[Bug] Bluetooth: the very first key presses while the keyboard is in deep sleep are lost

[bam80]

Description

Hi, let me quote the request for ZMK here, it apparently applies to QMK also:

Hi,
I'm new to ZMK, but a long time BT keyboards user.

One behavior driving me nuts is the loosing of the first key-presses if the keyboard is in deep sleep (and thus disconnected itself from BT host).
As it's turned out, one half of the problem is a regression in Linux BT stack (Bluez), when the key events gotten by host are just discarded until the keyboard fully re-connects (from host POV). The optional fix (workaround?) for Bluez was recently introduced:
bluez/bluez#737

That still wasn't the end of the story, unfortunately:
apparently, some BT keyboards (even modern ones, with BT 5.0) do not pass the key events to host at all, if they were in deep sleep!:
bluez/bluez#737 (comment)
That was quite unexpected to me, because even ancient BT 3.0 "Microsoft" keyboard didn't have the issue, as well as the other cheap one.

I really upset about this situation. The proper behavior in this case is essential to me.
My only hope is ZMK (and others free keyboard firmwares) do that properly.

Any insight how things are really going?
Could we implement the feature in ZMK, if it doesn't do that yet?
Thanks.

[tzarc]

The closest anyone has been to this area recently has been @KarlK90, so I’ll defer to their familiarity in this area (if they have any comments at all).

Ultimately this is low priority for the QMK team who have a whole host of other things going on as well — your best bet is to try to rectify it yourself.

[bam80]

Thanks for reply @tzarc, I refactored the feature request to a bug, as apparently there is really no reason for a keyboard to hold these keypresses.