- From: Joseph Scheuhammer <clown@alum.mit.edu>
- Date: Thu, 9 Mar 2017 14:38:38 -0500
- To: "White, Jason J" <jjwhite@ets.org>, Rich Schwerdtfeger <richschwer@gmail.com>, Alexander Surkov <surkov.alexander@gmail.com>, Bogdan Brinza <bbrinza@microsoft.com>, Dominic Mazzoni <dmazzoni@google.com>
- Cc: ARIA Working Group <public-aria@w3.org>
On 2017-03-09 1:44 PM, White, Jason J wrote:
> So far as I can remember, we don't specify anywhere what checking user agents should perform.
We do, and it amounts to doing little validation, although there is
some. There is an "Author Errors" section in the Core-AAM:
https://rawgit.com/w3c/aria/master/core-aam/core-aam.html#document-handling_author-errors
There does not appear to be anything specific in that section about
using an aria-* attribute with a role that does not support it, other
than the basic "do not validate". The closest is the paragraph that
begins, "If a WAI-ARIA property contains an unknown or disallowed value,
..." That's the case where the aria-* attribute applies, but its value
is not one given in the spec; for example, aria-current="fido". There
are rules to handle that.
> If that's the case, then I suppose it's implementation-defined, and authors shouldn't rely on any specific behavior if they decide to use ARIA attributes contrary to specification.
I suspect that if the AAPI object with the separator role has means to
expose aria-expanded, then the user agent will map it, but if there is
no such means, then the user agent will ignore it. That suggests the
validation occurs below the user agent and at the level of the AAPI
itself. Right, authors can't assume any consistent behaviour in such cases.
--
;;;;joseph.
'Die Wahrheit ist Irgendwo da Draußen. Wieder.'
- C. Carter -
Received on Thursday, 9 March 2017 19:39:15 UTC