Draft Minutes - QA Working Group Teleconference 2003-06-23

QA Working Group Teleconference
Monday, 18-November-2002
Scribe: Peter Fawcett

(KD) Karl Dubost (W3C, WG co-chair)
(PF) Peter Fawcett (RealNetworks)
(LH) Lofton Henderson (CGMO - WG co-chair)
(LR) Lynne Rosenthal (NIST - IG co-chair)
(AT) Andrew Thackrah (Open Group)

Visiting: Dave Marston (MIT)

(DH) Dominique Hazael-Massieux (W3C)
(KG) Kirill Gavrylyuk (Microsoft)
(MS) Mark Skall (NIST)

(PC) Patrick Curran (Sun Microsystems)
(dd) Dimitris Dimitriadis (Ontologicon)
(SM) Sandra Martinez (NIST)

Summary of New Action Items:
AI-20030623-1  Lofton:	To write up a few sentences for the 
conformance clause (section 3) to address the dov of dov issue with 
8.1. - Monday (June 30).

Agenda: http://lists.w3.org/Archives/Public/www-qa-wg/2003Jun/0061.html
Previous Telcon Minutes: [...replace w/ correct link before circulating...]
[Optionally, additionally: previous f2f minutes if one immediately 
preceded telcon]


1.) roll call 11am EDT, membership

2.) Any routine business

- any new Last Call reviews?
No. No reviews to assign.
- is anyone doing these 2 parts from Xquery/XSL WGs? [8]
Dave is looking at Functions and Operators.
- overdue Action Items (beat the rush, look now & email Olivier!)
Skipping today, haven't been reviewed yet.
But will become a regular thing to get the list cleared up.

3.) approve MathML's LC review reply?
- default 'yes'.   [2a], [2b]
Approve the MathML LC review reply.
Any objections. No.

4.) SpecGL [1], LC-issues [3]
- see Crete raw minutes for recent stuff, current state
- LC-29.4 loose end [9]
Lynne has had a comment on the LC29.4. It had been discussed at Crete 
but it is unclear what was decided.

The comment that is being addressed (see ref 9) has to do with 
checkpoints being too broad or are overlapping.

In principal Lynne agrees with what he is saying and that we are 
trying to apply them in the documents.

It applies to the whole document, not to any one checkpoint.

Lofton says we came up with definition of Conformance Requirements 
which should end up in text.

Lynne doesn't remember definition but yes, if it exits it should go 
in the  document.

Lofton is reading it as a suggestion that this is guidance both to us 
and for us to pass on through our guidelines.

Put in verbiage some where, extec or here.

- approve SpecGL Use Cases [7]
Dom sent draft and comments have been made by Lofton and others.

With in SpecGL we require use cases.

So we will include use cases in ExTech. If any one has use cases to 
include please use the same format and submit them.

Any comments on the ones we have so far? (See ref 7).

Andrew, yes they are fine so far.

Lofton, most people seem to have a separate document for use cases 
but since we already have this existing ExTec document it seems to be 
a good place to put them. We can include a link to the ExTec as use 

It's open for people to write up more use cases if they feel that 
they are needed.

- Finish SpecGL-by-SpecGL [5]
Mark did a review of SpecGL by SpecGL.
Covered part in Crete. Believe that it was covered up to GL8

8.1 We don't conform.
We do have levels (A, AA, AAA) but we don't have modules or profiles.

There is discretionary choices with in a GuideLine in that the
implementer must make a choice as to wether they will or will not  do 
profiles/modules/levels. Since it is a choice it is discretionary but 
it is not labeled as being discretionary.

Dave is uncomfortable with having decisions made on DOV as being discretionary.

It's a Meta discretion.

Dave, there are some other GLs that are more discretionary than these 
DOV, say the GL about TOC having easy way to find conformance 

Lofton can see how we could pass this requirement by making a comment 
with in section 3 (the conformance clause) about how there is some 
allowances for implementors to make decisions while trying to avoid 
getting into the meta notion which would be very confusing.

Minutes from crete indicate that mark will rewrite 8.1

Lofton will take an action item (Monday June 30) To write up  a few 
sentences in the conformance clause (section 3) to address this issue.

8.2-8.5 are not applicable.

9.1 -9.3 - Conform

9.4 - We don't want to limit people from doing things but we don't 
allow for 'extensions'.
The conformance requirements can be thought of as part of a uniform mechanism.
There are other part's that can also be thought of as part of a 
uniform mechanism.

9.5 - Mark says we don't conform, but Lynne disagrees.
Those who 'extend' specGL in their own specifications, will extend 
specGL  and will publish it in it's own specification (implementation)

9.6 - Provide mode to make conforment requirements.

We allow 'extensions' that improve the interoperability (say do an 
even better job of linking between requirements and use cases than we 
require) but not extensions that are less conforment.

Lynne -  this doesn't apply in our case really anyway as we are a
consumer, not a producer of our spec.

This GL only applies to producers, not to consumers. Place language 
in Conformance Requirement that this only applies to producers, 
rather than just in descriptive language.

10.1-11.1 conform. (but)

10.2 - Do we need to make a clear distinction between the conformance
requirements for different documents.
Each document is implementable with out being constrained or having 
requirements form other documents.
They are independently implementable.

11.2 - We don't conform. There is requirements about what language to 
include and he's right. It's an editorial thing that should be 
cleaned up.

13.3 - We are not consistent in many different places and this is an 
editorial thing as well. Should be cleaned up by editors.

13.3 - the wording of this is bad and would be hard to test.
What does "analogous wording to express analogous provisions" really mean.

It comes from one of the w3c style documents.
No one objected to this language in the Last Call Comments.

14.1 - Mark has already agreed to do the wording for Test Assertions

14.2 - Once that's done the mapping can be provided.

Synopsis of current plan from Crete.
- Move TestGL (?)  to Note Status
- Move OpsGL and SpecGL to a state suitable for CR.
- Will sit in CR State till next Tech Plenary.
- Try to get WG's to fill out checklists. Don't need to conform but 
we will use the feedback from their filling them out.
- Minutes are not clear as to decision to go to Rec or not. Lofton 
thought that there was some leaning as to not finally go to Rec.
- Progress TestGL during this time while Ops and Spec are sitting in CR status.

Received on Tuesday, 24 June 2003 23:04:21 UTC