W3C home > Mailing lists > Public > public-html@w3.org > July 2007

Re: DI element [Re: html 5 and accessibility issue]

From: Robert Burns <rob@robburns.com>
Date: Tue, 3 Jul 2007 17:33:20 -0500
Message-Id: <43298F43-4A06-4E23-946C-F6FE8BF76C6A@robburns.com>
Cc: Thomas Broyer <t.broyer@gmail.com>, public-html@w3.org
To: Andrew Ramsden <andrew@irama.org>


On Jul 3, 2007, at 5:11 PM, Andrew Ramsden wrote:

> Thomas Broyer wrote:
>> But the point was that Rob was trying to style a table where a tbody
>> were to be inserted/implied between a *tr* and a *td* rather than
>> between a table and a tr.
>
> Another good point :)
>
> I put together some test cases using "table td" and "table tbody td":
> <http://wg.irama.org/html/test/table/implicit-tbody.html>
>
> ... And indeed it was the second selector that always matched (in  
> FF2). In fact viewing the generated source (DOM source), there was  
> always a tbody inserted. And the same is true of other browsers  
> tested (IE6, Opera9.2, Safari3.0.2).
>
> You learn something every day :)
>
>
> I might be missing something though. I can't see that it would it  
> be a problem for UAs to treat di the same way?

I think the problem is that existing browsers will not know to infer  
the <di> element. What I understood from the original proposal was  
that authors would need to make the <di> element explicit to make use  
of this facility (this is how UAs parsing tables as "application/xhtml 
+xml" work with tbody). So the proposal would be to the work the way  
XHTML works with tbody and not the way HTML works with tbody.

I think that should work for backwards compatibility and graceful  
degradation, though we'd need to test it to be sure.

The problem I see with using for="ids" is that it puts too much  
flexibility in the authors hands. You'll end up with @for pointing to  
an id that's not even a <dt> or to the <body> or whatever. This could  
treated  as a silent error, but I think the nested element structure  
is simply easier and less intimidating for novice authors and it  
accomplishes precisely what we're looking for and no more.

Take care,
Rob
Received on Tuesday, 3 July 2007 22:33:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 May 2012 00:16:02 GMT