Model Loader

spx_core/system/model.py defines the Model class that the server instantiates for every YAML document. It is responsible for:

• Holding top-level containers (attributes, actions, communication, etc.).

• Spinning up dedicated Timer, Polling, and Scenarios children when those sections appear in YAML.

• Driving the lifecycle (prepare, start, run) for the entire simulation.

Configuration Example

YAML

model:
  name: plant
  timer:
    step: 0.1
  polling:
    interval: 0.05
  scenarios:
    voltage_drift:
      enabled: true
      duration: 5.0
      actions:
        - noise:
            output: "#attr(voltage)"
            std: 0.01
  attributes: { ... }
  actions: [ ... ]
  communication: { ... }

JSON

Key behaviors

• Automatic extraction: The loader pulls timer, polling, and scenarios from the model definition before instantiating other containers. This keeps the YAML tidy—no need to declare those sections separately.

• Unique identifier: Each model gets a uid (UUID string) you can query through the API for observability.

• Real-time flag: model.real_time controls whether the polling loop runs continuously (true) or waits for manual ticks (false). You can flip it through the API for deterministic testing.

Lifecycle expectations

1. prepare() is called on every child (attributes, actions, communication, timer, polling, scenarios).

2. start() launches timer/polling threads when real_time is true.

3. run() is driven by either the polling loop or manual API calls (POST /run).

Practical tips

• Always keep attributes, actions, and communication at the top level—nested models inherit the same pattern.

• Use the SDK to validate the structure before deploying; the server loader relies on the same class registry.

• To disable real-time execution in tests, call PATCH /models/<id> with { "real_time": false } and then trigger run() steps manually.