The synergy of functional programming and very large scale integration

Abstract

In 1977 functional programming received a fresh and timely examination from Turing Award recipient John Backus. This work raised awareness of inherent deficiencies in traditional von Neumann architecture and languages. In his subsequent writings Backus uses results from combinatory logic to further develop a high-level functional programming form (FP) in which variable-free programs are built from a set of primitive programs using a few combining forms and recursive definitions. The symmetric properties of the underlying program forming operations (PFO's) in FP provide a strong mathematical structure with which to study the "algebra of programs." These PFO's generate programs which are semantically simple, are clearly hierarchical, and facilitate program correctness proofs. Unfortunately, FP style code does not execute as quickly on traditional von Neumann machines as do traditional languages. Very large scale integration (VLSI) chip design technology may address this problem. Development of VLSI offers low cost architectural possibilities for concurrent processing to which FP is well suited. In return, FP offers the clarity, structure, and mathematically sound software which VLSI applications require to justify their design, development, and implementation.