Templates

Templates (spx_core/system/templates.py) let you package reusable model fragments and register them in the class registry. Other models can then reference the template by name via the modules or instances containers.

Configuration Example

YAML

templates:
  multimeter:
    attributes:
      voltage: { default: 0.0 }
      current: { default: 0.0 }
    communication:
      ascii:
        port: 0
        mappings:
          "MEAS:VOLT?": "#out(voltage)"

JSON

What happens under the hood

• Each entry is validated against the Model schema using the same validation engine as the SDK.

• The template is stored in class_registry as { "class": Model, "template": <definition> }.

• When a module references multimeter, the runtime clones the template definition and instantiates it as a normal model.

• If an attribute references a meta parameter ($param(name)) and the attribute type is missing/str, template normalization can promote that type from meta_parameters[name].type.

Meta parameter typing example

templates:
  regulator:
    meta_parameters:
      k__setpoint_c: { type: float, default: 60.0 }
    attributes:
      # Type is auto-promoted to float from meta_parameters
      k__setpoint_c: { default: "$param(k__setpoint_c)", type: str }

Best practices

• Keep templates focused: one device or subsystem per template.

• Use parameters inside templates sparingly; prefer leaving knobs for the consuming model to set via parameters.

• Version templates (e.g., multimeter_v2) when breaking changes occur so older models remain compatible.

• Store template YAML alongside unit tests in spx-examples/library/domains/... and sync them into the server repo during releases.