W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > October to December 2003

Re: Untestable success criteria

From: Geoff Deering <gdeering@acslink.net.au>
Date: Mon, 24 Nov 2003 20:41:40 +1100
Message-ID: <3FC1D254.1020100@acslink.net.au>
To: Jens Meiert <jens.meiert@erde3.com>
Cc: joeclark@joeclark.org, w3c-wai-gl@w3.org

Jens Meiert wrote:
>>>All I am doing is making a point that colour combinations are machine 
>>>testable.
> 
> 
> Excuse me for my comments, but I want to add two notes: First of all, where
> should the definition of legibility be set to, what is readable, what is not?
> That might not be a current problem, but I imagine cases where a future
> validator says 'your page ain't readable by visually impaired users' where the
> color scheme e.g. bases on dark blue on light gray background (which is often
> legible); maybe the site author also provides alternative mechanisms to
> improve legibility which are not analyzed.

I think this is where the developer should choose fonts appropriate for 
the device; screen, print, whatever, and set them in relative units.

There are other techniques like using a choice of skins/switcher css for 
the user.  Users need to also take some responsibility for configuring 
their computer and user agent so that their base system responds 
according to their needs; ie font sizes, types etc.

> On the other hand, I basically agree that such a mechanism is constructable,
> I'm convinced that it ain't that difficult to create an application which
> analyzes color values and calculates a value which represents a legibility
> factor. But if you want to go further, you will soon encounter other problems
> which will not be testable, some of them include 'real' Accessibility issues,
> others will be non-testable Usability criterions (and as I often remarked on
> this list: unusable means inaccessible).
> 

Its pretty straight forward to be able to parse the following and 
compare the contrast between the color values and conclude which has 
enough variation in color contrast.  There is probably a need for more 
complex rule sets for various problematic color combinations, like red 
and blue, when combined tend to make it difficult for aging eyes to 
focus on.  But its machine testable.

eg

Ex 1

content {
	color: #000000;
	background-color: #FFFFFF;
}

Ex 2

content {
	color: #33CCCC;
	background-color: #6699CC;
}


Geoff
Received on Monday, 24 November 2003 08:57:53 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:47:26 GMT