[html4all] Numbered headings (was HTMLWG)

[Vlad Alexander (XStandard)]

Hi Henri,

> I suggest first coming up with a suggestion for outline
> generation behavior when a UA gets a mix of <h>/
> <section> and <h1>-<h6>.

1. Keep the old semantics of h1 to h6.

2. section/h is a child of any numbered heading. For example:

<h1>Vehicles</h1>
<h2>Cars</h2>
<section>
	<h>Domestic</h>
</section>
<section>
	<h>Imports</h>
</section>
<h2>Trucks</h2>
<section>
	<h>Domestic</h>
</section>
<section>
	<h>Imports</h>
</section>

The above markup would have the following outline:

Vehicles
	Cars
		Domestic
		Imports
	Trucks
		Domestic
		Imports

3. section element must contain one h element (as the first element) and cannot contain numbered headings.

Regards,
-Vlad
http://xhtml.com




-------- Original Message --------
From: Henri Sivonen
Date: 2007-11-26 1:11 PM
> On Nov 26, 2007, at 18:20, Vlad Alexander (XStandard) wrote:
> 
>> This would not affect legacy pages. This will affect legacy tools.  
>> For example, an accessibility checker tool or markup validator  
>> reports to users on incorrect use of headings.
> 
> A new spec is inherently incompatible with legacy validators and  
> checkers. Browsers ignore unknowns, but validators are all about  
> making noise about the unknowns. A legacy checker wouldn't know about  
> <section>/<h>, either. A new spec needs new validators.
> 
> <h> has been considered. The reason why HTML5 uses <h1> with extended  
> semantics is that the way HTML5 is defined, the outline algorithm  
> produces (supposedly--I haven't implemented it) a sane document tree  
> even if the input mixes <section>s and old-style <h1>-<h6>. XHTML2  
> failed to define the interaction of <h> and <h1>-<h6>.
> 
> So if you propose revisiting <h>, I suggest first coming up with a  
> suggestion for outline generation behavior when a UA gets a mix of <h>/ 
> <section> and <h1>-<h6>.
>