Module codec 

Canonical codec — DAG-CBOR, the deterministic encoding the content address is computed over.

Content-addressing REQUIRES a single canonical encoding so that identical data → identical bytes → the identical ContentAddress, reproducibly across implementations and toolchains. DAG-CBOR is that form (sorted map keys, shortest-form integers, no indefinite-length items). rkyv — used elsewhere as a local zero-copy cache — is explicitly NOT the address codec: its byte layout is version/feature/target-bound, so committing or sharing rkyv bytes is a cross-toolchain liability. The address is over this form.

Determinism here rests on the encoded values being order-stable by construction (the runtime’s serialized structures use BTreeMap/Vec, and the morphism/closure rows are sorted before encoding), so the canonical bytes do not depend on iteration order.

Citation: IPLD DAG-CBOR codec specification (https://ipld.io/specs/codecs/dag-cbor/); RFC 8949 (CBOR) §4.2 (deterministically encoded CBOR).

Enums

CodecError

Errors from the canonical (DAG-CBOR) codec.

Functions

address_of

The content address of value: hash its canonical DAG-CBOR encoding. This is the bridge from “a definition” (any serializable value) to its place in the Merkle-DAG — address_of(definition) is the node’s identity.

canonical_decode

Decode bytes (canonical DAG-CBOR) back into a value — the inverse of canonical_encode.

canonical_encode

Encode value to its canonical DAG-CBOR bytes — the form the content address is taken over. Equal (order-stable) values produce equal bytes.