Abstract

. Polymorphic equality is a controversial language construct. While being convenient for the programmer, it has been argued that polymorphic equality (1) invites to violation of software engineering principles, (2) lays a serious burden on the language implementor, and (3) enforces a runtime overhead due to the necessity of tagging values at runtime. We show that neither (2) nor (3) are inherent to polymorphic equality by showing that one can compile programs with polymorphic equality into programs without polymorphic equality in such a way that there is no need for tagging or for runtime type analysis. Also, the translation is the identity on programs that do not use polymorphic equality. Experimental results indicate that even for programs that use polymorphic equality, the translation gives good results. 1 Introduction Often, statically typed languages, like ML, provide the programmer with a generic function for checking structural equality of two values of the same type. To avoid ...

Citations