Re: [Usability] HIG has severe "MS Windows" flaws.

["Blad, John Erling" <john erling blad aftenposten no>]

>> Denis Fernandez
>> But the correct solution is to code menu
>> triggering as a broadening cone, instead as a rectangular region, so that
>> the user can drift off from the parent entry on his way to the new
>> submenu.

> Calumn Benson
> Hmm... I believe this is pretty much how it does work (and certainly
> seemed to when I tried it there just now)-- I vaguely remember the
> original discussions about it three or so years ago.  Maybe it needs
> some tweaking, but I've never had a problem with it personally.

There are actually a number of ways to do this. Some are better than others(!) and some are nearly impossible to implement "right". The cone idea seems like a good one at first but it is very difficult to "learn". The reason is that you don't have one cone but one for each submenu.

Also how do you know when to dismiss the submenu? You can not any longer move in diagonal down to open a new submenu. :)

Probably the "right" implementation would be neural networks that learn how to reach one specific item in a submenu from a starting point, given the speed and acceleration of the pointing device, and given the knowledge of other submenus at the same level. 

Such a network would need to be trained for an average user and with option to retrain for new users, especially when they have special needs.

I do think a simple timeout with a few alternate values given the existence of other submenus at the same level will suffice for most users. (That is, you hold the timeout a little bit longer if there are no submenus attached to the menu item you move over.)

John