WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: Recommending Widget Markup that Doesn't Work in Reality

for

From: Alastair Campbell
Date: Feb 11, 2016 2:51AM

Brooks newton wrote:

> Do you recommend ARIA-enabled solutions for complex widgets, when you know
> full and well that there is no combination of OS, browser and assistive
> technology that will render the particular widget accessible, in terms of
> actual performance?


No.



> Over the years I've seen a lot of theoretical solutions that pass WCAG
> compliance muster, which would in no way pass a more stringent standard,
> such as the performance objectives central to standards like the Twenty
> First Century Communications and Video Accessibility Act (CVAA).
>

I'm not really convinced by that line of thinking, part of testing WCAG2 is
things like "All functionality of the content is operable through a
keyboard interface", and "Information, structure, and relationships
conveyed through presentation can be programmatically determined or are
available in text."

Combine those criteria with the concept in WCAG2 of accessibility
supported: "Technology features can only be relied upon to conform to WCAG
2.0 success criteria if they are used in a way that is 'accessibility
supported'." I.e. works in common user-agents.

That means testing in common user agents, and if it doesn't work, it does
*not* "pass WCAG compliance muster."



> Do any accessibility experts out there ever tell their clients that some
> design patterns are inherently inaccessible, given the current state of
> technology?


Yes, most commonly because an enthusiastic developer has tried to apply
ARIA in a way that isn't supported (yet), and there needs to be a practical
solution in the meantime. Usually that situation could have been avoided if
(realistic) accessibility were considered earlier, and there is often a
simpler solution than the one they were attempting.


Last question: Why don't we have an accessible, freely available
> centralized repository of fully vetted ARIA design patterns, with a full
> compatibility matrix that displays OS, browser, AT, etc. compatibility?


Because that is a shed-load of work that would be out of date before it was
completed.

The matrix that would involve is mind-boggling. Browser testing is browsers
multiplied by OSs, you have to multiply perhaps 5 OSs by 4 or 5 browsers.
In practice, with certain OSes not having many browsers, you can usually
test in less than 10 combinations to cover most of the usage. Add ATs and
you are multiplying 5 * 5 * 5 ish? Taking off non-existant combinations
you'd still probably be looking at over 30 combinations? (I'm open to
corrections on that!)

Then multiply by the number of combinations of ARIA design patterns within
different contexts, and it would probably take longer than a month if all
the accessibility agencies dropped their paid work to tackle it together...
not practical.

Steve Faulkner (& I assume others at TPG) put together:
http://html5accessibility.com/ which is browser support for HTML5 tags.
That's a lot of work to do and keep updated (thank you Steve!). Then
multiply that by ATs and design patterns...

In practice I think most already have a good idea of what works in general,
and can test specific examples quite efficiently. In our work that is
usually what's needed, and we can't publish those examples because they are
not ours to publish.

There was a move to try something like this a couple of CSUNs ago, and I
won't speak for others, but for me it turned into quite a daunting,
repetitive and dull exercise; there was always something more interesting
to do.


Are the competitive pressures between companies /agencies keeping us from
> reaching a consensus on what advanced cookbook solutions work now and which
> theoretical models need additional support before we can recommend with
> confidence to clients?


Overall I think the accessibility community is pretty good at sharing
information, just look at what people publish (compared to other
industries, or other areas of web development).

For ARIA these come to mind:
http://whatsock.com/tsg/ (probably the closest thing to what you are asking
for)
https://www.w3.org/TR/html-aria/
http://heydonworks.com/practical_aria_examples/

The "theoretical models" aspect is covered by the W3C in documents like:
https://www.w3.org/TR/wai-aria-practices/#aria_ex

It is just that the context of the implementation often affects its
practicality...

-Alastair