Empirical software evolvability - code smells and human evaluations

  • Authors:
  • Mika V. Mantyla

  • Affiliations:
  • SoberIT, Department of Computer Science, School of Science and Technology, Aalto University, P.O. Box 19210, FI-00760, Finland

  • Venue:
  • ICSM '10 Proceedings of the 2010 IEEE International Conference on Software Maintenance
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Low software evolvability may increase costs of software development for over 30%. In practice, human evaluations and discoveries of software evolvability dictate the actions taken to improve the software evolvability, but the human side has often been ignored in prior research. This dissertation synopsis proposes a new group of code smells called the solution approach, which is based on a study of 563 evolvability issues found in industrial and student code reviews. Solution approach issues require re-thinking of the existing implementation rather than just reorganizing the code through refactoring. This work also contributes to the body of knowledge about software quality assurance practices by confirming that 75% of defects found in code reviews affect software evolvability rather than functionality. We also found evidence indicating that context-specific demographics, i.e., role in organization and code ownership, affect evolvability evaluations, but general demographics, i.e., work experience and education, do not