Abstract
Real-time distributed applications based on object-oriented technology raise many complex issues, most of them still open.We elaborate on the reasons why traditional paradigms, found adequate for tackling simple problems, cannot help in addressing these complex issues. A detailed illustration is given with a problem in modular avionics, which has been addressed by resorting to proof-based system engineering, an emerging discipline aimed at coping with real-world problems complexity. Drawing from our experience with partners in industry, we report on why such paradigms and technologies as object-orientation, distributed transactions, distributed architectures, transactional monitors, can be contemplated for the construction of real-time distributed systems, provided that appropriate on-line decision-making algorithms and protocols are resorted to.