With the fix for 4654713, screen insets are now correctly updated when the Windows taskbar is moved onto a secondary monitor.  However there is still one edge case where screen insets are reported incorrectly.  If the taskbar is on a secondary monitor which is then removed from the Windows desktop, the taskbar is automatically moved to the remaining monitor.  However, the insets for this monitor are still reported as 0,0,0,0.  I wager that this matters almost never, but is a bug all the same.  My first guess is that we need to recalculate the HMONITOR structures on a WM_DISPLAYCHANGE as we do for WM_SETTINGCHANGE.  Such a change would require some ordering/locking/threading investigation.