different products of the family or refactoring the code into product line representations that eliminate duplications and facilitate reuse. We suggest two heuristics for improving the accuracy of existing feature location techniques when locating distinguishing features – those that are present in one product variant while absent in another. Our heuristics are based on identifying code regions that have a high potential to implement a feature of interest. We refer to these regions as diff sets and compute them by comparing product variants to each other. We exemplify our approach on a small but realistic example and describe initial evaluation results.