W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > January to March 2005

Text only and testing needs Re: accessible banking:

From: Charles McCathieNevile <charles@sidar.org>
Date: Tue, 08 Feb 2005 13:00:58 +0100
To: "Nissen, Dan E" <Dan.Nissen@unisys.com>, "wai-ig list" <w3c-wai-ig@w3.org>
Message-ID: <opslvnnw0fw5l938@saturne>

On Mon, 7 Feb 2005 10:03:56 -0600, Nissen, Dan E <Dan.Nissen@UNISYS.com>  
wrote:

> ... the last time I tried to get a comprehensive response
> together on how to comply with the stated objectives of the members of
> this list, I came up with 8 separate environments that needed to be
> separately tested for usability,
>
> A text only site would presumably meet all the objectives of this group

As a member of this list I have to state that a text only site would not  
meet my objectives in providing accessibility in general. It is something  
that can enhance accessibility for a small group of people with specific  
disabilities, if done right, but I generally consider that this involves  
more work than just making the original version work for those people...

> ...  So, the investment to get
> to where this group would be happy is:

> 1. A text only site tested in Firefox on 3 or more platforms

Windows, Mac OS X, some flavour of Unix - for example. (At sidar we also  
test on Mac OS 9 using the screen reader OutSpoken, which is what people  
have until VoiceOVer comes out with the next version of OS X, but I am not  
sure if our tester uses Firefox).

> 2. A text only site tested on Internet Explorer on several versions of
> IE

Which ones? Do you need to test across different versions of Windows? It  
depends where you are perhaps - for South America I would suggest that you  
really need to go as far back as Windows 95, and in general 98 (although I  
believe it is no longer supported, it is still used

Still, I would skip the text-only site, so this means the comments apply  
below...

> 3. The main site that works with both IE and Firefox for non-disabled
> persons to meet the image needs
> 4. Test one of these on Macintosh under probably 3 browsers

I would use Firefox, IE, Safari and Opera

> 5. Test under Opera in combination with several OSs

I do this for Symbian (a common mobile phone platform), Mac OS X, Windows,  
and another Unix.

> 6. Test under Lynx on Linux and Windows

I use Lynx extensively on Macintosh. But I have not noticed that it varies  
at all across platforms, and I would be confident testing it on my main  
platform (by chance for me that is usually OS X. Only thing to watch for  
is what versions are available across platforms - for Mac OS 9 in  
particular it tends to be an older one).

So I get something like the following for a full testing setup:

Windows 98, XP:

Opera, Firefox (latest full release version)
IE 5, 6

Mac OS X

Safari, Opera, Firefox (latest full release version)
IE 5.5

Unix running X (Gnome and KDE for full points)

Mozilla (which is not quite the same as Firefox - particularly with regard  
to accessibility options), Konqueror, Opera

Mobile platform

Opera, "default browser" (depends on your platform - ideally you'd at  
least do WinCE with IE and Symbian).

OS of choice

Lynx (latest version, latest version available across all platforms)

This is 20 testing setups, although you can run them all with a high-spec  
Mac running VirtualPC and a Linux installation, plus a couple of mobile  
devices.

> Each test might require over 100 different web pages be tested for a
> typical banking application.

True, but the basic testing to be done is in the order of a few hours per  
setup and a couple of days of analysis to design sensible testing patterns  
in the first place. If the test setup is well designed it should be  
possible to use a lot of results from early tests that you know will apply  
for later tests - for example alt-text behaviour can be predicted across  
almost all browsers without needing to test it 20 times, CSS testing can  
be done in parallel with other stuff, and so on.

In my experience the hardest things are javascript (because it is so often  
badly written in the first place), layout (because it is relatively  
inconsistent across platforms and people try to do stuff that is  
unreliable), testing with assistive technologies that use non-standard  
system access methods to improve performance, and testing badly-designed  
sites (which throw up more problems so take longer to get through).

> This kind of investment is significant to
> the banks, etc. who need to absorb all this to get a fraction of their
> clients going.

Hmmm. Setting up the infrastructure to do this costs something (mostly in  
terms of adjusting workflow and management processes). The ongoing costs  
are also non-zero. But my experience is that they are not massive in terms  
of a typical large-scale banking system. As frames of reference I am using  
things like

+ the decision by the Australian Human Rights and Equal Opportunies  
Commission that providing TTY devices is a "reasonably" (in the Australian  
legal sense, applicable across the kind of places most of us live in)  
cheap thing for telcos to accommodate the proportion of their clients who  
need them,
+ the cost of providing physical access to banking facilities,
+ what amounts to an educated guess at the size of the "fraction of  
clients" we are talking about,
+ what I know of the overall costs of such projects,
+ an assumption of ongoing commitment to accessibility...

> Yes, it is a "right" to have access, but it is not clear
> exactly how much of this is required to provide the rights.

True. In my experience it is not clear to anyone what the actual usage  
patterns are, let alone who is being excluded and therefore leaves no  
readily visible information.

Developing a methodology that accurately quantifies how much stuff is  
"needed" is a very difficult job. Like developing text-only sites, I think  
that to a large extent better returns can be made on doing simple  
qualitative analysis to identify potential problems, then fix them.

> And, I'm sure I can find a lot of people who would assert that Linux is
> only cheaper if your time is very inexpensive.

I'm sure we could have this argument for 3 years and not reach a consensus  
:-). Note that the price of people's time versus productn costs varies  
enormously in a global context, changing a working environment almost  
always costs something, and I don't think that investing in this argument  
is an effective way to achieve accessibility either :-)

> We need good standards that allow us to not have to do all of this
> testing and building of separate solutions, and we need to work hard on
> the vendors of browsers to do the needed work to make them compatible.

Indeed. As noted above and elsewhere, we also need expertise in using what  
is available.

Cheers

Chaals

-- 
Charles McCathieNevile - Vice Presidente - Fundacion Sidar
charles@sidar.org                      http://www.sidar.org
     (chaals is available for consulting at the moment)
Received on Tuesday, 8 February 2005 12:08:23 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 19 July 2011 18:14:19 GMT