Literal unit syntax

A unit suffix is written attached to the magnitude (5MiB, not 5 MiB) — the same lexical slot as a width suffix like 200u8. Any suffix that isn't a numeric width is taken as a unit name and resolved against the field's declared type at evaluation time. The magnitude defaults to i64 (integer) or f64 (with a decimal point); the resolved value's type is the alias's underlying type.

@document
type Config {
  buffer:  std.ByteSize     // base unit: byte
  radius:  std.Distance     // base unit: millimetre
  timeout: std.Duration     // base unit: nanosecond
  sizes:   list<std.ByteSize>
}

buffer  = 4MiB              // 4 * 1024 * 1024  = 4194304
radius  = 3km              // 3 * 1_000_000     = 3000000
timeout = 30s             // 30 * 1e9 (ns)     = 30000000000
sizes   = [256KiB, 1MiB]   // each element resolves

§ 1Declaring units

Units live on a numeric type alias as repeated @unit(name, factor) decorators — the same alias-decorator mechanism as @min / @max. factor is the number of base units in one of that unit, and is an ordinary expression.

@unit("B", 1)
@unit("KiB", 1024)
@unit("MiB", 1024 * 1024)
@unit("kB", 1000)            // SI: powers of 1000
@unit("MB", 1000 * 1000)
type ByteSize = i64

§ 2Rules