Skip to main content

​
Modular Configuration

Krono separates configuration from execution. Configuration changes are reactive—updating a setting immediately triggers the necessary internal logic (e.g., reconnecting the socket or clearing buffers).

​
Event-Driven State

State management in Krono relies on a strict unidirectional data flow.
  1. Status Events: The connection state (connecting, connected, error) is tracked separately from data.
  2. Config Events: Changing depth or symbol triggers a soft reset of the pipeline without destroying the instance.

​
Performance First

Performance is the primary constraint for a websocket-based orderbook visualizer. Krono implements several improvements to ensure low-latency updates and minimal CPU usage:
  • Mutation-free Updates: Data passed to the UI is immutable to ensure predictable React rendering behavior.
  • History Storage: The history buffer uses an associative array to handle the sparsity of orderbook price levels.
  • Lazy Calculation: Derived metrics (like total liquidity or spread percentage) are calculated only when requested or when the specific data point changes.
The architecture is designed to handle the full throughput of Kraken’s WebSocket feed, which can burst to hundreds of messages per second during high volatility.

​
Distinct Asset Pairs

Managing available trading pairs is handled by the AssetPairs entity, distinct from the Orderbook.
  • Separation: You don’t need an active Orderbook connection to fetch available symbols.
  • Caching: Symbol metadata (tick size, display names) is cached to minimize API calls.
  • Validation: Ensures the Orderbook is only initialized with valid, tradeable pairs.