W3C home > Mailing lists > Public > w3c-wai-er-ig@w3.org > October 2003

RE: EARL And Validator

From: Charles McCathieNevile <charles@w3.org>
Date: Wed, 29 Oct 2003 04:09:40 -0500 (EST)
To: Shadi Abou-Zahra <shadi@w3.org>
Cc: 'WAI ER IG List' <w3c-wai-er-ig@w3.org>
Message-ID: <Pine.LNX.4.55.0310290403190.13041@homer.w3.org>

Hi Shadi,

as I understand it you can't use the full checkpoint URI for the Assertion
because then it becomes a global claim about what that URI represents, and as
soon as two of them conflict you have a problem. So you use a local URI for
each assertion, which refers to the checkpoint which was used as a test case.
In my Xform toolos I ask people to provide manually determined data about
each WCAG checkpoint so that is the test case but you are quite correct that
in most cases tools do smaller atomic testing, so should identify their
particular test - this seems to be what Chris is doing already. (Chris, is
that right?)

At some point it would be good to have a way of explaining that something
which passes tests X, Y and Z therefore passes test A - this has been assumed
throughout the development of EARL but never got into the spec. There is a
schema for built-in logic properties for cwm (TimBL's RDF processor) which
would allow us to deal with this.

One difficulty is that this is getting towards rules and queries, which are
not yet standardised for RDF - perhaps we should think about defining a
temporary structure until then. (I don't know - it hasn't been an issue given
that this group was out of charter and couldn't republish a spec anyway...)

cheers

Chaals

On Wed, 29 Oct 2003, Shadi Abou-Zahra wrote:

>
>hi charles,
>
>what do you think of using the WCAG checkpoint for the Assertion (for
>example "http://www.w3.org/TR/WCAG10/#tech-text-equivalent") and the
>tool specific test for the TestCase (for example
>"http://www.some-vendor.com/tool/#testcaseXY").
>
>some test cases could also be conducted manually (for example the test
>"check if the alt-text resembles the image" located at
>"http://www.some-vendor.com/manual/#testcaseXY").
>
>my aim is to be able to collate the results of different tools but at
>the same time allow the tools to have their own sets of test cases. only
>if i can map an arbitrary test to a checkpoint can i accomplish this.
>
>here is what information i could end up with:
>
>* summarize by assertor
>  X assertors conducted tests on checkpoint Y
>
>* summarize by assertion
>  X tests were conducted to evaluate checkpoint Y
>
>does that make sense?
>what do you think of this?
Received on Wednesday, 29 October 2003 04:10:05 GMT

This archive was generated by hypermail 2.2.0 + w3c-0.30 : Thursday, 9 June 2005 12:10:42 GMT