02:11 <ezzieyguywuf> it seems that when my pointer first enters my surface, a middle-click event is registers even if I'm not clicking anything. is this expected behaviour, or could it be oddness in my wayland server and/or hardware?

02:12 <danieldg> no, that's not expected

02:15 <danieldg> you can try libinput debug-events to see if it's the hardware

02:15 <danieldg> or verify with wev or just WAYLAND_DEBUG=1 what is actually being sent

02:16 <danieldg> check if you have any other mice that might have their middle button held down too

02:32 <ezzieyguywuf> danieldg: I will try these things thank you

02:57 <ezzieyguywuf> the WAYLAND_DEBUG stream is very verbose, I've tried filtering it for just wl_pointer events. the 'MIDDLE_CLICK' text is from my application, which reports changes in state e.g. if a button goes from released to pressed. my application assumes all buttons begin in the released state.

02:58 <ezzieyguywuf> oh, here's the output (filtered): https://dpaste.com/8ZDZYKXTB

02:59 <ezzieyguywuf> for comparison here is similarly filtered output, but showing what I consider a valid left-click event https://dpaste.com/2W8MUZFFK

03:00 <danieldg> ... that's not a middle click

03:01 <danieldg> you can see the button event that all clicks generate

03:02 <ezzieyguywuf> 🤔

03:03 <danieldg> look at the 'button' lines

03:03 <ezzieyguywuf> yea I see

03:11 <ezzieyguywuf> ah, thanks again for the help, turns out there's some silliness going in in how I initialize my state container

06:34 <wlb> weston Merge request !1327 closed (kiosk-shell: bind key to weston surface)

20:13 <colinmarc> is there a good way to approach frame scheduling if all my clients are using syncobj (and I'm doing the wait on the gpu)?

20:14 <colinmarc> or is the approach I've seen online of delaying the frame callback superseded by one of the new presentation protocols?

20:15 <colinmarc> there are ways to get timestamps back from the gpu... I just don't really want to do that

22:24 <DemiMarie> Colin Marc: Generally you want to wait on the CPU, so that a buggy client doesn’t lock up your compositor.

22:25 <DemiMarie> There is a way to do that but I don't remember the details.