Relationships — Concept

A relationship is a typed edge from one entity instance to another. Causet manages the edge store; you declare the shape and optionally wire emit_events so projections can materialize queryable join tables.

Why relationships exist

Not every connection between entities belongs on a single entity’s snapshot. A user follows dozens of artists — storing following: [] on the user works, but so does a first-class edge with uniqueness enforced by the engine.

Relationships give you:

• Engine-level uniqueness — unique: true deduplicates (from_id, to_id) at storage time
• Optional automatic events — emit_events.created / removed without a separate op: emit
• Cross-stream writes in core — relationship_create is allowed in core (unlike mutating another entity’s fields)

Relationships vs entity state vs projections

Typical pattern: relationship + emit_events → projection listens to those events → query joins projection tables.

Cardinality mental model

• many_to_many — users follow artists; artists have many followers
• one_to_many — one owner, many owned records (each to has at most one from)

Next steps

• Defining Relationships — syntax reference
• Examples — concert-app patterns
• Listeners — synchronous cross-entity updates as an alternative