Fix
- #59 follow-up — BLE adapter probe before enrollment. Reporter @craigo1975 saw 10-30s lag on every H6199 command. Root cause: HA running in a VM without Bluetooth passthrough still receives BLE advertisements via the passive scanner stack, but
BleakClient.connecttakes ~25-40s of retries per command before REST fallback fires. The integration now probesasync_scanner_count(connectable=True)before adding a device to the BLE dispatch path. Zero connectable adapters → stay cloud-only.
Notes
- Trade-off: adapters attached after the integration is set up won't re-enable BLE until the integration is reloaded.
- Affected users (HA in VM / Docker without BLE passthrough) should see commands respond at REST speed (<1s) again after upgrading.