Cloning in Customization Classes: A Case of a Worldwide Software Product Line

• Jefferson Barbosa UFC
• Rossana M. C. Andrade UFC
• João Bosco F. Filho UFC
• Carla I. M. Bezerra UFC
• Isaac Barreto FCPC
• Rafael Capilla Universidad Rey Juan Carlos

Resumo

Cloning-and-owning, in the long run, can severely affect evolution, as changes in cloned fragments may require modifications in various parts of the system. This problem scales if cloning is used in classes that derive products in a Software Product Line, because these classes can impact in several features and products. However, it is hard to know to which extent cloning in customization classes can impact in a project. We conduct a study, within an SPL that generates mobile software for over 150 countries, to analyze cloning practices and how cloned parts relate to the maintainability of customization classes. We collect and identify clones inside customization classes during a period of 13 months, involving 70 customization classes and 5 branches. In parallel, we collect the respective issues from the issue tracking tool of the SPL project, obtaining over 140 issues related to customization classes. We then confront the time spent to solve each issue with its nature (i.e., if it relates to cloned code or not). As first result, we verify that issues related to cloning take in average 136% more time to be solved. Our study helps to understand how cloning relates to maintainability in the context of mass customization, giving insights about cloned code evolution and its impacts in a software product line project.

Palavras-chave: Clone, Customization, Software Product Line