Minutes QA WG 2002-08-07

[Sorry to send them late, I was on vacations]

QA Working Group Teleconference
Wednesday, 7-August-2002
Scribe: Dominique Hazael-Massieux

(KD) Karl Dubost (W3C, WG co-chair)
(DH) Dominique Hazael-Massieux (W3C - Webmaster)
(LH) Lofton Henderson (CGMO - WG co-chair)
(DM) David Marston (IBM)
(SM) Sandra Martinez (NIST)
(JM) Jack Morrison (Sun)
(LR) Lynne Rosenthal (NIST - IG co-chair)
(MS) Mark Skall (NIST)
(AT) Andrew Thackrah (Open Group)

(dd) Dimitris Dimitriadis (Ontologicon)
(KG) Kirill Gavrylyuk (Microsoft)

(PF) Peter Fawcett (RealNetworks)

Summary of New Action Items: 
- AI-20020807-1 DH to add a reminder in the review matrix [1] about
public vs member materials
- AI-20020807-2  DM to send a proposal to where the profiles should fit
in the 1 to 7 list GL 2 of spec GL [2]
- AI-20020807-3 KD to begin a liaison with the TAG to see who owns the
issue of architectural division of a spec (profile/modules/levels)
- AI-20020807-4 KD to ask for publications in TR space in 3 weeks

1. http://www.w3.org/QA/Group/2002/06/reviews

Agenda: http://lists.w3.org/Archives/Public/www-qa-wg/2002Aug/0048.html
Previous Telcon Minutes:

1. Roll Call
See above

2. Logistical Topics
* F2F Meeting in Tokyo
Details given at http://www.w3.org/QA/2002/10/f2f . No new informations.
You can book a room in the hotel indicated by Olivier by fax (follow the
link in the forementionned page). MS wonders if other hotels are
available in the area in case this one would be full. Olivier Thereaux
(on IRC) indicates that there are other hotels, but not necessarily as
cheap as this one since it's a rather expensive area.

* Public/private WG materials
The review matrix is kept in Member only space [1].
LH asks to not send finished review to the WG list which is public,
since the content of the review is not endorsed by the WG as a whole,
and could contain contentious data.
DH takes the action item (AI-20020807-1) to put a warning on the page.

1. http://www.w3.org/QA/Group/2002/06/reviews

* IG/WG mailing list uage (aka www-qa@w3.org vs www-qa-wg@w3.org)
KD sent a mail [3] reminding that technical discussions should happen on
the IG mailing list to get a broader audience, and drawing attention on
the good way to reply to a mail on the WG list. LH shares this opinion,
and DH points that he would like to see the WG list used only for
logistical purposes.
LH asks if even low level details email should go to the IG list. KD
points that low level details can quickly lead to a larger issue on
which the IG would be interested and having this kind of mail on the IG
list could motivate people to read the documents.
Hence, the WG resolves to move technical discussions as much as possible
to the IG list.

3. http://lists.w3.org/Archives/Public/www-qa-wg/2002Aug/0046.html

* "Week in QA" [4] Periodicity
The last edition of "Week in QA" was on July 15 [5], which means that
there should have been a new edition at the beginning of the month for
the "2 weeks" period. Provided that there was no much activity during
this period and that it would be already one week late, it's decided
that the next one will a "4 weeks" period, written by KD.

4. http://www.w3.org/QA/weekinqa
5. http://www.w3.org/QA/2002/07/weekinqa-20020715

3. Spec GL [6] Discussion
LH would like to have after next week special teleconf in a state good
enough to request its publication and ask for public feedback on the new
direction of the specification. 

6. http://www.w3.org/QA/WG/2002/08/qaframe-spec-0804.html

People main concern during last week teleconference was around GL 3, 4
and 7 namely on the confusion between profiles, modules and levels [7].
He tried to avoid suggesting that one relationship was the good way or
another one was the bad way. He did the same thing in GL 3, 4 and 7. But
in GL 7, he just pointed back to GL 3 and 4 since they were already
addressed in these GL. He is now seeking endorsement of this approach
DH thinks that we should really indicate if there is a good and a bad
way to do that but that's probably not in the scope of the upcoming
publication. MS notes that something we may decide we don't want to
address because of the big number of cases. Public feedback on this
issue seems to be in order.

DH suggests pointing to open issues in the SOTD, marking them up
differently in the spec so they get more visible and finally listing
them in the announcement of the newly published doc to the IG list.
Consensus on this approach

- open issues get more visibility through this
- issue on good and bad relationships between dimensions of variability
marked as open in the next published draft.

