Investigating Intentional Clone Refactoring

Abstract

Software clone refactoring has been studied from many perspectives,including empirical research on clone refactoring history, IDE supportfor tracking clone change, and recommendation systems for clonemanagement.  Most of the work relies on having access to and being ableto analyze the history of clone refactoring. However, refactoring clonedcode is not equivalent to clone management, as code refactoring can bemotivated by goals unrelated to cloning. In this position paper, weintroduce a dataset of intentional clone refactoring, which is producedby keywords matching in commit messages within the version control systemof Linux kernel. By investigating two important clone evolution scenarios--- clone removal and inconsistent changes --- in subsystems of Linuxkernel, we find that intentional clone refactoring accounts for only asmall proportion of all detected clone evolution.