2005-06-14 14:39 UTC One False Step
A month ago I posted about my attempts to reverse-engineer IE's drag-and-drop mechanism, including a crash that I found.
I'm still working on that; I'm currently trying to determine exactly what the dataTransfer object is. It seems that it's a magical object, that IE creates a new one for each event, but that, most importantly, setting anything on that object actually always takes effect, even if you set it on the object from a previous event rather than the latest event.
Except when it crashes.
Now, the dataTransfer model that it seems IE is doing is seriously ass-backwards, in my opinion. But the good thing about the fact that it crashes so easily when you try to prove this model is indeed the one IE implements is that there's no way anyone will be relying on that behaviour, so we can specify something much saner in the HTML5 spec.