Dynamic imperative languages for runtime extensible semantics and polymorphic meta-programming

Abstract

Dynamically typed languages imply runtime resolution for type matching, setting-up an effectible ground for type-polymorphic functions. In statically typed object-oriented languages, operator overloading signifies the capability to statically extend the language semantics in the target program context. We show how the same can be accomplished dynamically in the Delta dynamic language, through simple member-function naming contracts. Additionally, we provide a software-pattern for dynamically extensible function semantics, something that cannot be accommodated with static function overloading. We demonstrate how meta-programming, i.e. Grafting of parametric program capsules solving generic problems known as meta-algorithms or meta-components, become truly polymorphic, i.e. can accept an open set of parameter values, as far as those dynamically bind to eligible elements compliant to the meta-program design contract. In Delta, inheritance is dynamically supported as a runtime function, without any compile-time semantics, while all member function calls are resolved through late binding. We employ those features to show how Delta supports the imperative programming of polymorphic higher-order functions, such as generic function composers or the map function. © Springer-Verlag Berlin Heidelberg 2006.