High-level direction for the project. Priorities may shift; this is a guide, not a contract.
- Stability — Keep the core pipeline (parse → Zig → JIT/compile) and engine NIFs stable; fix regressions and improve error messages.
- Documentation — Clear “why Zixir,” use cases, config, deploy, and standard library so evaluators and operators have one place to look.
- CI — Automated build and test (e.g. GitHub Actions) on push/PR so changes are validated consistently.
- Developer experience — First-run experience (fewer noisy warnings), VS Code extension install path, and clear distinction between
mix zixir run(short path) and full compiler pipeline (MLIR/typecheck).
- List comprehensions — Complete codegen so list comps execute correctly (parser and type inference already exist).
- Pattern matching — Broaden codegen and tests for more pattern shapes and edge cases.
- Observability — Document and, where useful, provide a simple integration path (e.g. Prometheus/ Grafana) for production.
- Examples and content — More runnable examples (pipelines, retries, cache, Python call) and a consolidated use-cases doc.
- Optional MLIR (Beaver) — Keep optional; improve “quick start with Beaver” on Unix for users who want full MLIR.
- Python NIF — Document when NIF vs port is used; improve build/deploy story for the NIF path.
- GPU — Keep toolchain-dependent; document one “quick start” (e.g. CUDA) for users who need it.
- Ecosystem — Hex package, possible VS Code extension publish, CONTRIBUTING and community guidelines.
Last updated: 2026. For implementation status and gaps, see PROJECT_ANALYSIS.md.