Programming language implementations for context-oriented self-adaptive systems

Cardozo, Nicolás [Systems and Computing Engineering Department, Universidad de los Andes, FLAG Laboratory, Bogotá, Colombia] Mens, Kim [UCL]

Context: The context-oriented programming paradigm is designed to enable self-adaptation, or dynamic behaviour modification of software systems, in response to changes in their surrounding environment. Context-oriented programming offers an adaptation model, from a programming language perspective, that maintains a clean modularisation between the application and adaptation logic, as well as between the components providing adaptations. Objective: We use three implementation techniques for context-oriented programming languages to assess their appropriateness to foster self-adaptive systems. These approaches take advantage of the capabilities offered by the host programming language to realize self-adaptation as proposed by context-oriented languages. Method: We evaluate each of these approaches by assessing their modularity and complexity when defining adaptations, and by comparing their run-time performance on a simple benchmark. Results: Our results show a higher modularity than that for common architecture-based self-adaptive systems, while maintaining comparable performance. Conclusion: We conclude that context-oriented programming is an appropriate paradigm to realize self-adaptation.