WebSocket Progress Streaming

Long-running Studio operations (characterisation, heatmap, model scan) stream progress updates via WebSocket. A progress bar appears in the UI showing the current step, percentage, and description.

Supported Operations

Operation	Steps	Typical Duration	What It Does
characterize	~50	5-30s	f-I curve (20 pts) + sensitivity (15 params × 2)
heatmap	N×M	10-60s	Two-parameter sweep (e.g. 15×15 = 225 sims)
scan	118	30-120s	Classify all 118 models by firing pattern

Protocol

Connect via WebSocket to /ws/progress, send a JSON request, receive progress frames:

Client → Server: {"op": "characterize", "config": {"name": "AdExNeuron", "current": 15.0}}

Server → Client: {"type": "progress", "step": "trace", "pct": 0, "msg": "Running default simulation"}
Server → Client: {"type": "progress", "step": "fi_curve", "pct": 5, "msg": "f-I curve 1/20"}
Server → Client: {"type": "progress", "step": "fi_curve", "pct": 10, "msg": "f-I curve 2/20"}
...
Server → Client: {"type": "progress", "step": "sensitivity", "pct": 60, "msg": "Sensitivity 3/15: tau_m"}
...
Server → Client: {"type": "complete", "pct": 100, "result": {...}}

Message Types

Type	Fields	Meaning
progress	pct, step, msg	Operation in progress
complete	pct=100, result	Operation finished successfully
error	msg	Operation failed
heartbeat	—	Connection alive, worker still running

Frontend Integration

The progress bar appears below the header when isSimulating is true and progressMsg is non-empty. It shows:

• Current step description (e.g. "f-I curve 12/20")
• Percentage bar (0-100%)
• Percentage number

When WebSocket is unavailable (e.g. behind a proxy that strips WS), the frontend falls back to the standard HTTP endpoint without progress.

API

WebSocket: /ws/progress

Request:

{
  "op": "characterize",
  "config": {
    "name": "AdExNeuron",
    "params": {"a": 4.0, "b": 0.0805},
    "current": 15.0,
    "duration": 200.0,
    "dt": 0.1
  }
}

Operations: - characterize — full model characterisation with f-I + sensitivity - heatmap — 2D parameter sweep (requires param_x, param_y, x_min/x_max/x_steps, y_min/y_max/y_steps in config) - scan — scan all 118 models (no config needed)

Implementation

The backend runs the long operation in a daemon thread, pushing progress messages to a queue.Queue. The async WebSocket handler polls the queue and forwards messages to the client. This avoids blocking the event loop while the CPU-intensive simulation runs.

Client ←WebSocket→ AsyncIO handler ←Queue→ Daemon thread (simulation)