Performance

This SDK keeps the implementation simple.

Characteristics

• One in-memory token cache per client instance
• No background polling
• No built-in concurrency limiter
• No request batching

Practical advice

• Reuse a single OpenLogistics instance (or one client) when possible
• Avoid constructing a new client for every request
• Cache serviceability and rate responses instead of repeating identical calls
• Handle ShiprocketAPIError so a failed refresh does not leave a request silent

The SDK does not add queues, background workers, or connection pooling. Keep concurrency control in the calling application.