- From: Henri Sivonen <hsivonen@iki.fi>
- Date: Tue, 30 Sep 2008 19:48:12 +0300
- To: "Bruce Lawson" <brucel@opera.com>
- Cc: "Gez Lemon" <gez.lemon@gmail.com>, "Steven Faulkner" <faulkner.steve@gmail.com>, "Al Gilman" <Alfred.S.Gilman@ieee.org>, "W3C WAI-XTECH" <wai-xtech@w3.org>
On Sep 30, 2008, at 15:42, Bruce Lawson wrote:
>> It doesn't resolve the whole issue of ensuring that ARIA is used
>> correctly with a native markup language, but if the ARIA
>> specification
>> is available in a machine readable format [...] we could build a
>> basic
>> validator to ensure at least the ARIA part is used according to its
>> specification.
>
> Most authors would be using it with HTML - so that would be a good
> start. There are several options:
>
> 1) an HTML validator that simply ignores any ARIA information - so
> it doesn't conformance check the ARIA at all, but doesn't fail a
> page because ARIA is present. (That's useful for knowledgeable early
> adopters who have an organisational policy that requires valid
> markup. Perhaps they are committed to WCAG AA which "requires"
> validity, for example).
>
> So perhaps it throws a warning ("please check the syntax and
> applicability of the ARIA attributes you use") but no error.
You can't get further than this using a DTD-based validator.
> 2) an HTML validator that also syntax-checks ARIA attributes to make
> sure it's used according to its spec
>
> 3) a full validator that ensures ARIA is used correctly with the
> native markup language. I believe this is impossible now as there
> are issues about how to specify this.
>
> So (3) is impossible, (2) would be very useful, although the spec
> might change, while (1) could be built now, and at least removes
> impediments to adoption, even it isn't "helping".
Validator.nu is doing the impossible #3 by making up stuff that isn't
in the specs (yet?).
--
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/
Received on Tuesday, 30 September 2008 16:48:53 UTC