Re: Common failures (was: Common failures and baseline)

On Wed, May 24, 2006 at 09:58:37AM +0200, Johannes Koch wrote:
> With 'accessible version' and 'alternate version' you mean different web
> units like in Situations A and B in "How to Meet Success Criterion 4.2.1"?
Without making any claim on gregg's behalf, yes, I am sure that's what was
> >RE your second question:  I am not sure what you mean by "the technique
> >must not interfere with a listed common failure."   I'm not sure what
> >interfering with a failure means?
> With interfering I mean the following: You write some content. The result of
> performing the test procedure for a listed common failure for SC n.n.n on
> your content is "FAIL". Someone invents a technique, which he thinks to be
> sufficient to meet SC n.n.n. In the reply to Tim, that I quoted, you said
> this is possible (non-listed technique). The procedure for testing your
> content with this technique would result in "PASS". So it is not clear
> whether your content PASSes or FAILs the SC n.n.n.
> IMHO this must not be possible. The test for a non-listed sufficient
> technique must not result in "PASS" when the test for a listed common
> failure results in "FAIL". The test for a non-listed common failure must
> not result in "FAIL" when the test for a listed sufficient technique results
> in "PASS".
While this would be desirable, I don't think it is attainable. The failures
can only relate, I think, to the listed techniques: if a failure occurs, then
the listed techniques haven't been properly implemented and consequently
(subject to the next sentence), the success criterion isn't met. However, if
you use an unlisted technique, its success or otherwise should depend solely
on the test procedure for that technique. Thus the content can fail to
implement listed techniques (due to a common failure or otherwise) but still
meet the guidelines by implementing an unlisted technique.

Here is an example of why I think your stronger requirement isn't feasible.
Suppose my baseline includes XHTML and a version of the RDF specifications
currently under development by the Protocols and Formats working group (see
their Role Taxonomy and States and Properties, available publicly from the PF
home page at

Now assume that my XHTML fails to provide certain information required by a
success criterion, and in doing so it instantiates a "common failure".
However, the necessary information is provided in RDF in accordance with the
role taxonomy, for which, let us suppose for the sake of the example, there is
no listed technique. The content therefore satisfies the success criterion due
to the correct implementation of an unlisted technique, using a technology
included in the baseline.

I think the best way of handling this is to say that once you move outside the
listed techniques and implement your own, you are responsible for defining
a test procedure to establish whether the success criterion has been met.
Ultimately, the success criterion is the determinant of conformance, and I
don't think the working group could write its techniques and failures in such
a way as to anticipate and handle any technique which an author might come up
with, given whatever technologies may be included in the baseline for the
content in question.

Received on Wednesday, 24 May 2006 11:43:08 UTC