Automatic detection of architectural violations in evolutionary systems

Abstract

Software applications evolve over the years at a cost: their architecture modularity tends to be degraded. This happens mainly because software application maintenance often leads to architectural degradation. In this context, software architects need to elaborate strategies for detecting architectural degradation symptoms and thus maintaining the software architectural quality. The elaborations of these strategies often rely on tools with domain-specific languages (DSLs), which help them to specify software architecture rules. These tools also enforce the adherence of these rules in the evolving program. However, their adoption in mainstream software development is largely dependent on the usability of the language. Unfortunately, it is also often hard to identify their usability strengths and weaknesses early, as there is no guidance on how to objectively reveal them. Usability is a multi-faceted quality characteristic, which is challenging to quantify before a DSL is actually used by its stakeholders. There is even less support and experience on how to quantitatively evaluate the usability of DSLs used in software maintenance tasks. To this end in this dissertation, a usability measurement framework was developed based on the Cognitive Dimensions of Notations (CDN). The framework was evaluated both qualitatively and quantitatively using two textual DSLs for architecture rules in the context of two evolving object-oriented systems. The results suggested that the proposed metrics were useful: (1) to early identify the DSL usability limitations to be addressed, (2) to reveal specific features of the DSLs favoring software maintenance tasks, and (3) to successfully analyze eight usability dimensions that are critical in many DSLs. However, along with these results this evaluation also revealed that this kind of tools lack support for communication among the stakeholders, creating a gap in the software development. To solve this problem we proposed heuristics for tools that use DSLs for detecting architecture degradation symptoms. These heuristics will permit the exchange of information between the stakeholders, thereby, also increasing the tool usability. Finally, we chose TamDera as the tool to implement these heuristics in our study domain. Therefore, we implemented in the new version of TamDera the communication support for the stakeholders by using a new architecture and a new environment with the developed heuristics.