Language first philosophy [parallel computers]

Parallel computers are an obvious way of providing increased computing power and utilizing the benefits of VLSI technology in a cost-effective manner. However, it is difficult to argue that parallel computing has been an unqualified success since the first serious attempts at real parallel structures some 20 years ago. The author argues that this can be attributed, to a large extent, to the concentration on the hardware and operational models of parallel computing rather than the real problems which are concerned with how one can write programs to make use of parallel machines for other than a small class of problems whose structure is particularly suited. It is suggested that the properties of very high level declarative languages are a good starting point, but questions whether the essential operational nature of many computing problems can be accommodated by any current approaches without compromising the declarative principles