On Tue Dec 5 11:01:38 2023 +0000, Alexandros Frantzis wrote:
This is to keep the `NtUserLogicalToPerMonitorDPIPhysicalPoint` call happy, since it will fail if the point is outside the window rect. I have tried to stick with the current win32u exports, to avoid leaking some of the current internal assumptions (e.g., `get_win_monitor_dpi(hwnd) == system_dpi`). Perhaps at some point it may be worth exposing some general facilities from win32u for drivers to perform such dpi mappings with more versatility. That's also the reason I have left a 'FIXME' for thread->win dpi mapping in the WAYLAND_ClipCursor code, since getting the current thread dpi for calculations will require a duplication of the related internal win32u code.
Okay, I don't have a good knowledge of this DPI stuff, I don't have a good idea what to do yet. Probably fine to keep it like that but maybe with a comment there too.