- From: Lofton Henderson <lofton@rockynet.com>
- Date: Fri, 09 May 2003 08:46:27 -0600
- To: Mark Skall <mark.skall@nist.gov>, Alex Rousskov <rousskov@measurement-factory.com>
- Cc: www-qa@w3.org
At 09:56 AM 5/9/2003 -0400, Mark Skall wrote: >At 12:17 AM 5/8/2003 -0600, Lofton Henderson wrote: > >>At 08:54 AM 5/7/2003 -0600, Alex Rousskov wrote: >> >> >>>On Wed, 7 May 2003, Lofton Henderson wrote: >>> >>> > CP1.3: Commit to complete test materials. [Priority 3] Conformance >>> > Requirements: the WG MUST commit to produce or adopt a complete Test >>> > Materials before Recommendation, where complete is defined as: at >>> > least one test case for every identifiable conformance requirement >>> > of the specification. >>> >>>I apologize if I missed the discussion about it, but not all >>>"identifiable conformance requirements" are testable and, hence, can >>>have at least one test case. Should "testable" qualifier be added? >> >>I don't have a problem adding such a qualifier. Objections anyone? > >I object. The reason is that I don't accept Alex's premise. Every >requirement should (MUST) be testable. To clarify, are you saying: 1.) that "testable" is part of the our definition of conformance requirement (or should be), and therefore this is inherently redundant? 2.) or, that we should have (e.g., in SpecGL) a requirement (i.e., checkpoint) that every conformance requirement in a target spec must be testable? Note. "Conformance requirement" is not in QA-glossary, nor in SpecGL "4. Definitions". Note2. I recall a while back that we discussed the difference between conformance requirement and test assertion in telecon, and someone (Andrew?) drew a distinction that we liked, and Mark did an AI to take a sample GL and show what its *test assertions* would look like, as opposed to its conformance requirements. >(In fact, I thought this statement was included somewhere in our guidelines) Not yet. Or if we have addressed it, I can't remember where it is, off the top of my head. We have talked a lot about "test assertions", but little about "requirements". There is a SpecGL issue group about TA-req. As I recall, at least one LC commenter noted in passing the lack of definition of "conformance requirement". (Or maybe I did -- but if you look at the issue group, then you'll see that we need to agree about it before resolving the questions about test assertions.) > If a requirement is not testable, it should be reworded to be testable > or be eliminated from the specification. If it can't be tested, it can't > be verified that it was done correctly and is, thus, of no use. Adding > the suggested qualifier would sanction having non-testable requirements. Since there is dissent on this, I would propose that we close it in the OpsGL context for now, *without* the suggested change. We will sort it out with the SpecGL TA-req issues group, and if that implies any adjustment to OpsGL is appropriate, then we will do it. (I'll link this thread from that issue group, to make sure we don't overlook it. So ... continue ...) Okay? -Lofton.
Received on Friday, 9 May 2003 10:52:38 UTC