Types and data definitions

Primitive types are i8, i16, i32, i64, u8, u16, u32, u64, f32, f64, bool, str, and bytes, the type of byte literals such as b"\x89PNG". There is no built-in int; use a width or a transparent alias. Numbers have local literal inference and explicit conversions rather than implicit widening.

The unit type () denotes completion without data. It can be an expression, function result, or enum or tuple component, including Option[()] and Result[(), E]. Standalone unit bindings, parameters, collection elements, and class fields are unsupported. A reference to unit, &() or &mut (), is an ordinary reference: borrowed unit payloads have this type, and a &() parameter lends a unit argument like any other temporary. There is no None type or value.

Tuples are structural products. Classes and enums are nominal: identical fields or payloads do not make separately declared types interchangeable. Import aliases preserve a declaration’s identity; transparent aliases create none. Generic instances with different arguments have different nominal identities.

Dimensions are compile-time types with no values. Nominal bases and normalized products, quotients, powers and exact roots can be arguments to generic parameters constrained by Dimension.

Type inference stays within local expressions and calls; every function interface declares its types. Collection, reference, callable, and generator types follow their respective facilities, ownership, function, and iteration contracts.

Data definitions may refer forward and recur through permitted storage boundaries. References cannot be stored in classes, enums, or tuples. Generators can occur inside standard Option and Result wrappers, but not class fields, user enum payloads, or tuple components.

Quantities combine numeric storage with a compile-time dimension and check that arithmetic combines those dimensions correctly.