Logic Programming: Context, Character and Development

Logic programming has been attracting increasing interest in recent years. Its first realisation in the form of PROLOG demonstrated concretely that Kowalski's view of computation as controlled deduction could be implemented with tolerable efficiency, even on existing computer architectures. Since that time logic programming research has intensified. The majority of computing professionals have remained unaware of the developments, however, and for some the announcement that PROLOG had been selected as the core language for the Japanese 'Fifth Generation' project came as a total surprise. This thesis aims to describe the context, character and development of logic programming. It explains why a radical departure from existing software practices needs to be seriously discussed; it identifies the characteristic features of logic programming, and the practical realisation of these features in current logic programming systems; and it outlines the programming methodology which is proposed for logic programming. The problems and limitations of existing logic programming systems are described and some proposals for development are discussed. The thesis is in three parts. Part One traces the development of programming since the early days of computing. It shows how the problems of software complexity which were addressed by the 'structured programming' school have not been overcome: the software crisis remains severe and seems to require fundamental changes in software practice for its solution. Part Two describes the foundations of logic programming in the procedural interpretation of Horn clauses. Fundamental to logic programming is shown to be the separation of the logic of an algorithm from its control. At present, however, both the logic and the control aspects of logic programming present problems; the first in terms of the extent of the language which is used, and the second in terms of the control strategy which should be applied in order to produce solutions. These problems are described and various proposals, including some which have been incorporated into implemented systems, are described. Part Three discusses the software development methodology which is proposed for logic programming. Some of the experience of practical applications is related. Logic programming is considered in the aspects of its potential for parallel execution and in its relationship to functional programming, and some possible criticisms of the problem-solving potential of logic are described. The conclusion is that although logic programming inevitably has some problems which are yet to be solved, it seems to offer answers to several issues which are at the heart of the software crisis. The potential contribution of logic programming towards the development of software should be substantial