Re: RDF Core test driven development and QA Test Doc

[Dropping RDF Core from cc]


I had noticed that QAWG is engaging more with the Semantic Web activity 
work ... that's good.

One of my procedural concerns is to do with the tensions between the 
following two paras:


> We have no intention to "prohibit and/or obstruct" what you are doing.  
> In my view, what you are doing is outside of the scope of the normative 
> content of TestGL 1.0.  (This is my view, and I'd invite other QAWG 
> participants to comment.)
> 
> Regards,
> -Lofton.
> 
> P.S.  We are pushing to publish again soon, but that effort has been 
> diluted in the last few months by the effort around moving other parts 
> of QAF to CR.  Your message indicates that we should prioritize 
> publishing an updated WD.
> 
> [1] http://www.w3.org/QA/WG/#docs
> 
> 

Having gone to CR on both the Ops GL and the Specs GL you have expressed 
the opinion that these do not need to change as a result of any comments on 
the still maturing Test GL. This seems to be systematically prejudicial.

For instance, it appears that you have recently become convinced that not 
all test suites should be conformant; but to change the Ops GL to not 
normatively require a commitment to at least A conformance with the Test GL 
will require going back to last call.

Essentially the move to CR says that the docs are mature and the design 
won't change - but you are still considering some of the basic input into 
the design (the test experience of W3C WGs). As it becomes clear what the 
test GL should and should not say, then the other docs might need 
substantive change.

Even after finishing the test GL, I believe that freezing the design before 
  completing other informative deliverables that you have committed to, 
such as the Cost Benefit Analysis promised to the DIWG, may be a mistake.


It may transpire that the CBA shows that the most controversial ten tests 
in any W3C test suite deliver 95% of the value in resolving 
interoperability problems that arise from the spec itself (rather than 
implementors understanding what they have to do and not doing it). This may 
well suggest substantive change throughout the QAF to try and encourage the 
creation of very small but highly contentious test suites.

cf Graham Klyne's msg, which I forwarded to the QAIG

Taking a complete QAF (normative and informative parts) to Last Call as a 
whole seems the most straightforward and honest approach. It would also 
allow you to have a single mailshot to all your pre-last call reviewers 
asking them to check that their comments have been adequately addressed in 
the completed QAF. It still leaves you with the problem of being more 
visibly disciplined about tracking comments on your documents, particularly 
after deciding that they are stable.


Jeremy

[Having dropped RDF Core from the CC list should I have moved this to the 
www-qa list? This is now meant merely as a peronal message to the QAWG, it 
is not a comment on any document]

Received on Sunday, 4 January 2004 18:26:41 UTC