Re: [DRAFT] 2002-05-30 QA WG Teleconference minutes

Karl,

Thanks for the minutes.  I have three comments, in-line below...

At 11:09 AM 5/31/02 -0400, Karl Dubost wrote:

>[...]
>- Issue 66:  Glossary dated versions
>
>LH: I didn't hear any strong objections to that. Any discussion.
>
>MS: We have a dated version, so we can point on it, until we have a 
>process to modify it.
>
>LH: I didn't think to the process. [Lofton gives an example with spec 
>guidelines.] Ex: The spec guideline points to the definition in the 
>glossary that is not in the TR space. If we change the glossary and at the 
>same we can't change also the specs in the TR space, there's a possibility 
>of differences between the meant definition and the definition of the 
>glossary. How frequently we update the glossary? If definitions are 
>contentious, we raise an issue and update it. But each version will be 
>dated. Does it answer your question MS?
>
>MS: Yes it answers my questions.
>
>DH: ??? (missed it Dom)
>
>LH: How to devise the dated version scheme.
>
>KD: not only the QA framework is concerned, it's why there's a need for a 
>dated version. Because a spec in another group can also point on the glossary.
>
>LH: dated version in the WG space.
>
>KD: same kind of dated document than spec or notes.
>
>LH: agreed

I think the conclusion was a new Action Item for Karl&Mark 
(A-2002-05-30-3), to make a proposal for dated version scheme, any needed 
sync of the /QA/ dated versions with /TR/ publications, any constraints on 
frequency of updating /QA/ dated versions, etc.

>[...]
>- #61 [Spec]:  standard terminology for classes of products
>
>LH: how to ensure that the specification use our terminology and taxonomy 
>for classes and products.
>
>DM: Taxonomy is complex and depend on the specifications? a process, an 
>api, a format.
>
>LH: leave the resolution until the proposal.
>
>DM: how much abstract we want to be? If we want to adopt a general 
>consumers/producers point of view or to precise on particular ones: So if 
>we go to particular we may need profiles and flavors. Do we want to grant 
>a certain level of flavors and invite people to have different 
>possibilities if they have particular cases.
>
>LH: SVG have different between Viewers and interpreters. That's the same 
>for MathML. We have already take a step in the direction of 
>differentiation. Should we go back or go further?
>
>DM: We should ask to www-qa list to see if people think it's good or not.
>
>LH: closed it and wait for comments. Put it online on ML

As I recall, we decided: postpone it until Originator (MS) is available 
(f2f);  and (I think), add to the "Proposal:" section of the issue 
something like, "Yes, require standard terminology, by adding 
lower-priority checkpoint for it."



>- #60 [Ops]:  "In the charter," in ckpt statements.
>
>LH: Some of our checkpoints it seems to have only influence on the charter 
>and people writing charters and not specs. I propose to leave it as it 
>until Kyrill be able to discuss it.
>
>- #69/70 [Spec]:  flavors of conformance (see DM message at [4])
>                 (clarify DM proposal, not necessarily resolve)
>
>LH: David sent a suggestion in response. not sure to have understood.
>
>DM: See the mail. Can we have strict conformance, not so strict, 
>discretionnary conformance, etc. My suggestions: we expose all of that to 
>a wider audience.
>
>LH: Dan Connolly was the origin of the discussion: "Flavors are ennemy of 
>interoperability". I'm not sure to understand the guideline 3 or the 
>interpretation of it.
>
>KD: there are two issues: Having flavors of conformance AND the way you 
>declare your conformance clause.
>
>LH: experience Sandra? about Guideline 3
>
>SM: Not sure about the precise meaning of the documents
>
>LH: David?
>
>DM: There's a kind of philsophical approach
>
>DH: having a practical approach, looking at the spec who had success with 
>the interoperability
>
>Sandra: Flavors is not defined.
>
>LH: I have difficulty to understand. Ask to Mark/Sandra to discuss and try 
>to clarify the guideline 3. Other interesting thing about your email. We 
>should have a table called dimension of variability.
>         Ref to a mail?

We didn't call this an Action Item at the time, but could we do that, just 
so that it doesn't get lost?  (If that's okay, then:  A-2002-05-30-4, and 
let's have a due date of 2002-06-13, the f2f.)

-Lofton.

Received on Friday, 31 May 2002 17:11:29 UTC