A sketch-based technique for the derivation of self-interpreters

Abstract

We investigate a view of compiler generation which does not involve the direct specification of the source→target relationship. Here, the concentration is on the role of self-interpreters in this context, and their derivation. The method is based on a category theoretic model of language using finite limit sketches and requires the automatic derivation of a target partial evaluator and a source interpreter, expressed as a target program. We describe a technique to derive a self-interpreter as this represents a significant step towards the derivation of both partial evaluators and interpreters.