Resource Models and Loading

RECAP exposes canonical, load-aware resource models. Callers choose relationship hydration without choosing a separate wire or model shape.

Model identity and loading

Full models contain stable identity and schema fields. load="none" leaves relationships unloaded; targeted includes and load="eager" hydrate them. The old shape="ref" spelling is deprecated and normalizes to shape="full". Specify load="none" explicitly for unloaded relationships.

plates = (
    namespace.query_maker()
    .resources(load="none")
    .filter(name="Sample Plate 001")
    .include(["template", "properties"])
    .all()
)

Use load="eager" when the complete relationship tree is required. Resource and process-run eager loading uses a bounded, depth-independent number of SQL statements, but can materialize more data than needed. Targeted descendant queries are preferable when only one level or template is relevant.

Property model access

Resources created through builders expose property access models directly. A resource loaded with include(["properties"]) may require build_property_model() before accessing nested values. Process-run queries using include_resources() hydrate assigned resources and their children for property access.

for resource in plates:
    resource.build_property_model()
    print(resource.properties.dimensions.rows.value)

Child resources

plate = namespace.query_maker().resources(load="eager").first()
well = plate.children["A01"]

for name, child in plate.children.items():
    print(name, child.id)

Child hierarchies can be defined in a template or extended at runtime through a resource builder. Child identity remains scoped by its parent while the root resource retains its own UUID. Repeated references to the same root or child are canonicalized within client lifetime.