Right now relatedTarget is specced as being null on blur/focus. Why don't these just match focusout/focusin?
Doing some spec archaeology, it looks like relatedTarget was specced as having the same value as target on blur/focus and then set to null in response to http://lists.w3.org/Archives/Public/www-dom/2010JanMar/0010.html.
It seems to me that there's no downside to making relatedTarget do the same thing on focus/blur that is does on focusout/focusin respectively. From a web developers perspective, the only difference should be whether the event bubbles or not and whether it fires before/after focus has shifted.
I know this is late in the cycle, but I just noticed this and this seems reasonable to include in the level 3 version of this spec.
Fixed in latest ED.
relatedTarget for blur copied from focusout
relatedTarget for focus copied from focusin
(The bug says to "do the same thing on focus/blur that is does on focusout/focusin respectively", but I think that's reversed (or else I'm not understanding something).