Data Model

The Primmel data model provides primitives for describing the structure of records, registries, and references in your domain. It plays the same role as a UML class diagram, but in a form tooling can validate against.

Primitives

enum — closed value set

enum RoastLevel {
  Light   { definition "Light roast" }
  Medium  { definition "Medium roast" }
  Dark    { definition "Dark roast" }
}

A closed set of named values, each with a definition. Use enums for any field whose value comes from a fixed list.

class#data — registerable record

class BeanLot#data {
  lot_id:           string   { definition "Supplier lot identifier"; modality SHALL }
  origin:           string   { definition "Country or region of origin"; modality SHALL }
  received_on:      datetime { definition "Date received at the roastery"; modality SHALL }
  certified_organic: boolean { definition "Whether the lot is certified organic"; modality MAY }
}

The #data suffix is the marker: only class#data declarations can appear in a data_registry. Each field has a name, a type, and a block with at least a definition and a modality (SHALL/SHOULD/MAY/ CAN/MUST).

Helper class — embedded complex type

class RoastProfile {
  name             { definition "Profile name (e.g. City+, Full City)" }
  target_level:    RoastLevel { definition "Target roast level" }
  development_time: string    { definition "Target development time in seconds" }
}

A class without the #data suffix is a helper class — a reusable complex type that lives as a field inside other data classes. It cannot be registered directly.

reference(Target#data) — typed foreign key

class RoastBatch#data {
  lot: reference(BeanLot#data) {
    definition "Bean lot used for this batch"
    modality SHALL
  }
}

A typed pointer to another record in another registry. At evaluation time, tooling follows these references to traverse the data graph.

data_registry — named store of records

data_registry BeanLotRegistry {
  title "Bean lots"
  data_class BeanLot#data
}

A named store of records. The data_class clause declares which class#data shape its records conform to. Registries are what processes read from (reference_data_registry) and write to (output).

Field types

Process clauses that touch data

process RoastBatch {
  reference_data_registry {     // this process reads from
    BeanLotRegistry
  }
  output {                       // this process writes to
    RoastBatchRegistry
  }
}

Where to see it in action

• Example 02: Data and registries — the dedicated walkthrough.
• Implementation package — data classes and registries at production scale.