13:52 <novakane> there is a thing a bit annoying with hide-cursor timeout, when you have a popup displayed when the cursor hovers over something, the popup is close when the cursor hide

14:10 <tleydxdy[m]> Conceptually those popups are meant to show when you are "using your mouse" right?

14:10 <tleydxdy[m]> maybe just the timeout is too short?

14:20 <novakane> yeah I probably should use a timeout a bit longer, but I don't know if it would make sense if the cusor, even hidden, is on a popup it would display it

14:20 <novakane> it is probably counter intuitive though

14:27 <ifreund> Yeah, I don't think a hidden cursor should e.g. cause button hover effects to trigger

14:30 <novakane> yeah it's a bit annoying if this is something that have a lot of text to read but I guess I need to keep my curosr moving above, making some small circle xD

14:40 <ifreund> I'd just use a much longer timeout

14:40 <ifreund> If you see a way to improve river's behavior here without negatively affecting other cases that would of course be welcome though

14:58 <tleydxdy[m]> would be nice if the mouse has a presence sensor XD

15:00 <tleydxdy[m]> idea, what if we made it so the cursor doesn't enter any new surface, but if it is being passed already we keep it that way

15:02 <tleydxdy[m]> probably also need to break the pass through if there is a new layout request since the mouse could end up on a different part of the surface

15:07 <tleydxdy[m]> so we have two attribute, hidden and passthrough, on timeout it only sets hidden, but on operations that could result in cursor landing elsewhere it sets no passthrough (if hidden)

15:08 <tleydxdy[m]> like switching tags, switching outputs, or starting a new layout request

15:08 <ifreund> > I don't think a hidden cursor should e.g. cause button hover effects to trigger

15:09 <tleydxdy[m]> yeah, but if it where hovering above something already and then it gets hidden

15:09 <tleydxdy[m]> This will keep it that way

15:10 <tleydxdy[m]> it's like a delayed no passthrough

15:10 <tleydxdy[m]> like it will not trigger anything new still

15:12 <ifreund> I don't want a button hover effect to continue if the cursor is hidden

15:13 <ifreund> Maybe I didn't say that clearly enough the first time

15:13 <tleydxdy[m]> yeah that's valid

15:13 <tleydxdy[m]> but it's the exact thing novakane was finding right?

15:14 <ifreund> no

15:14 <tleydxdy[m]> > there is a thing a bit annoying with hide-cursor timeout, when you have a popup displayed when the cursor hovers over something, the popup is close when the cursor hide

15:14 <tleydxdy[m]> this?

15:15 <ifreund> ok, yeah I guess that is exactly what I want to happen

15:16 <novakane> e.g on river github page, you hover one of the contributor on the contributors list on the right side, the popup with disapears with the cursor

15:16 <ifreund> I don't think there's anything that we should change here, just use a longer timeout or disable hiding entirely

15:16 <novakane> s/with/will/

15:16 <ifreund> or use hide when typing or whatever

15:16 <novakane> yeah anything else would probably have drawbacks

15:17 <novakane> I'll try 5000ms timeout then, gonna find the perfect spot :P

15:17 <tleydxdy[m]> do a bisect XD

15:18 <novakane> well yes but no :P

15:19 <tleydxdy[m]> has to be 4269ms, anything else has problems

15:21 <novakane> I hope you're right or I'll blame you :D

15:24 <ifreund> I guess I need to run pam in a child process for my waylock rewrite, if I block the wayland event loop for too long with a pam library call the server can send enough events to fill up the server side buffer and kill the client :/

15:24 <ifreund> that's pretty dumb

15:26 <ifreund> good news is that since ext-session-lock-v1 is a thing now that bug isn't a security issue!

15:27 <ifreund> I guess I now have a working ext-session-lock-v1 lockscreen in Zig, though it has 0 UX so far

15:39 <novakane> more secure without UX :P

15:40 <novakane> can't be compromised if nobody can unlock it

17:46 <mizzunet> dnkl the unstable experience of river once yambar was because I accidentally uninstalled libxshmfence.

17:46 <mizzunet> But I see that neither river nor yambar depends on this one.

17:46 <ifreund> that sounds weird

17:47 <mizzunet> Yeah

17:47 <mizzunet> This package with neovide-bin which I later uninstalled and this issue started occuring

17:48 <dnkl> weird, and I have absolutely no idea why libxshmfence matters...

17:49 <mizzunet> s/with/installed with

18:27 <dnkl> just noticed that drag operations (drag-to-select text in terminal, drag scrollbar in browser etc) "survives" the view being hidden, and then coming back. For example, via set-focused-tags. Is this intended?

18:29 <ifreund> dnkl: there's a TODO about that, I decided that it was too tricky to implement robustly until we have a scene graph

18:30 <ifreund> https://github.com/riverwm/river/blob/8f59075bc505910207bb803d3530c6daee36577c/river/Cursor.zig#L942

18:30 <dnkl> got it, thanks :)

18:30 <dnkl> wasn't really bothering me, but found it a bit.. weird

18:31 <ifreund> Yeah, it definitely is a bit weird but shouldn't cause any real issues I don't think