Re: Open issues on SpecGL (was: WG SpecGL updated)

I'm catching up on email after my vacation. As best I can tell, the
following are still open.

DHM>- on CP 12.2  "Require the ICS be completed as part of the
DHM>conformance claim. "
DHM>On whom/what would such a requirement be set? What conformance would
DHM>it break not to complete the ICS?
DHM>http://www.w3.org/QA/WG/2002/10/qaframe-spec.html#Ck-require-ICS

This isn't about conformance, it's about claims of conformance. When
the spec has any variability, we shouldn't accept claims of the form:
"Product Q conforms to XBlah 1.0 from the W3C."
Rather, the claim should identify all the implementer choices made.
Which class of product, as enumerated in the spec, does this product
implement? Which profile does it implement? Which modules? Which
degree of conformance does it attain? Does it also support deprecated
features? What level of the spec does it implement? What choices were
made on the discretionary items? Are there extensions, and what are
their names? Answers to all those questions will affect tailoring of
the test suite to this product. (See issues #95 and #96.)

Regardless of the resolution of issue #50, an open-systems approach
requires that conformance, and claims thereof, be verifiable by an
independent testing lab. That lab should be able to gather up all the
normative specs (may vary by class of product, profile, etc.), the
product and its ICS, and a test suite, then run the tests that
pertain to the choices made by the implementer.

(Fine point: the reason for asking for the names of the extensions
is to avoid inadvertant name clashes. A test case might have a
"junk" element that turns out to be an extension element for a
particular product!)
.................David Marston

Received on Wednesday, 23 October 2002 23:32:23 UTC