Non-testable, normative requirements (was RE: tone as a guildeline?)

Gregg Vanderheiden writes:
 > 
 > As to being normative but not checkpoints. I'm not sure what that
 > means. It is an interesting thought. At first it seemed like
 > everything normative needed to be checkable -- but thinking further
 > it would seem that only those things required for conformance would
 > need to be testable checkpoints. Maybe we can have normative
 > suggestions that are not required.

In doing this we would (de facto) be > defining a "core" of testable
requirements plus a set of additional > requirements to which entities
could claim conformance, but for > which adequate testing could not be
carried out. In the latter case > we would be essentially treating the
author's declaration as final > (a third party couldn't conduct some
tests and show that the > author's conformance claim was correct, or
indeed erroneous).

I am not objecting to the idea at all, merely pointing out its
implications. If we want those consequences then we can go ahead.

Received on Tuesday, 11 December 2001 03:20:43 UTC