Andrew Mohan

A framework for understanding conceptual changes in evolving source code

As systems evolve, they become harder to understand because the implementation of concepts (e.g. business rules) becomes less coherent. To preserve source code comprehensibility, we need to be able to predict how this property will change. This would allow the construction of a tool to suggest what information should be added or clarified (e.g. in comments) to maintain the codes comprehensibility. We propose a framework to characterize types of concept change during evolution. It is derived from an empirical investigation of concept changes in evolving commercial COBOL II files. The framework describes transformations in the geometry and interpretation of regions of source code. We conclude by relating our observations to the types of maintenance performed and suggest how this work could be developed to provide methods for preserving code quality based on comprehensibility.

Programming style changes in evolving source code

As programs evolve they become harder to maintain as the source code becomes more difficult to comprehend. One aspect of comprehensibility is related to the programming style exhibited by the program. However, as this aspect itself is changed during the maintenance process, we must attempt to maintain the original programming style quality to maintain the comprehensibility of the program. This paper presents the initial findings of an investigation of how one area of programming style (typographical style), changes over the lifetime of a program. The aim is to draw upon these findings to indicate how typographical style should be measured, and where necessary changed, to ensure program comprehensibility.

An approach to understanding program comprehensibility using spatial complexity, concept assignment and typographical style

This paper has briefly presented an approach to identifying the comprehensibility of a program and initial results from its application. The results obtained so far, indicate that this approach is useful in modelling the comprehensibility of a program as it evolves. However further work is required to calibrate this approach to more accurately reflect comprehensibility and to identify at what point corrective action should be undertaken to maintain the quality of the program.