Tracking the evolution of code clones

  • Authors:

  • Tibor Bakota

  • Affiliations:

  • University of Szeged, Hungary

  • Venue:
  • SOFSEM'11 Proceedings of the 37th international conference on Current trends in theory and practice of computer science
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

It is believed by many academic and industrial experts, that source code cloning (copy&paste programming) represents a significant threat to maintainability in an evolving software system. The real threat does not lie in the existence of duplications, but the fears are in connection with their evolution. There exist an abundance of algorithms for finding code clones in one particular version of a software system, but eliminating or even evaluating these clones often seems hopeless, as there may exist several thousands of them. Tracking the evolution of individual clones might solve the problem, as it could help identifying the inconsistently changing duplications: the clones which are really dangerous at a particular moment. In this paper we present an approach for mapping code duplications from one particular version of the software to another one, based on a similarity distance function. For the suspicious evolution patterns we introduce the term of "clone smells". By defining the relevant categories of the possible evolution patterns, the proposed method also gives a clue about why the reported code fragments might be dangerous. In the case study, clone smells were extracted, evaluated, and manually categorized throughout many versions of the jEdit system. The findings suggest that roughly half of the reported smells refer to inconsistent changes in the code.