Transputer Communications, Vol. 1(1), 3-15 (August 1993)

SUMMARY The study of computing science is split at an early stage between the branches dealing separately with hardware and software; and there is a corresponding split in later professional specialisation. This paper explores the essential unity and overlap of the two branches. The basic concepts are those of occam, taken as a simple example of a high-level programming language; its notations may be translated by the laws of programming to the machine code of a conventional machine. Almost identical transformations can produce the networks of gates and flip-flops which constitute a hardware design. These insights are being exploited in hybrid systems, implemented partly in hardware and partly in software. They are being extended to earlier phases in the software life cycle, where fully automatic transformations are not available. The results may be applicable in embedded real time systems, where considerations of safety make provable correctness a criterion. 1 Correctness of design The design of a complex engineering product like a real time process control system is decomposed into a progression of related phases. It starts with an investigation of the properties and behaviour of the process evolving within its environment, and an analysis of requirements for its optimal or satisfactory performance, or at least for its safety. From these is derived a specification of the electronic or program-controlled components of the system. The project then may pass through an appropriate series of design phases, culminating in a program expressed in a high level language. After translation into the machine code of the chosen computer, it is loaded into memory and executed at high speed by electronic circuitry. Additional application-specific hardware may be needed to embed the computer into the system which it controls. Each of these phases presents a conceptual gap, as wide and challenging as that between hardware and software. Reliability of the delivered system requires that all the gaps be closed. It is achieved not just by testing, but by the quality of thought and meticulous care exercised by analysts, designers, programmers and engineers in all phases of the design. The goal of our research is to discover and formalise methods which reduce the risks and simplify the routines of the design task, and give fuller scope for the exercise of human skill and invention in meeting product requirements at low cost and in good time. The goal of this paper is to convey an impression of …