Re: Black-box equivalence of parsing fragments directly into context node

Henri Sivonen wrote:
> This seems like it's still worth pursuing, considering that the tree 
> builder--per spec--shouldn't fire mutation events for insertions 
> performed by the parser, so the behavior should be that mutation events 
> don't fire during the parse anyway.

Right.

> Is there a general mechanism for deferring content-related events?

That doesn't seem like an HTML spec question (and same for a lot of 
other things in this thread).  Maybe follow up to the m.d.t.dom or 
m.d.platform?

> Firefox is aware of the outer form and throws away the inner one. 
> WebKit, Opera and HTML5 as currently drafted are unaware of the outer 
> form and create a tree with the inner form as the child of the outer. 
> IE8 experiences an error.
> 
> Since WebKit and Opera already have the HTML5 behavior and IE8 fails to 
> have either behavior, I guess going with the spec is OK.

Yeah, I suspect that's fine.

> Speaking of the load event for SVG subtrees: During normal parse, when 
> the load event for an <svg> element fires, can a handler react to it 
> before there has been an event loop spin?

Yes, absolutely.  The event fires synchronously, as required by the spec.

> That is, should I treat the 
> firing of that event as a potential trigger of document.write() entering 
> the parser?

Yes.

-Boris