[imps] Tree construction question

[Rob Jellinghaus]

On Fri, Feb 5, 2010 at 2:38 PM, Jonas Sicking <jonas at sicking.cc> wrote:
>On Fri, Feb 5, 2010 at 2:22 PM, Anne van Kesteren <annevk at opera.com> wrote:
>> The idea of the parsing algorithm is not that any input yields a valid
>> output. It just gives back a tree. Whether that tree is conforming is
>> orthogonal. A simple example of this would be unknown elements not being
>> removed (or transformed).
>
>Also note that Firefox does not yet by default attempt to implement
>the HTML5 parsing algorithm. We still use the parser we've used since
>Firefox 1.0. Back in the days this parser indeed tried to always
>output a valid HTML DOM. This idea was eventually abandoned because it
>turned out not to be practical and just caused more confusion and
>trouble than it helped anyone. So even this parser can these days
>output invalid DOM trees.

Thanks, Anne and Jonas.  Good to know the impossible is not necessary :-)  

It would be helpful if the HTML 5 spec added a sentence to that non-normative paragraph, explicitly stating the output of that example, and its anticipated noncompliance.  That would head off future confusion.

Cheers!
Rob Jellinghaus
Microsoft Technical Strategy Incubation