On Thu Jan 23 12:38:42 2025 +0000, William Horvath wrote:
> ~~I just tested this again on swaywm (recent git version), and it only
> fixes the issue the first time the minimize-to-tray action is performed.
> Once the floating tray icon is clicked and the window is restored, on
> subsequent attempts, the game window is no longer unmapped/hidden. It
> remains as an empty black window on screen (but the tray icon does show
> up again).~~
> ~~This is different from how this patch affects i3 (where I tested
> before), because on i3 it fixes the issue no matter how many times the
> minimization/restoration is done. Before the async request patches, the
> functionality worked identically on i3 and sway.~~
> ~~Sorry I didn't catch that earlier. It's still interesting, because it
> means that the outdated workaround/hack I pointed out in i3's source is
> not the only thing to blame here. I looked around for something like
> that in sway/wlroots' xwayland interop and couldn't find anything as
> suspicious like that.~~
> Ignore this. The difference where the window is a black rectangle on
> consecutive attempts is not a regression due to the async requests
> commit, it's a separate bug. This patch does restore the behavior to how
> it was before the async requests change on both i3 and sway.
Thanks for checking this, reading the edited comment I understand this actually works and everything is fine.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/7121#note_92585
I just tested this again on swaywm (recent git version), and it only fixes the issue the first time the minimize-to-tray action is performed. Once the floating tray icon is clicked and the window is restored, on subsequent attempts, the game window is no longer unmapped/hidden. It remains as an empty black window on screen (but the tray icon does show up again).
This is different from how this patch affects i3 (where I tested before), because on i3 it fixes the issue no matter how many times the minimization/restoration is done. Before the async request patches, the functionality worked identically on i3 and sway.
Sorry I didn't catch that earlier. It's still interesting, because it means that the outdated workaround/hack I pointed out in i3's source is not the only thing to blame here. I looked around for something like that in sway/wlroots' xwayland interop and couldn't find anything as suspicious like that.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/7121#note_92582