[html4all] role attribute keywords for embedded media

[Leif Halvard Silli]

Robert J Burns 2008-08-22 16.53:

> Re: role for embedded media [1]
> 
> On Aug 22, 2008, at 5:44 PM, Leif Halvard Silli wrote:
>> Robert J Burns 2008-08-22 12.57:
>>
>>>> And in that regard, for what the Wiki page call 'layout or
>>>> spacer', how about role="space" instead, to signify images which
>>>> represent space, and therefore a space character as fallback?
>>
>> What did you think about this? You see, I think that many authors
>> would not understand what kind  of @alt a role="spacer" or
>> role="layout" woud require. Many would think that they should have
>> alt="". But my view is that they, for safety, should have alt=" "
>> and not alt="".
> 
> I'm not sure I recognize the use case for it. I guess you're saying  
> some use spacers just as they would a U+0020 space? Or perhaps the way  
> XMl/HTML attribute values replace sequences of other whitespace values  
> with a single U+0020? I guess that could make sense. To me this would  


Yes, I had in mind that several space(r)s would just become one. 
Often alt="" would be fine, but alt=" " would be safer, so words 
would not be concatenated if embedded elements with decorative or 
layout purposes was all that separated them.

> just be needed for legacy UA processing. An HTML5 UA could simply read  
> the 'space' keyword and then know to replace the image with a U+0020  
> space. Though that does suggest we would need a separate 'space'  
> keyword and leave 'decorative' for frills and flourishes and the like.  
> What about layout v. space or layout v. decorative? Do we still need  
> three separate keywords (layout, space, decor)?


Is 'decor' a new proposal, instead of 'decorative'? I agree that 
'decorative' is not so good, as it is an adjective. The other 
roles appears to be substantives. I would prefer 'decor' or 
'decoration'.

My thought would be to have only have two: "space" and "decor" (or 
"layout" and "decor").

One could have had only one role, as well, role=decor, and say 
that authors had to choose between alt="" and alt=" ". But then 
there would be no way programmatically validate whether the @alt 
had been used correctly or not.

 
> [1]: <http://esw.w3.org/topic/PF/XTech/HTML5/RoleAttribute#head-35490581f69f2d7a9157e4ba18f3a20a18956cac 
-- 

leif halvard silli