Jeroen Wijenbergh

> > hmmm so is this only a 'fix' for when we expect it to close? Because when you shutdown your pc I imagine that the x server can get...

Maybe it fixes it for some scenarios? I don't know. The changes you made do make sense to me though It might indeed help for graceful closes.

What about this? This works for me, it simply just checks if its an XCB connection error, and if so it logs on debug. We have to disable logging after...

> What about this? This works for me, it simply just checks if its an XCB connection error, and if so it logs on debug. We have to disable logging...

> Thanks for looking at this again. > > I thought our main concern with my original PR was that we wanted to find the root cause of the messages....

Simulated key presses are now proper X events.  They still go through the qtile keybindings, so if an user wants to ignore the swallowing when simulating keypresses, they can...

> Unfortunately this isn't working for me as expected. With this config to call rofi: > > ```python > keys.extend([ > Key(["shift"], "f", lazy.spawn("rofi -show"), swallow=True), > Key(["control"], "f", lazy.spawn("rofi...

> Oh, I'm being an idiot, aren't I? We decided that `swallow=False` is unsupported when the binding triggers a keyboard grab, didn't we? Please could you add a note about...

> Using the same keybinding as you, `Key([], "b", lazy.spawn("rofi -show"), swallow=False),`, these are the events printed by xev: > > ``` > FocusOut event, serial 34, synthetic NO, window...

> > Maybe XTest needs to be used instead? > > I think this is the likely solution. Clients would need to explicitly support send-event, adding a handler for those...