An Analysis of Language-Level Support for Self-Adaptive Software

Abstract

Self-adaptive software has become increasingly important to address the new challenges of complex computing systems. To achieve adaptation, software must be designed and implemented by following suitable criteria, methods, and strategies. Past research has been mostly addressing adaptation by developing solutions at the software architecture level. This work, instead, focuses on finer-grain programming language-level solutions. We analyze three main linguistic approaches: metaprogramming, aspect-oriented programming, and context-oriented programming. The first two are general-purpose linguistic mechanisms, whereas the third is a specific and focused approach developed to support context-aware applications. This paradigm provides specialized language-level abstractions to implement dynamic adaptation and modularize behavioral variations in adaptive systems. The article shows how the three approaches can support the implementation of adaptive systems and compares the pros and cons offered by each solution.