AbstractWe are interested in evolution scenarios for language-based functionality. We identify different dimensions along which such functionality can evolve, including the following: (i) coding style; (ii) coding details; (iii) data model; (iv) crosscutting concerns; and (v) patches. We focus at language interpreters as examples of language-based functionality, but similar scenarios exist for type checkers, static analyses, program transformations, and other sorts of language-based functionality. Our experiences are based on using rule-based programming (with Prolog) for the implementation of language-based functionality, while evolutionary transformations of the functionality are perceived as meta-programs.