Zakariyya Von Forslun

> Is there any way to keep this fix in Xwayland-specific codepaths, even if it's even hackier? I think it might be possible. This is my patch for the change:...

> I think this approach does have the potential to be better, though I'm still not a fan of the fact that native Wayland views would remain activated despite not...

I have found that nested menus are still broken with this approach. This is because, upon mapping a submenu, the previous menu loses focus and deactivates the parent Xwayland view,...

> I'm OK with adding some more Xwayland-specific code to `Seat` if necessary, I only ask that it be cleanly separated from the rest of the code and not be...

@icp1994 Thank you very much for the offer! Could you check the patch I linked above and see if it fixes your issue?

> Unfortunately, it's still happening. Here's a minimal [river.log](https://github.com/riverwm/river/files/10372055/river.log) Thank you for testing! It is much appreciated. However, I am unable to determine the issue from that log file. Could...

> Still can repro, hope these are helpful: [river_patch_v2.log](https://github.com/riverwm/river/files/10373549/river_patch_v2.log) [river_patch_v3.log](https://github.com/riverwm/river/files/10373553/river_patch_v3.log) > > I exited river as soon as I could get the drop-downs/right-clicks to misbehave so anything "bad" should be...

> It doesn't happen immediately when I start the browser. It starts to happen after some browsing/scrolling around - the waiting time varies. Then I have to restart the browser...

Note: I can also reproduce it in sway, so it is possible that it is an issue with Vivaldi and/or Xwayland as opposed to river.

> @zakvf I think I like the look of the 3rd patch linked which adds the `activated_xwayland_view` field to `Seat` the most. [...] Do you feel like that patch is...