this is also something Scirra should think about, if there's a way to prevent that - maybe an "inflight" Errormessage?
In the better of worlds, that would be cool.
Now down to earth in 2012. How do you pull that off ? ^^
It is some sort of real time code analysis that would have to understand the logic you wanted to implement as opposed to the logic you actualy implemented.
I don't think such an AI process is doable by a startup-of-two.
Don't get me wrong, it would be damn awesome. But as far as reality goes, it is unlikely it will see the light of day.