Re: EARL Testcase Clarification

On Thu, 4 Dec 2003, Chris Ridpath wrote:

(rearranged)

> From: "Shadi Abou-Zahra" <shadi@w3.org>
> > this is the way i interpreted it as well. while the test is defined by
> > the tool, the testcase is the requirement of the guideline. this way a
> > tool can conduct multiple tests to judge conformance to a checkpoint.

Indeed, that's how I first interpreted it, and how Valet works in
practice.  It works well if you're happy to reference one guideline
per test case.

>
> OK. And what if the test is required by several guidelines?

That is a good point, and one I've looked at, but that hasn't
yet made it into anything operational.  The simple/practical
answer is to select one guideline as most-relevant and drop
references to the others.  An inference engine can still keep
the reverse (guideline requires pass the test) dependency.


> Would you just repeat the earl:Testcase block?
>
> Example:
>
> <earl:Testcase rdf:resource=http://guidelines.org/guideline.html>
>  <earl:testId rdf:resource=http://checks.org/test25.html />
> </earl:Testcase>
>
> <earl:Testcase rdf:resource=http://other-guidelines.org/requirement.html>
>  <earl:testId rdf:resource=http://checks.org/test25.html />
> </earl:Testcase>


My preference is for an additional layer of abstraction.  The EARL
testcase refers to your test (whatever that may be), which in turn
references each relevant guideline.

You'll presumably let us know if/when you implement anything relevant?

-- 
Nick Kew

Received on Thursday, 4 December 2003 13:47:47 UTC