Re: Cleaning House

On May 4, 2007, at 18:30, Philip Taylor (Webmaster) wrote:

> I did not intend to suggest that one could simply include
> additional elements in an HTML document without formality :
> rather, I was proposing a TeX-like approach whereby one
> could add elements in a more formalised manner, defining
> their syntax (but not their semantics) in terms of existing
> elements.  In your message above, <kappale> and <korostettu>
> might (for example) be defined as sub-classes of <div> and
> <span> respectively (if you intended them as I think).
> In the absence of any CSS rules corresponding to these
> new elements, a browser (user agent) would render them just
> as it would <div>s and <span>s respectively.

Umm. HTML already has the class attribute exactly for that. The HTML5  
draft contains a preliminary proposal for registering the semantics  
of class names to foster better multi-party understanding.

> But by providing
> corresponding CSS rules, you could indicate an alternative
> rendering that might better bring out their meaning.  There
> is no need (IMHO) for the browser (user agent) to /understand/
> the semantics,

Then you are not communicating semantics end-to-end but instead you  
are sending presentation with syntactic indirection.

> any more than a browser "understands" what <code>,
> <kbd>, <samp> or <var> really mean.

If you have a program that extracts code snippets from documents  
authored by people with whom you don't have an agreement about  
conventions, chances are that you'd want them to use <code> so that  
you could leverage the semantics rather than <foobar> plus style.

> The purpose of semantic
> markup is to separate content from form, to improve accessibility,

Sending presentation with syntactic indirection is placebo as far as  
improving accessibility goes (compared to sending presentation  
without syntactic indirection).

> and to provide an infrastructure whereby additional value can
> be derived from a properly tagged document using appropriate
> tools.

To make this work without bilateral agreements, the elements need to  
be commonly specified--not something someone just came up with on his  

> There is no reason at all why a Finn should not mark
> up his/her document using <kappale> and <korostettu> if he/she
> wishes

Yes, there is. Real software has built-in behaviors for the  
corresponding commonly known elements (in this case <p> and <em>), so  
the commonly known elements are more useful to markup consumers. (And  
example with a homegrown <a href='...'> replacement might have been  
better, though.)

Henri Sivonen

Received on Friday, 4 May 2007 16:12:13 UTC