W3C home > Mailing lists > Public > public-wai-ert@w3.org > February 2005

Re: What was the test, Re: EARL Scenarios

From: Charles McCathieNevile <charles@sidar.org>
Date: Thu, 24 Feb 2005 16:51:51 +0100
To: "Andrew Kirkpatrick" <andrew_kirkpatrick@wgbh.org>
Cc: public-wai-ert@w3.org
Message-ID: <opsmpk0pdqw5l938@saturne>

On Wed, 23 Feb 2005 21:06:44 -0500, Andrew Kirkpatrick  
<andrew_kirkpatrick@wgbh.org> wrote:


>>  For common tests where the rule is published, such as the ones Chris  
>> has  developed, the publisher of the test can provide a URI for it, and  
>> voila.

> Fair enough, but in practice there are few "common" tests that  
> evaluation tools share.

I think this is temporary. More and more tools are using the Markup and  
CSS validators, for example. Most tools do a simple test to determine  
whether an alt attribute exists for an img element.

With the publication of the WCAG Test Suite, we start to get tests that  
can be readily shared. (Further development, which we trialled in WAINu,  
is to pulish code or a service that performs a particular test, so you can  
do run-time configuration of the testing. Seems workable).

>> If, for example, I find a test that some tool does, and can't find a  
>> URI  used to identify it, there is nothing that stops me from making  
>> one up. If  another person makes up a different one for the same thing,  
>> then we  realise we should have looked around and talked about things  
>> first, but  the technical solution is the OWL property sameAs - which  
>> it is probably  useful to have implemented in any system that reads  
>> EARL.
>
> I couldn't know less about OWL, but there would potentially need to be  
> more complex referencing than 1:1 declarations like 'Same As'.  Some  
> tools do one test that others do in 5 separate tests, while in other  
> cases a tool  has a test that checks for situations 'a', 'b', and 'c'  
> and another tool has 2 tests, one that checks for 'b' and 'd' and  
> another that checks for 'a', 'c', and 'e'.
>
> This isn't really a problem if the solution is to just create another  
> URI to identify tests, but when it is necessary to align test results  
> the connections need to be there.  I fully support highly granular tests  
> so that a testing tool that tests for multiple things can have multiple  
> 'sameas' declarations, but I do think that come up with this list is  
> deceptively difficult.

Right. Beyond "SameAs", OWL lets us declare equivalence, and similar  
things.

For example, I can say that passing a particular collection of 5 tests is  
equivalent to passing a different collection of 6 tests. Or that passing a  
collection of 20-odd checkpoints is equivalent to passing the "test"  
"conforms to WCAG 1 level A"...

Some of this is written up (perhaps badly - comments welcome) in  
http://www.w3.org/2001/sw/Europe/200407/earl/ and Shadi did some more  
which is (I think) linked from there.

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 Thursday, 24 February 2005 16:00:42 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:55:52 UTC