Re: Purpose of Controls SC

Michael Gower wrote:

> I'm objecting to the WG essentially building our own spec. I believe this is happening because the web ecosystem which would allow Purpose of Controls to be met without a lot of heavy lifting by authors is very immature.



It is an immature space (in terms of implementations). The chicken and egg aspect I talked about earlier is that it will not mature without that stalemate being broken.



Websites won’t add things unless there are user-agents, and no-one will make an extension (user agent) if there is no prospect of websites adding things.





> We are being asked to give approval for something clearly in a state of flux …

> The closest parallel that has been cited to Purpose of Controls is 4.1.2 Name, Role, Value.



I think it is important to note that concept is not in flux, the details of the items are.



This SC (at AA) is much closer to the landmarks part of ARIA, it is far simpler than dynamic ARIA widgets.



As an author: Which of the links on my page match this list? Ok, that’s Home, Sitemap and about us. Right, which technique do I use? I’ll add coga-destination (or I’ll use title with the description.)

As a tester: Does the home link have something associated from the list? Yes/No.

As a user-agent: Check each link/field/button for meta-data with a calculation similar to accname, and either add symbols or highlight things with descriptions and make those available.





> When 4.1.2 was introduced it was light-weight for many authors. You essentially met it if you were just using standard HTML components. If you added in custom widgets, you could supplement HTML with a fairly small subset of aria attributes.



That’s sort of true, but in practice people already were using custom widgets and ARIA was the only accessibility supported way of meeting WCAG 2.0. Using ARIA widgets is still not straightforward, and adding these terms/descriptions is easy by comparison.





> While not expecting Purpose of Controls to be that light-weight, I also think it should be achievable and testable without overwhelming authors.



I think that is true now. The list might be long, but the way you use it is very simple. Is this a home link? Yes: add this token or this description.





> My preferred way of doing that is to base it on a spec (or multiple specs) with good credibility and momentum.



Those won’t gain momentum unless there is some prospect of it being required. This SC could provide the seed, a core set of conventional names that can later be moved into a separate spec.



Cheers,



-Alastair