Skip to main content

​
Canonical records

The presschain-models repository provides Rust records that are useful at API, projection and integration boundaries. They translate lower-level protocol state into named fields while preserving the identifiers and relationships applications need.

​
CapsuleRecord

Current fields include:
The model serializes field names in camelCase for JSON. A projection can use this record to render a Capsule card or article verification page without forcing the client to decode every raw contract tuple.

​
EvidenceRecord

Current evidence projections include:
The nested tally contains support, reject and participation counts plus status and finalization state.

​
ContributorRecord

Current contributor projections include:
Keep share_bps as an integer. Convert to a display percentage only at the UI boundary.

​
Metadata envelope versus projection record

Canonical Capsule metadata is publication input. It is normalized and hashed before or during publication. CapsuleRecord is a read model that may include convenient title, summary, canonical URL and cover fields. Do not attempt to reproduce an on-chain metadata hash by serializing a transformed projection. Use the canonical metadata envelope and canonicalization rules.

​
JavaScript serialization

Large EVM integers and 256-bit identifiers require careful JSON handling. A practical boundary representation is:
Use exact hex strings for bytes32 identifiers and decimal strings for integer quantities that can exceed the safe JavaScript number range.

​
Add projection provenance

When an API must be auditable, attach chain coordinates beside the semantic record:
This keeps domain models clean while still allowing consumers to trace a projected fact back to network state.

​
Compatibility discipline

Treat changes to canonical records as versioned API changes when they affect serialized shape or meaning. Adding a convenience UI field and changing the semantics of status are not equivalent changes. A strong projection layer makes those distinctions explicit instead of depending on every client to infer them.