Re: Confidence Claims - more discussion

Chris Ridpath wrote:
> 
> Shadi wrote:
> 
>> IMHO, it is bad practice not to record which tests passed.
>>
> Our checker records only those tests that have failed or tests that we
> can't tell and may have failed. If we recorded all the tests that were
> run then the report would be huge. I don't see anything wrong with
> recording tests that have passed but I think there's a simpler way of
> doing this.

Chris, we seem to have quite a high degree of unanimity amongst
tool implementors:-)

>> I think this is useful "proof" of what has been tested, how, and why
>> the tool claims certain assertions.
>>
> Yes, I agree. The conformance statement should state which tests have
> been run and if all the tests pass then the content conforms. For
> example the conformance claim could state that test #1 was run (all
> images have an alt attribute) and if the test passes then the content
> conforms.
> 
> I don't think it's necessary to state that the first image has an alt
> attribute and the second image has an alt attribute and the third image
> etc. You just need to state that all images have an alt attribute.

Well, yes.  But in effect, that's just a detailed statement of the
tool's capabilities.  This tool tests ALT attributes for every IMG;
therefore every IMG that isn't reported has passed that test.
It's no more than a static document.

-- 
Nick Kew

Received on Thursday, 2 June 2005 13:59:28 UTC