Highrez

OK thats interesting, it looks like it is sending once when the button is pressed and again when it is released. That should be plenty to point me in the right direction. I'm away for Easter but when I get back (Monday probably) I'll see if I can spot the stupid error in the code that causes that .

OK I think I found the difference between the working and not working. It is an odd one, that for whatever reason, if I send the VK_BROWSER_BACK browser back key to the keyboard buffer, it causes vivaldi to go back twice.
If I just post a WM_APPCOMMAND message to the browser window, it goes back once.

It has nothing at all to do with the original bug discussed in this topic nor the fact that its a chromium based window - other browsers (Chrome and chromium itself don't suffer this reaction to the BACK key being sent) so I would bet that this double back bit is actually a bug in Vivaldi itself as no other browsers (inc. Chrome react in this way to the VK_BROWSER_BACK key!).

Unfortunately, I don't have access to a keyboard with a physical "browser back" key to see how Vivaldi reacts the the physical key (I bet that also causes a double back action but cant prove it easily!).

Anyway... ll that aside, XMBC already has code in to to account for this, and normally that works. So the actual bug is that in XMBC it looks to see if the process is a "modernUI/Store" app because they don't respond to WM_APPCOMMAND and hence XMBC has to send the BACK keyboard key for those apps. The problem is that in my process detection, its setting the modernUI flag to true (incorrectly) whenever the foreground window changes to another in the SAME process and therefore XMBC sends the BACK command differently.

So defiantly a weird one - it should be fixed in the next release.

Thank you for the detailed explanation and for the fix.