Re: Comments on Last Call WD QA Framework: Specification Guidelines

I feel a need to respond to a couple of Marc Hadley's comments.

>Guideline 5 - Modules are non-hierarchical - can modules have
>dependencies on other modules ? If so, isn't this a hierarchy ?

There can be tricky relationships. For example, if the "Core Module"
is required of all implementations, and other modules are optional
but independent of one another, the spec may not address whether
the optional ones can depend on the core, but in practical terms,
they can. Also, consider a module whose rule is: when this module is
implemented, at least one of Module A or Module B must also be
implemented. That's not a hierarchy as contemplated in the "levels"
guideline. Then again, do we care if modules have a relationship
that could be called "hierarchical"? Not really, so the GL should
say modules are not *necessarily* hierarchical. (I'd be happy to do
away with levels and call them modules instead, but there are those
who advocate a separate designation to keep the word "level" magic.)

>Checkpoint 7.1 - conformance requirements imply a single section for
>deprecated features - is it not OK to include deprecations where they
>occur without a summary section ?

No, it's not! The Conformance Clause should be a single point of entry
into the conformance requirements, including all Dimensions of
Variability. Same for discretionary items. BTW, this requirement helps
the WG avoid contradictory provisions.
.................David Marston

Received on Monday, 10 March 2003 19:23:05 UTC