> ## Documentation Index
> Fetch the complete documentation index at: https://laminar.sh/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# SDK Lifecycle

<Tabs items={['TypeScript', 'Python']}>
  <Tab title="TypeScript">
    <Heading level={2} id="ts-laminar-flush">Laminar.flush()</Heading>

    Flush pending spans to the backend.

    ```typescript theme={null}
    // At end of script or Lambda handler
    await Laminar.flush();
    ```

    **Returns:** `Promise<void>`

    **When to use:**

    * End of CLI scripts
    * End of serverless function handlers
    * Before process exit

    **When NOT to use:**

    * In web server request handlers (degrades performance)

    ***

    <Heading level={2} id="ts-laminar-shutdown">Laminar.shutdown()</Heading>

    Flush spans and shut down tracing.

    ```typescript theme={null}
    await Laminar.shutdown();
    // Can call Laminar.initialize() again if needed
    ```

    **Returns:** `Promise<void>`

    **Note:** Force flushes, clears config/context, releases resources. Can re-initialize afterward.

    ***

    <Heading level={2} id="ts-laminar-get-http-url">Laminar.getHttpUrl()</Heading>

    Return the configured HTTP URL.

    **Returns:** `string`

    ***

    <Heading level={2} id="ts-laminar-get-project-api-key">Laminar.getProjectApiKey()</Heading>

    Return the configured project API key.

    **Returns:** `string`
  </Tab>

  <Tab title="Python">
    <Heading level={2} id="py-laminar-flush">Laminar.flush()</Heading>

    Flush pending spans to the backend.

    ```python theme={null}
    # At end of script
    Laminar.flush()
    ```

    **Returns:** `bool` — Success status (`False` if Laminar is not initialized)

    **When to use:**

    * End of CLI scripts
    * Graceful shutdown of web servers

    **When NOT to use:**

    * In web server request handlers (degrades performance)

    ***

    <Heading level={2} id="py-laminar-force-flush">Laminar.force\_flush()</Heading>

    Force flush spans, blocking until complete. Resets context.

    ```python theme={null}
    # In AWS Lambda handler
    def handler(event, context):
        Laminar.initialize()
        result = process(event)
        Laminar.force_flush()  # Blocks until spans exported
        return result
    ```

    **Returns:** `None`

    **Use for:** AWS Lambda and serverless where background threads may be killed.

    **Behavior:** Shuts down and reinitializes span processor internally. Clears isolated context—subsequent spans start new traces.

    ***

    <Heading level={2} id="py-laminar-shutdown">Laminar.shutdown()</Heading>

    Flush spans and shut down tracing.

    ```python theme={null}
    Laminar.shutdown()
    # Cannot re-initialize in same process
    ```

    **Returns:** `None`

    **Note:** Unlike JavaScript SDK, Python SDK cannot re-initialize after shutdown.

    ***

    <Heading level={2} id="py-edge-cases">Edge Cases</Heading>

    * **Input/output size:** Values >10MB replaced with `"Laminar: ... too large to record"`
    * **Uninitialized:** If `Laminar.initialize()` not called:
      * `@observe()` returns original function unchanged
      * Manual span methods return `NonRecordingSpan`
    * **Generators:** `@observe()` aggregates all yielded values for output recording
    * **Span lifecycle:** `start_span`/`start_active_span` must end in LIFO order. Violating can corrupt context hierarchy.
    * **Context isolation:** By default, `@observe()` uses isolated context. Set `preserve_global_context=True` to continue existing OTEL traces.
    * **force\_flush:** Clears isolated context. Subsequent spans start new traces.

    ***

    <Heading level={2} id="py-laminar-get-http-url">Laminar.get\_base\_http\_url()</Heading>

    Return the configured OTLP HTTP URL.

    **Returns:** `str | None`

    ***

    <Heading level={2} id="py-laminar-get-project-api-key">Laminar.get\_project\_api\_key()</Heading>

    Return the configured project API key.

    **Returns:** `str | None`
  </Tab>
</Tabs>