Module tuples 

The Tuple value descriptor (§4.5 structural tuples).

A tuple is an anonymous, positional product: a fixed number of elements, each a GcRef, in source order. Unlike records (§4.5) tuples carry no field names — identity is the element-type sequence alone (so (Int, Int) and (Int, Bool) are distinct, and (Int, Int) is one shape regardless of where it appears).

Each distinct tuple shape (element descriptor sequence) gets a TupleSchema. The schema is leaked to &'static (one per shape) because a tuple’s descriptor callbacks need a type-stable home for the element descriptors; this mirrors how the codegen leaks RecordSchema and function names.

The descriptor dispatches element-wise through the schema (§11.4) — there are no scattered type switches. A single TUPLE-shaped descriptor serves every tuple because the per-shape knowledge lives in the schema referenced from the payload. Structural equality and hashing (§5.5) recurse element-wise; a tuple is equatable/hashable iff every element is. The ordering a container imposes (ADR-138) recurses the same way — a Map[(Int, Int), V] walks its keys element-wise — which is a different question from the source-level <, and (1, 2) < (1, 3) is still refused at check time.

Structs

TuplePayload

The Tuple payload: a pointer to the static schema plus the element values (one GcRef per element, in schema order).

TupleSchema

The static shape of a tuple: an ordered list of element descriptors (positional, no names). Leaked to &'static once per distinct shape by the codegen.

Statics

TUPLE

Descriptor for the Tuple value type (§4.5). Structural equality, hashing (§5.5) and the container ordering (ADR-138) all recurse element-wise through the per-shape schema’s element descriptors. A tuple is equatable/hashable iff every element is; functions never are, so a tuple containing a function is neither.

Functions

point_schema

The cached 'static schema for a (Int, Int) point tuple. Used by Grid methods that return (x, y) points (§6.4). Built once and leaked; the two element descriptors are both INT. This avoids the codegen round-trip for tuple schemas when the runtime allocates points directly.