Actions Library

The server core bundles a set of action classes under spx_core/actions. They run inside the simulation loop to evolve attribute values over time. Each action inherits from the same base you use in the SDK, but the server version is optimized and instrumented for production logging.

How actions are structured

• ramps.py: ramp, step, and integrator functions for smooth transitions.

• saws.py: sawtooth generators, useful for repetitive test signals.

• interpolate.py: linear/multi-point interpolation between samples.

• noises.py: additive and multiplicative noise injection.

• pid.py: PID controller implementation that reads error attributes and writes control outputs.

• overrides.py: temporary overrides (force, restore).

• suspend.py: pause/resume other actions.

• call.py: call arbitrary registered functions with resolved parameters.

Each action consumes a YAML block in the actions: list. Parameters are evaluated just like in the SDK (#attr, #external, expressions).

Ramp example

YAML

actions:
  - ramp:
      output: "#attr(voltage)"
      start_value: 0.0
      stop_value: 230.0
      duration: 5.0

JSON

PID example

YAML

actions:
  - pid:
      output: "#attr(heater_power)"
      setpoint: "#attr(target_temperature)"
      feedback: "#attr(current_temperature)"
      kp: 2.0
      ki: 0.5
      kd: 0.1
      sample_time: 0.1

JSON

Noise injection

YAML

actions:
  - noise:
      output: "#attr(sensor.reading)"
      std: 0.01
      mode: proportional

JSON

Overrides and suspend

YAML

actions:
  - override:
      target: "#attr(fan_speed)"
      value: 1000
      duration: 2.0
  - suspend:
      target_actions: ["ramp", "pid"]
      duration: 1.0

JSON

Best practices

• Place critical control loops (PID, overrides) at the top of the list so they execute first.

• Use suspend to temporarily disable other actions during fault injection.

• Combine ramps with noise to emulate realistic sensor behavior.

• Keep durations explicit; otherwise actions may run indefinitely.