On Tue May 5 01:41:16 2026 +0000, navi wrote:
initially i've added simple tests but those didn't hold under certain combinations of dpi/scale and resolutions ~~after being confused for a while, i noticed the ratio between the dpi aware virtual screen pixel dimensions and the pointer device dimensions was always constant everywhere i tried (qemu w/ various output configurations, and a physical laptop)~~ ~~i still can't really find out where from, why, or how that ratio is derived, so i didn't yet update the implementation itself -- would appreciate some suggestions if any~~ nevermind, ci fails and i think qemu and the laptop just had very similar display sizes okay issue was, the resulting value was influenced by the dpi value, but it was always the dpi that was set at boot, which would be applied to the current resolution -- changing the dpi reflected on the displayRect, but would only reflect on the pointerDeviceRect after a reboot
i am much unsure how to test this, it's a very specific quirk that only shows up when scaling is changed at runtime, so should we try to match that behaviour at all? -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10649#note_138961