Long-lived instances of BleakClient/BleakScanner vs new instances

[x11x]

I just wanted to check: the preferred pattern is to create a new instance every time you need to scan (BleakScanner), or connect to a device (BleakClient).

Especially with BleakClient this seems important due to inconsistent state in BleakClient after disconnect as mentioned in #713 (comment) and a few other places.

I'm not sure if this is also critical in BleakScanner, but most of the examples (such as this one #933 (comment)) use a new BleakScanner instance every time (the BleakScanner.find_by_* class methods create a short-lived BleakScanner instance internally).

So is it correct to say, its not really recommended to keep a BleakScanner instance and call start(), stop() and start() again, etc. as needed? Likewise with BleakClient it is not recommended to call connect() after a disconnect()?

Since bleak is a relatively young project, I wonder if it is a good idea for bleak to enforce the recommended pattern and just raise an exception when connect() is called after a previous disconnect() on a BleakClient, or start() after stop() on a BleakScanner? It might help with the OOP complexities of inconsistent state in long-lived instances mentioned in #713 and I think elsewhere.

Maybe this should have been an issue instead of a discussion post but I just wanted to check I have understood (happy to open an issue for this suggestion if anyone things it is worth it).

[dlech]

Starting and stopping a scanner multiple times should be fine.