Assessing the effect of source code characteristics on changeability

Maintenance is the phase of the software lifecycle that comprises any modification after the delivery of an application. Modifications during this phase include correcting faults, improving internal attributes, as well as adapting the application to different environments. As application knowledge and architectural integrity degrade over time, so does the facility with which changes to the application are introduced. Thus, eliminating source code that presents characteristics that hamper maintenance becomes necessary if the application is to evolve. We group these characteristics under the term Source Code Issues. Even though there is support for detecting Source Code Issues, the extent of their harmfulness for maintenance remains unknown. One of the most studied Source Code Issue is cloning. Clones are duplicated code, usually created as programmers copy, paste, and customize existing source code. However, there is no agreement on the harmfulness of clones. This thesis proposes and follows a novel methodology to assess the effect of clones on the changeability of methods. Changeability is the ease with which a source code entity is modified. It is assessed through metrics calculated from the history of changes of the methods. The impact of clones on the changeability of methods is measured by comparing the metrics of methods that contain clones to those that do not. Source code characteristics are then tested to establish whether they are endemic of methods whose changeability decay increase when cloned. In addition to findings on the harmfulness of cloning, this thesis contributes a methodology that can be applied to assess the harmfulness of other Source Code Issues. The contributions of this thesis are twofold. First, the findings answer the question about the harmfulness of clones on changeability by showing that cloned methods are more likely to change, and that some cloned methods have significantly higher changeability decay when cloned. Furthermore, it offers a characterization of such harmful clones. Second, the methodology provides a guide to analyze the effect of Source Code Characteristics in changeability; and therefore, can be adapted for other Source Code Issues.