Architecture-sensitive heuristics for prioritizing critical code anomalies

Abstract

The progressive insertion of code anomalies in evolving software systems might lead to architecture degradation symptoms. Code anomalies are particularly harmful when they contribute to the architecture degradation. Although several approaches have been proposed aiming to detect anomalies in the source code, most of them fail to assist developers when prioritizing code anomalies critical to the architectural design. Blueprints of the architecture design are artifacts often available in industry software projects. However, such blueprints are rarely explored to support the prioritization of code anomalies in terms of their architecture relevance. This paper proposes and evaluates 2 sets of blueprint-based heuristics for supporting the prioritization of critical code anomalies. The prioritization is based on their potential impact on revealing architectural drift problems. The heuristics allow developers prioritizing critical code anomalies by exploiting architectural information provided in the blueprint. The contributions of this paper include: (i) a set of architecture sensitive heuristics to support developers when prioritizing critical code anomalies; (ii) an evaluation of the proposed heuristics in terms of their prioritization accuracy in 3 systems; and (iii) an empirical analysis on how the blueprints’ information might enhance the prioritization of critical code anomalies, as opposed to existing heuristics strictly based on source code analysis.