MetaAST is a meta-model (M² in MOF terminology) that sits one level above any concrete language's AST, providing a universal representation for code analysis tools. Instead of rewriting analysers for each language (quadratic growth: N×(N-1) converters), each language provides a single adapter that maps its native AST to MetaAST and back—a Galois connection of abstraction and reification. The Elixir library 'Metastatic' implements this: every node is a triple (type, metadata, children), deliberately mirroring Elixir's quoted expressions. A purity analyser or cyclomatic-complexity tool written once against MetaAST automatically works for Python, Ruby, Erlang, Haskell, and any other language with an adapter. The architecture has three layers: Core (universal constructs like binary ops and literals), Extended (common but not universal constructs like loops and lambdas), and Native (language-specific constructs wrapped without claiming universality). The result is linear rather than quadratic growth in tooling effort.
Sort: