Higher-order dataflow and its implementation on stock hardware

An important shortcoming of current data ow implementations of functional languages, is the treatment of higherorder functions. The solution usually adopted is through the use of closures. However, such an approach besides being expensive in practice, does not seem to harmonize with the basic principles of tagged data ow. This paper presents a technique for implementing higher-order lazy functional languages using only simple data ow concepts. The proposed approach when implemented on stock hardware, competes with the modern reduction-based implementations of functional languages.