A Framework for Aspect-Oriented Programming of Embedded Systems
Commercial programming frameworks for embedded software that enable aspect-oriented software development and property-based debugging will require semantically rich deep-structure program representations. Such representations convey the preprogrammed generic analyses, and enable the subsequent application-driven analyses and transformations, that are needed to support weaving and composition of crosscutting software aspects. In contrast, it has been expedient, until now, for researchers in aspect-oriented programming to use only the simplest surface-structure representations of programs.
Research in aspect-oriented programming, which is still in its infancy, will advance considerably in the coming years, spurred on, for example, by DARPA's investments in the PCES project. We fully expect that effective mechanisms for aspects, such as concurrency, efficiency, effective use of the memory hierarchy, etc., will emerge in the next 2-4 years. Tools that support the automation of such mechanisms will have the potential to cut costs and errors dramatically throughout the embedded systems arena.
Under an STTR Phase I contract of DARPA's Program Composition for Embedded Systems (PCES) Project, GrammaTech and its partner the University of Wisconsin (Madison) are studying how GrammaTech's dependence-graph technology can be adapted and applied for aspect-oriented programming.
Application of Horwitz-Prins-Reps program integration (which is based on program dependence graphs and slicing [abstract; paper]) to the problem of merging independently woven aspects has particular appeal. Its potential benefits include automatic detection of interference between aspects, automatic merging of non-interfering aspects, and reduced occurrences of aspect interference.
Related Links:
- DARPA PCES program home page (www.sainc.com/pces).
- AspectC and its Tool Support (PCES February Meeting).