Identifying software architecture erosion through code comments

Abstract

Software architecture erosion or the as-implemented architecture is not complying with the as-intended architecture is one of the major problems faced by many organizations. There is no easy way to trace design decisions or tracking back or reconstructing those decisions by looking at the source code level elements is one of the major reasons for software architecture erosion. Other than that the mistakes or carelessness of the programmer may lead the system to an eroded status eventually. Lack of domain knowledge, lack of knowledge about intended architecture and unable to identify possible violations of as-intended architecture (by identifying architectural degradation) are some other reasons for software architecture erosion. There are various methodologies and tools for architecture conformance checking and analyzing the static architecture and provide comparison results which can be used to determine whether the architecture of a system is altered or not [10]. Most of them require high end tool support and providing the implemented architecture and the intended architecture each time the analysis needs to done. As the main research objective it identified a missing area of software architecture conformance checking methodologies and analyzed and identified a way to prevent software architecture erosion using that. This research is more focused on unconventional usability of the code comments and how it can be leveraged to capture the architecture of the application and how it can be used as an effective architecture conformance checking mechanism. This research states a methodology which uses Java Doc comments to inject architecture specific information into the code base and a mechanism to capture them and compare them with a pre-defined architecture rule set. An empirical and theoretical evaluation has been done to prove this concept actually works in real life scenarios. It opened up a new area of architecture conformance checking to the future researchers of the field of software