Final Minutes of QA WG Teleconference, 2003-1-13

QA Working Group Teleconference
Monday, 13-January-2003
--
Scribe: Peter Fawcett

Attendees:
(dd) Dimitris Dimitriadis (Ontologicon)
(KD) Karl Dubost (W3C, WG co-chair)
(PF) Peter Fawcett (RealNetworks)
(KG) Kirill Gavrylyuk (Microsoft)
(LH) Lofton Henderson (CGMO - WG co-chair)
(SM) Sandra Martinez (NIST)
(LR) Lynne Rosenthal (NIST - IG co-chair)
(MS) Mark Skall (NIST)
(AT) Andrew Thackrah (Open Group)

Regrets:
(DH) Dominique Hazael-Massieux (W3C - Webmaster)

Summary of New Action Items:

AI-2003-1-13-1: Karl to draft new language for checkpoint 9.4 by next telcon.

Agenda: http://lists.w3.org/Archives/Public/www-qa-wg/2003Jan/0038.html

Previous Face to Face Minutes: http://www.w3.org/QA/2003/01/f2f-minutes

Minutes:

1) Roll Call

2) Miscellaneous topics

- PF will take over lead editorship since Kirill is going to be too busy.

- reschedule monday meetings to previous wed slot. (10am eastern)
Set for Wed 22 if Dom is available.

3) OpsGL [1] last call details

- In order to pass SpecGL we need to have usage scenarios to go to last call.
Do we have time?
If there is an interest and time, we  should do it but if there is no time
we will just have to fail that part when we go to last call.

- Commitment table:
Lofton's Action Items - 3 of them, One having to do with language one 
to do with
structure and one to do with nesting of levels.
After looking at it Lofton has decided that two of the issues aren't 
really issues
but that the language needed to be clarified.

Did the addition of the requirements language help or make it worse?
PF it made it better.

Go over editorial notes.
Row 6 - we say complete but "complete" is not necessarily possible.
Mark, require testing all MUST statements.
Must be a test associated to all requirements.
Andrew, but there may be a requirement that is not testable.
Mark, but specs should make requirements that are testable.
Lofton proposed going forward with out mention of the untenability of 
requirements
and wait for it to be raised against the last call text (by Andrew or any one
else).
Agreed.

Lofton, is there any reason not to change test suite to test materials.
No, change it.

Row 1 - Lofton, what time line do we want to tie this to?
Lynne, yes but having an ICS is only a priority level 3.
Lofton, we want them to do the review, and an ICS is a nice way to do that,
we could require reviews and suggest that an ICS is a way to do this.
Tie requirement to milestones

- CP3.2 research (AI-20030107-4)
Still to do, has not been done yet.

4) SpecGL [2] last call details

- Loftons reworking of guideline 13 according to what was decided in 
last face to face.
Made one deviation from that decision which was to say "Clearly 
identify conformance requirements."
rather than "Clearly identify conformance criteria."
Language is fine.

- Alex's comments;
4 comments where he asserts that they are untestable.

- #9, guideline 2 - he asserts that we do not need this guideline.
Do we need it?
Karl, it might be hard to conform to this guideline in some cases.
Some have done it well but others have not done it at all.
Kirill would like to keep checkpoint 2.1 as it was useful in the development of
SOAP.
Lofton thinks it is testable, there is a finite number of classes of products.
Also agrees that 2.1 is useful.
Karl, Keep the parts having to do with conformance. Drop Guideline 2 
and move what has
to do with conformance to another related guideline.
Lynne, yes we could do that and there are good places for the 
conformance related stuff
but the explanation part of Guideline 2 is important and should remain there.
Lofton agrees with Lynne's observation that things shouldn't move 
this close to last call.

We have reaffirmed that we want to keep guideline 2 and that we want 
to keep the checkpoints
basically as they are and Lynne will try to draft a rational for 2.1.

- #24-26, guideline 9
- #24, Alex has a problem with the example part of of the guideline,
that it is wishful thinking.
Mark thinks it can be removed. Lynne agrees, other language has 
changed and the point it
was trying to make isn't really there any more.
Lofton suggests adding verbiage from checkpoint 9.5 since that is 
what it is trying to
justify. But doesn't feel strongly.
Decision is to remove it. Agreed.
- #25, Checkpoint 9.4. Yes all agree that this is a problem,
wouldn't have a clue to know if it's been passed or not.
Karl, What we mean to say is that it is a task of the WG to define a
uniform definition or mechanism of doing extensions.
For example CSS defines an explicit way to define an extension.

Action item to Karl to draft new verbiage for this and send to WG 
before next telcon.

- #26, Alex thinks it's too narrow, Lynne sees how it could only 
apply to producers.
Mark, thinks we should keep it, it is specifically narrow and it's priority 3.
Lofton, there are two issues here, one is that the scope it two 
narrow, the other
is a dispute of the notion that extensions are bad for conformance.
Having a mode for publishing conforming documents is the important 
part of this.
and yes we do feel that extensions can be bad for conformance. This should
be reworded a little perhaps to make it less narrow but it should be kept.



5) Adjourn

Received on Monday, 27 January 2003 18:46:14 UTC