I'm just wondering. The normal keys in UniPCemu's non-numeric pad keys are currently affected by numlock, giving fake shifts if required according to documentation. Is that correct behaviour? Or is it actually supposed to act on the numpad keys versions of them instead? The documentation I can find seems to imply that the num lock state has effect on the non-numeric keypad keys instead?

Anyone knows exactly how those cursor pad keys work? Are they prefixed(during presses) and suffixed(during release of said key) by those fake shifts? Or are the numeric pad keys(those under the numlock key) supposed to adjust?

http://www.quadibloc.com/comp/scan.htm mentions something about some 'cursor pad' keys? Are those the ones supposed to have adjustments? Are those the normal insert/delete/home/end/page up/page down and arrow keys?

Edit: OK. I notice that the Ctrl-Alt-Del actually gets sent to the emulated machine (Although SDL2 at least seems to have issues with properly reporting the num locked keys when num lock is off(it's reporting them as the insert/home/... keys in that case)). But somehow, the BIOS/OS won't get the message until the del key is pressed/released for a second time?

Eventually found the bug that was causing Ctrl-Alt-Del to not work properly: it was checking for modifier keys when searching for the latest pressed key(to be properly typematic). When searching through them, when multiple keys are pressed at one time(in this case alt and del), it would skip the modifier key(alt or ctrl, whichever is added last by the main routine) and send the keystroke of del(and performing alt when enough keys had been released to be counted as a first key in a sequence(in the case of modifier keys, they would only start to be sent when no other modifier keys were actually pressed).

So that's fixing the Ctrl-Alt-Del bug in UniPCemu! 😁