Re: role vs aria-role

On Tue, 18 Aug 2009 09:08:56 +0200, Jim Jewett <jimjjewett@gmail.com>  
wrote:
> It is too late to say "Don't use a bare 'role' ", but I don't think it
> is too late to say "Also support 'aria-role' " or even "if 'role' and
> 'aria-role' disagree, believe 'aria-role' ".
>
> The shipped implementations won't have that extra piece of
> functionality, but things that do work won't stop working.  (And since
> aria-* is invisible metadata, I suspect it won't even lead to many
> opportunity costs -- the people who switch to the more consistent
> 'aria-role' in the very short term are likely to be people who were
> already having error-prone on this attribute.)

Besides consistency there does not seem to be much in favor of this. And  
consistency is not enough for duplicating a feature in my opinion,  
especially as it does increase the complexity of the language. Not just  
for implementors, but also for authors. (Also, if we did this introducing  
another bunch of more proper aliases might be considered nice to have too,  
e.g. encoding="" instead of charset="". And I definitely do not want to go  
there.)


-- 
Anne van Kesteren
http://annevankesteren.nl/

Received on Tuesday, 18 August 2009 07:40:42 UTC