7. http://lists.w3.org/Archives/Public/www-qa-wg/2002Aug/0029.html

[Jack left during this discussion]

Going on into the specifics of GL 3, 4 and 7:
Nothing special in the general verbiage of GL 3.
* In the CP lists, 3.4
<<Minimal requirements on what object(s): a class of products? I.e.,
does this mean, "For each profile, define the minimal required
features/support for each class of product"? Or does it mean to define
some profiles-wide minima that are minima for the whole concept of
profiles (in which case, it sounds like "rules for profiles">>
AT opts for the pro-profile interpretation, wondering if another CP
would be necessary for the profile-wide interpretation.
LH cites the case of XHTML modularization as an example of this
profile-wide requirement. MS and DM sees this more as a module type of
issue. LH sums up: there should be a distinction in the concept and that
3.4 should be understood as pro-profile.
DH points that the case of XHTML modularization seems a side case, where
profiles are the object of the conformance clause, not the subject. LH
asks if there should be another class of products for this case
(profile). DH points that not having specification in this list makes
our own GL not usable with this list.DM thinks that category 1
(foundation or abstract) would fit for it.
KD is worried by the weight of the past and wondered if the TAG [8]
shouldn't have a look at that to impose consistency on this. DH thinks
that this would be the role of our WG, but KD explains that there a more
architectural question behind it.

8. http://www.w3.org/2001/tag/

=> AI-20020807-2  DM to send a proposal to where the profiles should fit
in the 1 to 7 list GL 2 of spec GL [2]
=> AI-20020807-3 KD to begin a liaison with the TAG to see who owns the
issue of architectural division of a spec (profile/modules/levels)

The discussion leads to CP 2.3
http://www.w3.org/QA/WG/2002/08/qaframe-spec-0804.html#Ck-minimal-requirements, with the following marked issue:
<<So what? This "core" is not a useful concept for conformance of
classes of products, in a profile environment, is it? From a UA
conformance perspective for example, what use is it to know that
profiles A, B, C all have subset D in common? Okay, it may be an
interesting fact, but it tells you nothing that you need to know to
verify conformance of the A-conforming UA or the B-conforming UA or the
C-conforming UA.>>
Agreement that this place should be where to speak about cross-profile
min req.
Back to GL 3, this resolves the questions pointed by LH.

Going on GL 4:
In the general verbiage, LH has a couple of specific questions:
- examples of levels designed in the 1st edition
(most commonly vs always)
=> most commonly (DOM example given in the ml:
- examples of level 2 defined by a bunch of new modules
DH suggests dropping this theoretical-only principle. DM suggests
discouraging the approach of building level by modules, since it would
lead to confusion between level and profiles: modules are not a just a
packaging division, they should have a logical foundation.
DH formulates how this kind of grouping modules just look like profiles,
and would be a bad usage of levels.
Resolved: LH will flag this issue in the next published version

LH has a minor question in 4.3: should this be added "The conditions or
constraints normally will be tailored according to class of product"?

Going on GL 7:
- <<By definition, level 1 is the minimal requirement with respect to
the levels dimension of variability. (@@min reqt on what? i.e., what
class, what product?)>>
DM thinks that this apply on the scope of what the levels are actually
dividing. After some discussions of the various cases of interaction
between level and profiles, it is proposed and agreed to drop the
sentence, moving the discussion to the email list.

CP 7.1
(http://www.w3.org/QA/WG/2002/08/qaframe-spec-0804.html#Ck-choose-level-grouping) has reworded in response of KD's comment
<<@@Ed Note. Should we also say, "N/A for specifications which predate
these guidelines?" I think not: the same could be said about *any*
checkpoint.>> gets agreement from the WG.

GL 10: Discussion about terminology about "conformance level"
DH and LH suggests using "degrees" and referring to the historical use
of the word "level" for this pattern. DM notes that this affects our own
conformance clause since we're copying the WAI usage of "Level A"
LH indicates that agreement on this topic means changing our conformance
clause to "Degree A conformance". MS suggests "1st degree" instead. LH
notes that there is a risk of a push back from WAI because of the
ambiguity that level A is the minimal whereas 1st degree looks like the
best. DM doesn't find any conflicting usage of the word degree doing a
search on W3C site, except "degree of confidentiality". MS agrees that
moving that way seems clearly justified

Resolved: Moving to 1st/2nd/3rd degree in our conformance clause and
adapting the verbiage of the relevant CP but reminding the example of
usage of the word "level" by WAI and not discouraging it/

Special telecon next wednesday on spec G
Adgenda: remainder of today's agenda and priority of CP list

Dominique HazaŽl-Massieux - http://www.w3.org/People/Dom/

Received on Monday, 19 August 2002 10:40:59 UTC