Title:

Calculi for higher order communicating systems

This thesis develops two Calculi for Higher Order Communicating Systems. Both calculi consider sending and receiving processes to be as fundamental as nondeterminism and parallel composition. The first calculus called CHOCS is an extension of Milner's CCS in the sense that all the constructions of CCS are included or may be derived from more fundamental constructs. Most of the mathematical framework of CCS carries over almost unchanged. The operational semantics of CHOCS is given as a labelled transition system and it is a direct extension of the semantics of CCS with value passing. A set of algebraic laws satisfied by the calculus is presented. These are similar to the CCS laws only introducing obvious extra laws for sending and receiving processes. The power of process passing is underlined by a result showing that the recursion operator is unnecessary in the sense that recursion can be simulated by means of process passing and communication. The CHOCS language is also studied by means of a denotational semantics. A major result is the full abstractness of this semantics with respect to the operational semantics. The denotational semantics is used to provide an easy proof of the simulation of recursion. Introducing processes as first class objects yields a powerful metalanguage. It is shown that it is possible to simulate various reduction strategies of the untyped λCalculus in CHOCS. As pointed out by Milner, CCS has its limitations when one wants to describe unboundedly expanding systems, e.g. an unbounded number of procedure invocations in an imperative concurrent programming language P with recursive procedures. CHOCS may neatly describe both callbyvalue and callbyreference parameter mechanisms for P. We also consider callbyname and lazy parameter mechanisms for P. The second calculus is called Plain CHOCS. Essential to the new calculus is the treatment of restriction as a static binding operator on port names. This calculus is given an operational semantics using labelled transition systems which combines ideas from the applicative transition systems described by Abramsky and the transition systems used for CHOCS. This calculus enjoys algebraic properties which are similar to those of CHOCS only needing obvious extra laws for the static nature of the restriction operator. Processes as first class objects enable description of networks with changing interconnection structure and there is a close connection between the Plain CHOCS calculus and the πCalculus described by Milner, Parrow and Walker: the two calculi can simulate one another. Recently object oriented programming has grown into a major discipline in computational practice as well as in computer science. From a theoretical point of view object oriented programming presents a challenge to any metalanguage since most object oriented languages have no formal semantics. We show how Plain CHOCS may be used to give a semantics to a prototype object oriented language called 0.
