Staged Parser Combinators in Scala: Have Your Cake and Eat It (Too)

Mon, 02 Mar 2026 00:00:00 +0000

Parser combinators have a reputation of poor performance—worst-case exponential time—and deemed good for prototyping but not for production. Or so the story goes. While techniques like Packrat parsing or static analysis offer linear-time guarantees by addressing naive backtracking, they come with their own trade-offs.

In this post, I want to explore an optimization route that complements the others: avoiding the performance penalty of abstraction.

By leaning into Scala 3’s metaprogramming capabilities, our combinators combine code fragments. By moving to a staged continuation-passing style (CPS) encoding, we make the control flow explicit and continuations are fully evaluated at compile time. The result is a parser that runs over 10x faster than a standard non-staged version. Interestingly, this approach is deeply rooted in program optimization history, closely mirroring the First Futamura Projection.

Partial Evaluation on Scattered Space

Staged Parser Combinators in Scala: Have Your Cake and Eat It (Too)