Runtime Constraint Checking Approaches for OCL, A Critical Comparison

There are many benefits of checking design constraints at runtime---for example, automatic detection of design drift or corrosion. However, there is no comparative analysis of different approaches although such an analysis could provide a sound basis for determining the appropriateness of one approach over the others. In this paper we conduct a comparative analysis and evaluation of different constraint checking approaches possible for the Object Constraint Language (OCL). We compare several approaches including (1) direct translation to implementation languages, (2) use of executable assertion languages, and (3) use of aspect-oriented programming languages. Our comparison includes both quantitative metrics such as runtime performance and qualitative metrics such as maintainability of constraint checking code. We found that there is no universal approach; for the production use the implementation language-based approaches perform better if memory footprint or runtime speed is important, and for the development use the other approaches are more appealing if maintainability is more important