David Lechner

This has now been tested on Windows, Linux and Mac. It is working as expected (when used with #1147 and #1148). The test case is running the `discover.py` example and...

> bleak==0.18.1 Is this still a problem with Bleak v0.19.5? If it is, we will need to see debug logs and Bluetooth packet captures to see what is going on.

We really need the Bluetooth packet logs to see what is going on behind the scenes. Usually, this failure is caused by devices not properly responding to the requests Windows...

Thanks for the logs. I don't see any obvious differences between the "good" and "broken" ones. Do you have better luck if you make the timeout longer? I'm not able...

It just looks like stopping scanning is happening at a different time. All of the same requests and responses are made of the device. The "broken" one disconnects earlier, I...

> since the difference between the good and bad tests Doh, I missed that part. Can you check `asyncio.get_event_loop_policy()` for both cases?

Thanks. I went back and set up an environment like you described and I am able to reproduce the problem. There seems to be some strange side effect that is...

I made some progress in understanding the issue (quite by accident) by printing the thread in the logs with `%(threadName)s` (note to self, add this to BLEAK_LOGGING) and adding a...

Some other reports of GetGattServicesAsync hanging forever... - https://stackoverflow.com/questions/63819254/c-ble-why-does-getgattservicesasync-hang-forever-is-there-a-work-around-such - https://stackoverflow.com/questions/55594814/bug-in-windows-bluetooth-le-getgattservicesasync-method - https://github.com/OpenBluetoothToolbox/SimpleBLE/issues/31

I think I have narrowed this down to the `pywin32` package. This is a sufficient `requirements.txt` to reproduce the problem: ``` bleak pywin32 ```