Paradigm composition

While a multiparadigm language is often attractive in concept, to make it practical there must be a consistent method for combining different paradigms in an application. Depending on the paradigms in question, this may have a large effect on the feasibility of this mixing. In general, the main difficulties are: (1) combining the execution models, and (2) combining the data structures, introduced in various constructs.

In Prop , this challenge of mixing paradigms is met in a number of ways:

• First, algebraic datatypes (and extensions thereof) are used as the uniform data structure in all the pattern matching constructs.
• Secondly, different pattern matching constructs are encapsulated in an object oriented manner using the class structure of C++ . Using virtual functions (i.e. late binding) and multiple inheritance, different pattern matching constructs, such as rewriting and inference, can be seamlessly combined.