RE: [www-qa] Re: Conformance and Implementations

At 04:05 PM 10/23/01 -0600, Alex Rousskov wrote:
>On Tue, 23 Oct 2001, Mark Skall wrote:
>[...]
> > XML is a compromise.  It's not as precise as these languages but
> > much easier to use and less costly to develop specs using it.
>
>Dry, spec-oriented English is a compromise. It's not as precise as
>these languages but much easier to use and less costly to develop
>specs using it.
>
>In fact, XML per se is almost as informal as constrained English so
>the benefits of using it instead of English must be minor!

This encapsulates one aspect of this dialog that I have been thinking 
about.  The benefit that you can reasonably obtain by authoring the 
standard (REC) spec in XML might widely vary, depending on the type of 
standard.

The fact that an API spec like DOM uses IDL to define the DOM interfaces, 
and "IDL Definitions (definitions)" has been integrated into [1], brings 
with it tremendous benefit.  The test suite can be (is) generated by 
transformations on the spec's XML source.

On the other hand, a spec like SVG has lots of testable assertions that 
(verbosely) define how given pieces of the SVG content are rendered into 
visual, graphical effect.  It would an enormous (and not necessarily 
desirable) undertaking to write SVG otherwise.  In this case, the most you 
might be able to achieve with XML is:

<testAssertion id="this-particular-assertion's-id">
...some rendering specification... (English or another human language)...
</testAssertion>

The content, "...some rendering specification..." still suffers from the 
imprecision of human language.

Having said that, the above still would bring considerable value -- the 
ability to automatically extract the (possibly imprecise) test assertions 
for various test-suite-related activity, and the ability to generate 
precise references from outside the spec to the test assertions within the 
spec.

-Lofton.

[1] http://www.w3.org/XML/1998/06/xmlspec-report.htm

Received on Tuesday, 23 October 2001 20:25:09 UTC