Extracting polyvariant binding time analysis from polyvariant specializer

Abstract

Polyvariant binding time analysis allows a program to be transformed for improving propagation and usage of static information. It could be a useful instrument for better specializability. We show that the process of such a transformation is of the same nature as the whole specialization process. Moreover, we present a practical method for realizing polyvariant binding time analysis based on the double application of a polyvariant specializer. The proposed technique is restricted to first order programs with strict semantics.