Re: QAF Ops Guidelines review action [LONG]

jeremy and evan:
>> Doing that a hundred times over would have made the document unreadable, 

> unionOf is only one of many parts
> of OWL
 what I was getting at is that much of the normative content of S&AS is 
expressed in the terse mathematical way in which unionOf is defined - for 
every row of every table we should do the same, if we were to follow in 
detail the Specification guidelines (which we were not specifically 
reviewing, but it is hard to review a compound spec in parts). (It is unclear 
how one would address the mapping rules and the abstract syntax rules)

So the specific checkpoint I had in mind was:
"The specification MUST provide a normative list of test assertions. "

The discussion of this is weak, in particular the link trail from the CR 
document peters out with an under construction before one finds actual 
examples of what this is intended to mean.

I understand that they mean "Tests can point directly to the test assertion," 
i.e. every atomic test should have a testable assertion that it corresponds 
to, and that ideally this testable assertion should use RFC 2119 language.

I am pretty clear that there are more than 100 atomic tests concerning the 
maths in S&AS and so the "hundred" is probably not an exageration.

There are 238 "<tr" in S&AS, I believe most would need attention.
(I note that many have anchors which is a recent? improvement from the point 
of view of being able to refer to them)


The links petering out suggests that a different quality goal is that 
multi-document recommendations should be published as a whole at each 
maturity level - don't they have that already:
Checkpoint 3.1. Synchronize the publication of QA deliverables and the 
specification's drafts. [Priority 2]

they may have played by the letter of this, but certainly not by the spirit 
with links off to unfinished work.


Received on Wednesday, 17 December 2003 08:50:01 UTC