[whatwg] several messages about XML syntax and HTML5

[Ian Hickson]

On Thu, 7 Dec 2006, Sander Tekelenburg wrote:
> At 00:45 +0000 UTC, on 2006-12-05, Ian Hickson wrote:
> > 
> > The point is the browsers all do the same thing, and that's well-defined
> > and documented [...]
> > if all the browsers do Z, then since the author presumably checked
> > his page with at least one browser, and it did Z, and he didn't correct Y,
> > then the assumption that Z is what Y was supposed to be is IMHO a safe
> > one.
> 
> OK, agreed. For HTML5 documents that seems a reasonable assumption. (Assuming
> the browser indeed is HTML5 compliant.)
> 
> I'm still somewhat sceptical about the reality of this though, as it relies
> on the author checking the document with at least one HTML5-compliant
> browser. This reliance would provide Microsoft an easy attack vector on
> HTML5: give away free authoring tools (or even lure people into paying for
> them) that produce code that triggers HTML5-compliant browsers to, as per the
> HTML5 spec, stop processing the document, and have InternetExplorer present
> such documents as if they're fine.

Your assumption seems to be that the interoperable handling of HTML 
documents is to somehow abort processing. This is not the case. Each error 
has explicitly defined error-recovery behaviour. The user need never know 
there was an error. The error-recovery behaviour described in HTML5 is as 
close to what IE does today as humanly possible.


> What then? Will every other browser really tell the user that it won't 
> try to interpret what the author might have meant?

No, no browser will say that. That's the way XML works, not how HTML 
works.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'