WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: Hardware and Software required for Accessibility Testing

for

From: deblist
Date: Dec 3, 2009 4:45PM

On Thu, 3 Dec 2009, Keith Parks wrote:
> I realize I'm talking about an ideal world here, which the "Table
> Navigation" thread clearly shows we have not yet reached, in terms of
> the standards themselves, or their implementation by AT. But the
> principle that in order to "do the right thing" on accessibility,
> developers need to be testing their sites with multiple versions of
> screen readers, magnifiers, voice-recognition technology, etc., seems
> to be setting the bar a bit too high.

By the same token, though, you could say that if you write
standards-compliant HTML + CSS you shouldn't have to test on
multiple browsers, and until about a year ago that wasn't true at
all. The fact is that the user agents weren't at all standards
compliant, and if you didn't test on multiple browsers you'd find
a site that didn't work at all in Internet explorer 6. (it's not
like that status quo has changed; all that's changed is the
willingness for sites not to work on Internet Explorer 6.) Until
a couple of years ago, any Web developer who actually wanted
to make sure that a site worked in most major browsers had to
test in multiple browsers -- and this is not entirely untrue
even today.

As you say, it's not an ideal world. The reality is that as a
sighted user who browses with non-screenreader adaptive tech, I
find that *many* so-called accessible sites aren't accessible to
me, because of a combination of the necessary fuzziness of the
standards and inadequate standards-compliance on the part of
adaptive technology.

I don't test on every available tool myself, just as I didn't use
to test on every available browser + platform combination. But I
do try to test on Firefox 3, IE 7/8, Opera 9/10, Lynx, and
Safari; XP and OS X; Jaws 10, DNS 10, Mouseless Browsing, and
NVDA. It's a thoroughly incomplete subset and I often don't hit
every item on my own incomplete list, but if standards-compliance
were enough to make all of these work, trust me, I wouldn't still
be testing with such a large set of choices.

-deborah