W3C home > Mailing lists > Public > www-validator@w3.org > July 2014

Re: Implied ARIA Roles

From: Jukka K. Korpela <jkorpela@cs.tut.fi>
Date: Wed, 23 Jul 2014 07:57:41 +0300
Message-ID: <53CF40C5.5090704@cs.tut.fi>
To: www-validator@w3.org
2014-07-22 15:50, Craig Francis wrote:

> When validating a document with an @aria-sort attribute, the validator
> requires that a @role should also be specified.


> As I'm adding this on a <th scope="col">, isn't this already implied by
> the markup?

No, because the specifications do not say that the scope attribute has 
an impact on default ARIA semantics.

> I've also brought this issue up on SO:
> http://stackoverflow.com/questions/24863531/using-aria-sort-in-validated-html5

I wrote a longer reply there.

In a nutshell, the validator acts according to the specifications. The 
specs require role attributes (on <th>, <tr>, and <table>), if you use 
aria-sort. This sounds redundant, but if you wish to raise an issue, it 
needs to be an issue about HTML specifications.

(I don't think it would be successful. We cannot expect that in general 
a <table> element represents a grid. It can be argued on good grounds 
that it is reasonable to expect so if a cell has a scope attribute, 
since such attributes would be nonsensical in a pure layout table. But 
the rules for setting default ARIA semantics on the entire table on the 
basis of an attribute on a cell would get complicated, and ARIA and its 
relation to HTML is already a mess^H^H^H^H rather complicated.)

Received on Wednesday, 23 July 2014 04:58:11 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 14:18:11 UTC