bynk-project

Bynk's project model: discovery, the unit dependency graph, path resolution, cross-unit consistency checks, and the schema-registry document's read/write halves.

It holds:

• discovery — walks a project's roots and finds every .bynk file (and, in project form, every unit's own manifest).
• graph — the unit dependency graph and its cycle/consistency checks.
• paths / roots — path resolution and the include/exclude root rules a project's layout is interpreted through.
• consistency — cross-unit checks (directory-kind, directory-name, group- kind, path-name alignment).
• schema_registry — reads and writes bynk.schema.lock, the schema registry document.
• parse_cache / json — supporting caches and JSON helpers the above use.

What deliberately stays out of this crate: schema-registry reconciliation (bynk-check-coupled, stays in bynk-emit), diagnostic modes and project-analysis result types (facts about how the pipeline is driven, not about the project itself), and test codegen — all downstream concerns, not project modelling.

Where it sits

This crate depends on bynk-syntax only (plus toml/serde for the manifest and lock-file formats) — it has no dependency on the checker or the emitter. bynk-check, bynk-emit, and bynk-ide each depend on it directly for discovery and path resolution, rather than re-deriving their own.

Use

[dependencies]
bynk-project = "0.290"

use bynk_project::roots::Roots;

let roots = Roots::Single(std::path::PathBuf::from("src"));

See the API docs for the full surface.

License

Licensed under either of Apache-2.0 or MIT at your option.