Toward Languages and Formal Systems for Distributed Computing

Therefore, an object can guarantee its survivability under circumstances where the objects cannot have a unique global view. Moreover, it introduces the notion of timing polymorphism that allows a server object to dynamically select one execution body (a method) that can be executed within the specied timing constraint. DROL is the rst step toward a programming language that incorporates the notion of object autonomy. For details, refer to [6] and [7]. Especially, we discussed the realization of the best eort property in [7]. We plan to implement it on the Apertos Operating System [13] in the near future. We also presented the possibility of formalizing distributed computation. We note particularly the following two criteria: whether communication is synchronous or asynchronous, and whether objects refer to the same clock or dierent clocks. Based on these criteria, we classied distributed computation into four forms. As a rst step, we proposed two formal models for concurrent real-time computations and real-time computations with dierent clocks. These models are extensions of CCS with the notion of global and local time, and thus can provide a powerful means of reasoning about globally and locally temporal properties and behavioral properties in the computations. We presented some approaches for describing asynchronous real-time computing. It is important to note that the synchronous real-time property of a DROL program can be validated by using RtCCS and DtCCS. We are also developing a static verication framework for DROL based on RtCCS and DtCCS. Particularly , since DtCCS can analyze locally temporal properties, it provides a very powerful method of verifying DROL's applications for distributed systems where objects cannot share the same global clock, thus must follow dierent clocks. A short remark was also given to relate the notion of asynchrony and real-time with a higher level abstraction of distributed computing called the Computational Field Model. We feel that we have just come to the starting point for investigating distributed objects.