Rémi Bernon <rbernon(a)codeweavers.com> writes:
On 3/31/21 7:52 PM, Alexandre Julliard wrote:
Rémi Bernon <rbernon(a)codeweavers.com> writes:
@@ -599,15 +599,15 @@ static void map_event_coords( HWND hwnd, Window window, struct x11drv_win_data * TRACE( "hwnd %p, window %lx, data %p, input %p\n", hwnd, window, data, input ); if (window == root_window) pt = root_to_virtual_screen( pt.x, pt.y ); - if (window == data->whole_window) + else if (window == data->whole_window) { pt.x += data->whole_rect.left - data->client_rect.left; pt.y += data->whole_rect.top - data->client_rect.top; - } - if (GetWindowLongW( hwnd, GWL_EXSTYLE ) & WS_EX_LAYOUTRTL) - pt.x = data->client_rect.right - data->client_rect.left - 1 - pt.x; - MapWindowPoints( hwnd, 0, &pt, 1 ); + if (GetWindowLongW( hwnd, GWL_EXSTYLE ) & WS_EX_LAYOUTRTL) + pt.x = data->client_rect.right - data->client_rect.left - 1 - pt.x; + MapWindowPoints( hwnd, 0, &pt, 1 ); + } This is wrong, you'd still need to map them for the client window.
Can this happen? I don't think we register input for the client windows.
As far as I can see the event's window is always the whole_window, for which we selected pointer motion events, and that received input.
It was happening at some point. I think we might as well keep the check in case we change things and get client window events again, but otherwise the if() should be removed. -- Alexandre Julliard julliard(a)winehq.org