Functional programmers are taking a closer look at Zig, a language that aligns with their design criteria. The author evaluates languages using three axes: expressing domain ideas without implementation noise (e.g., avoiding manual memory management overhead), leveraging type-system facilities for correct-by-construction systems, and minimizing surprise code that diverges from implementation understanding. Zig stands out for its comptime system, which the author suspects offers a simpler, more flexible alternative to the type-system programming seen in Haskell—after over a decade of Haskell experience, programming the type system is now a hard requirement for serious language consideration.

The author, eager to avoid functional systems languages, critiques the industry’s misalignment with imperative programming needs, particularly around monads. They argue monads are not abstract math but a computational context description, enabling languages to lack built-in notions of time (among other things). Without this, languages can mold to user needs and see improved compiler optimization ceilings. Zig connects to systems programming because its performance-oriented approach dissatisfies common functional languages like Haskell, OCaml, and Lisp/Clojure, which rely on garbage collection (GC). The author views GC as a tradeoff that creates a performance ceiling due to a forest of pointers and risks developer disconnect from machine/runtime systems. Since 1995, CPU compute has grown roughly 10,000x faster than memory access timing, making languages designed for earlier machines suboptimal for today’s hardware. Zig innovates by discouraging the forest of pointers approach and encouraging better manual memory management through Arenas and Allocators, allowing users to control memory more effectively and exploit modern machines better than predecessors. Zig 0.16 reworked its IO system to an interface design, with the author noting the Zig team’s independent creation of a Reader monad-like IO system speaks to the universal nature of monads and confirms they’re on the right track.