This book mainly presents the theory behind using objects to act as interacting entities in a globally distributed computation, which will lead to greater scalability and system performance. A theoretic foundation is presented to show that scalable distributed object systems cannot be effective without being interactive, based on asynchronous method calls. The proposed calculus is called asynchronous sequential processes (ASP).

The first chapter starts with an analysis of language and concurrency. In the next chapter, by presenting formalism and languages, and especially the methodology to deal with concurrency, the authors show that ambient calculus and &pgr; calculus are not very far apart; they both use expression evaluation and channels as fundamental concepts.

Using a few examples, the authors illustrate the ASP calculus. Determinism has been proven using tree topology and a Fibonacci program. It is true that tree topology is used in programs, but thinking it is used heavily in parallel and distributed contexts is a weak assumption. Further confluent and nonconfluent features are also presented in the textbook. The nondeterministic nature of ASP seems to be less convincing, however; having been proven based on Fibonacci’s fib(n), the final value of s.result should be nondeterministic. It is a good start to prove the nondeterminism, and useful patterns for concurrency. The authors also include migration, groups, and channels for confluence as further extensions. Finally, in support of the theoretic foundations, a proactive implementation is presented.

I believe that the calculus defined for ASP is a balance between asynchronous communications and synchronizations due to wait by necessity. It is a strong basis for parallel and distributed systems. The calculus is certainly locationless, because convergence and determinism are valid whatever the physical location. The book presents a solution for the migration of an activity. At the root of ASP, groups of active objects are made possible by systematic asynchronous communication. The group concept is widely used in the network due to multicast technology.

This calculus mainly presents out-of-order future updates. The theoretical foundation provides a strong basis to argue that the object language could be a future paradigm, one that can scale well in a distributed or a parallel system. However, the scalability could be mainly applied to distributed systems.