Modern computing is characterized, among others, by two major facts. On the one hand, methods are needed to master the ever growing complexity of information systems. On the other hand, distributed and open systems allow configurations to change dynamically and thus require tools to allow processes to cooperate, communicate and work in synergy.
A theoretical model μ 2Log, supported by an efficient implementation, is proposed to cope with these two problems. Following Linda, it offers tuple-based dataspaces as communication and coordination means and manipulates passive and active objects by the same kind of three basic primitives: tell, read, and get to respectively place, check the existence and remove an object. Our model supports multiple local dataspaces, multiple processes acting on these dataspaces as well as a distribution of these dataspaces and processes. Remote coordination is ensured by agents of some platforms granting access via aliases to other agents running on other platforms. As a result, a peculiar feature of our approach is that it neither requires a centralized server nor a centralized dataspace.