Re: ISSUE-54 / ISSUE-4.: versioning and doctype-legacy-compat

Hi Thomas,

On Jan 20, 2009, at 4:58 PM, Thomas Broyer wrote:
>
>
>> # Semantic parsers (if any)
>
> Are they any different from browsers?
> I mean, if a document is labelled as HTML 4.01, would such a parser
> ignore an <article> or <dialog>?

Certainly that example doesn't highlight the problem. However if a  
document is labeled ambiguously and the semantic analyzing UA find a  
<small> element, should eat treat it as the <small> from HTML4 or the  
<small> from HTM5. These are basically homographs for two different  
terms in the HTML vocabulary. So without marking the version of the  
document (or without the spec editors sticking to standard namespace  
tenants), the lack of version labeling within the document means the  
semantic analysis breaks.

Another thing that concerns me here is that we're using the HTML spec  
to make a statement about DTDs rather than making those arguments  
within other workgroups and task forces where those arguments should  
be heard. All this debate over versioning, DTDs, compatible DocType  
declarations, etc. could be avoided if we simply created an HTML5 that  
followed the DocType construction we've inherited and make arguments  
to the relevant WGs that DTDs and DocTypes should be dropped/ 
eliminated/repalced/etc.

Therefore another item Karl might have added to his list is:

  # DTD based SGML parsers

which would gain full HTML5 parsing support right out of the box if we  
simply provided a DTD and a corresponding DocType declaration that  
served those UAs (which would simultaneously solve the problem of  
needing a legacy compatible DocType produceable by legacy serializers).

Take care,
Rob