ArchTacRV: Detecting and Runtime Verifying Architectural Tactics in Code

A software architectural tactic is a design decision for realizing quality goals at the architectural level. With the evolution of code, the designed architectural tactics might be degraded over time. In practice, the existing systems provide limited support for checking the consistency between an architectural tactic and its implementation. Kim et al. specified the generic structure and interaction behavior for a subset of architectural tactics in Role-Based Meta-modeling Language (RBML) to facilitate the design of tactics. Based on Kim et al.'s work, this paper first presents a machine learning-based method to assist users in detecting the behavior methods of the tactic structure in code, then proposes a runtime verification (RV) method for checking the behavioral consistency between the tactic specification in RBML and its implementation. We conducted experiments for the behavioral methods detection approach by comparing five machine learning models on a dataset with seventy-four open-source projects containing ten types of tactics. For each tactic, we selected an open-source project to show the effectiveness of the RV approach. Finally, we design and implement a prototype tool named ArchTacRV to help developers efficiently maintain the architectural tactics.