SDKs

Chirpier exposes the same product model across JavaScript, Python, and Go.

All three SDKs share these important behaviors:

• same bearer token for logs and control-plane APIs
• unknown events auto-create event definitions
• helper methods for events, analytics, policies, alerts, deliveries, and destination tests

Choose your SDK

If you only need raw HTTP calls, use API Reference.

Production behavior

All three SDKs batch logs locally and flush them in the background. Understanding these defaults helps with production deployments.

Batching and flushing

Logs are queued in memory and flushed automatically:

• Batch size: 500 logs (configurable). The SDK flushes when the queue reaches this size.
• Flush interval: 500 ms (configurable). If the batch size is not reached, logs flush on this timer.
• flush(): Sends all queued logs immediately. Blocks until the batch is delivered.
• shutdown() / close(): Flushes all remaining logs and releases resources. Always call this before your process exits.

Retry and backoff

Failed batches are retried with exponential backoff and jitter:

• Retryable errors: Network failures, 429 (rate limit), and most 5xx responses.
• Non-retryable errors: 401, 403, 404 are not retried.
• Backoff: Delay doubles per attempt, with random jitter added. 429 responses respect the Retry-After header.
• Default attempts: 10-15 depending on SDK.
• Logs are not dropped: Failed batches are re-queued for retry, not discarded.

Concurrency

• JavaScript: Single async worker with lock-based queue coordination.
• Python: Single background thread processing batches sequentially.
• Go: Configurable worker pool (default 3) for concurrent batch delivery. All methods accept context.Context for cancellation.