Effects of source code regularity on software maintainability: An empirical study

Identifying the subset of a software system that is affected by a change task is an essential part of software maintenance and is traditionally addressed through model-source code tracing. However, due to the high costs associated with practicing traceability procedures, producing traceable systems is often not regarded as a feasible option in the software industry. As a cost-effective alternative to establishing and maintaining low-level trace information on software systems, we propose to improve maintainability of software systems through the regularity property of a system's source code, which can be achieved by consistently observing a set of design rules, called requirement component traceability rules, throughout the system. To understand how this form of design uniformity in a system's source code influences maintainability, we conducted an empirical study of five developers performing various change subset identification tasks on two functionally equivalent sets of source codes with different degrees of regularity. High regularity is shown to significantly improve software maintainability. We express our findings as a number of detailed hypotheses and a theory of the effects of source code regularity.