Skip to main content
Version: V2-Next

Glossary

Model Forge uses a small, fixed vocabulary. Code and documentation use only these terms, defined below. The decision behind this vocabulary is ADR 060.

Core terms​

Model Artefakt (artifact)​

Umbrella term for any persistable CORE unit: Element, DataStructure, Mapping, DataSource, DataSink, Pipeline, DataSet. Every artifact has a stable CORE URN identity and is versioned.

Element​

The umbrella for a single modelled domain type — a Class, an Enumeration, etc. An Element is a JSON Schema 2020-12 document whose $id is a versioned CORE URN. It carries version, name, description, URN.

DataStructure​

A stored grouping of Elements. Importing one JSON Schema document (a root, optionally with $defs) automatically creates one DataStructure that references the resulting Elements by URN — the module / provenance grouping ("these Elements were authored and imported together"). The DataStructure is named after the imported document's title and is managed as its own registry artifact through the ModelForge facade. A DataStructure is a stored artifact, not a View — it carries no content of its own and its members are fetched individually by URN (see ADR 060).

Mapping​

A declarative field mapping from a source Element to a target Element, both referenced by URN. Per target field it describes an operation (copy, concat, const, format, …) over the source payload. A Mapping is a stored, versioned artifact with its own CORE URN.

DataSource​

The declaration of an inbound data source (e.g. an MQTT topic or a database query) whose payload binds to an Element by URN. A DataSource is a stored, versioned artifact with its own CORE URN.

DataSink​

The declaration of an outbound target (e.g. an HTTP endpoint) whose payload binds to an Element by URN. A DataSink is a stored, versioned artifact with its own CORE URN.

Pipeline​

A dataflow graph of nodes and edges that wires DataSources through Mappings into DataSinks. A Pipeline references the participating artifacts by URN and is itself a stored, versioned artifact with its own CORE URN.

DataSet​

An integration composition: a manifest that references Elements, Mappings, Pipelines, DataSources and DataSinks by URN (the *Refs arrays). Where a DataStructure groups Elements by provenance, a DataSet composes artifacts of all types into a deployable unit (see ADR 051).

Bundle​

A View that renders an artifact together with the (complete) graph of the artifacts it connects to, as one self-contained document. The bundled (embedded $defs) and inlined schema projections are both Bundles.

Format​

A View that serialises an Element in a chosen representation — JSON Schema or XSD. An Element's format is recorded per version (see ADR 058).

View​

Umbrella for a read-side projection of stored artifacts — currently Bundle and Format. Views are generated on demand and never stored (see ADR 055).

Relationships at a glance​

  • DataStructure and DataSet are both manifests of artifact URNs: DataStructure is Element-only and import-scoped; DataSet is all-types and composition-scoped.
  • A Bundle is a View over an artifact and its dependency graph; Format is a View choosing the serialisation. Both are kinds of View.
  • There is no RDF/Linked-Data (JSON-LD) projection.