Semi-Expli it First-Class Polymorphism for ML

We propose a modest onservative extension to ML that allows semi-expli it rst- lass polymorphism while preserving the essential properties of type inferen e. In our proposal, the introdu tion of polymorphi types is fully expli it, that is, both introdu tion points and exa t polymorphi types are to be spe ied. However, the elimination of polymorphi types is semi-impli it: only elimination points are to be spe ied as polymorphi types themselves are inferred. This extension is parti ularly useful in Obje tive ML where polymorphism repla es subtyping. 1999 A ademi Pres