Functional Programing

Welcome to the rst functional programming column! Functional programmers and formal methoders have much to say to each other, and this column is intended to further such communication. Functional programming supports formal methods. A large number of proof systems and other formal methods packages have been implemented in functional languages. ML, the doyen of functional languages, takes its name from its rst application as a meta language for writing theorem provers. Functional languages may also be suitable for writing executable speciications, for rapidly generating prototypes from speciications, or as the target language of a system that transforms speciications into eeciently executable code. Conversely, formal methods support functional programming. Functional languages are well known for their amenity to mathematical reasoning, and formal methods can supply tools to support such reasoning. The application of mathematical laws to functional programs is not just a dream for the future { it underpins most of the techniques for generating eecient code from functional languages. Contributions for future columns are invited { my postal and e-mail addresses appear on this page. I intend to interpret functional programming in the broad sense, ranging from lazy languages with absolutely no side eeects, such as Haskell, to strict languages with disciplined use of side eeects, such as Standard ML. The rst column deals with Erlang, a language of the latter sort. One impediment to the spread of functional programming is that most of the eeorts have been academic in nature, and relatively little eeort has gone into matters like programming environments that can make or break the suitability of a system for use in industry. Mike Williams of Ericsson has led the eeort to develop Erlang, which is provides a convincing demonstration that functional languages can have industrial relevance. All of us interested in transfering ideas from academia to industry will be keeping an eye on Erlang to see how well it succeeds. Although it doesn't incorporate all the latest academic niceties { for instance, it lacks a type system { if successful it may pave the way for a generation of functional languages and other ivory tower innovations to stroll out of the classroom and into the software houses.