11. Capstone: schema a real config

Put it all together — vocabulary, constraints, document root, and instances for a deployment config.

Before you start: Tooling day-to-day

The capstone is the workflow you'll actually use: pick a config your project needs (deployments, pipelines, fixtures), declare its vocabulary as types, point a @document at them, and author real instances. A symbol_set closes an enum-like field — wcl check rejects any symbol not in the set — which is usually the first constraint worth adding.

Work schema-first: get one instance validating, then tighten (defaults for the common case, @min/@max where numbers have bounds, optionals where absence is legal). When the schema stops surprising you, the config is done.

§ 1Exercise: A deployment config from scratch

Model apps deployed to environments: a closed Env vocabulary, an app block with defaults, and two instances.

// A tiny deployment config — schema and data in one file.
symbol_set Env { dev  staging  prod }

@block("app")
type App {
  @inline(0)   name: identifier
  env:         Env
  @default(1)  replicas: u32
  @default([]) domains: list<utf8>
}

@document
type Deploy {
  @children("app") apps: list<App>
}

app web {
  env      = :prod
  replicas = 3u32
  domains  = ["example.com", "www.example.com"]
}

app worker { env = :staging }

§ 2Exercise: Make it yours

Extend the schema with one thing your real config needs — a nested block (@child), a numeric bound (@min), or an optional field — and keep the document validating.