Effects of cloned code on software maintainability: A replicated developer study

Abstract

Code clones are a common occurrence in most software systems. Their presence is believed to have an effect on the maintenance process. Although these effects have been previously studied, there is not yet a conclusive result. This paper describes an extended replication of a controlled experiment (i.e. a strict replication with an additional task) that analyzes the effects of cloned bugs (i.e. bugs in cloned code) on the program comprehension of programmers. In the strict replication portion, the study participants attempted to isolate and fix two types of bugs, cloned and non-cloned, in one of two small systems. In the extension of the original study, we provided the participants with a clone report describing the location of all cloned code in the other system and asked them to again isolate and fix cloned and non-cloned bugs. The results of the original study showed that cloned bugs were not significantly more difficult to maintain than non-cloned bugs. Conversely, the results of the replication showed that it was significantly more difficult to correctly fix a cloned bug than a non-cloned bug. But, there was no significant difference in the amount of time required to fix a cloned bug vs. a non-cloned bug. Finally, the results of the study extension showed that programmers performed significantly better when given clone information than without clone information.