- From: Lofton Henderson <lofton@rockynet.com>
- Date: Thu, 17 Jan 2002 09:08:54 -0700
- To: danield@w3.org
- Cc: www-qa-wg@w3.org
At 04:47 PM 1/17/02 +0100, Daniel Dardailler wrote: > > About #3. No new discussion drafts before telcon, but hopefully Friday. > >Should we talk about Ian's reorg of the TOC ? >(not sure where it was sent) Good idea. Ian's message only had limited circulation amongst us (chairs and editors). I have attached it below. Please have a look. For information: the editors (dd, KG, LH) had a teleconference last week, and we endorsed the overall outline and many of the details of Ian's proposal. Per the 1/3 telcon, email from LR, email from DD, we had been targeting a better implementation of Gd/Ck style. Ian made a great contribution in giving us the outline that actually showed the suggestions in practice. The next WG-only drafts (expected Friday) will largely reflect this. -Lofton. >Sender: ian@w3.org >Date: Thu, 10 Jan 2002 23:28:41 -0500 >From: "Ian B. Jacobs" <ij@w3.org> >Organization: W3C >X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.4.16 i686) >X-Accept-Language: en >To: danield@w3.org, karl@w3.org, lofton@rockynet.com >Cc: ij@w3.org, w3t-archive@w3.org >Subject: Comments on QA Framework: Processes and Operating Guidelines >X-RCPT-TO: <lofton@rockynet.com> > >Lofton, Daniel, Karl, > >Below are some comments on the 2 Jan 2002 draft of "QA Framework: >Process & Operational Guidelines": > http://www.w3.org/QA/WG/qaframe-ops.html > >I essentially ripped up and reorganized the document (a rough >outline at least). Please don't take this the wrong way: I often >do this as a way to get into a document and discover the issues >the author had to wrestle with. I recently did this with the XML >events draft as well. My comments may be a bit cryptic; don't >hesitate to ask me to explain them. I hope some of the comments >are useful, but feel free to ignore them. You can also forward >them to the QAWG if you wish. > >The document adopts the guidelines/checkpoints model, which I'm >happy to see. For each WAI document that adopted this model, the >WG wrestled with the question of the depth of the table of >contents. Each WG initially wanted a three-tiered table of >contents, with guidelines grouping checkpoints and topic headers >grouping related guidelines. In each case, the extra topic >groupings were dropped to produce a shallower TOC. > >In other words, the TOC went from: > > Topic 1 > G1 > checkpoint 1, checkpoint 2, > G2 ... > G3 ... > Topic 2 > G4 ... > G5 ... > ... > >To the shallower: > > G1 > checkpoint 1, checkpoint 2, > G2 > G3 > ... > >The primary reasons for this were: > > * The guidelines seem to suffice as an organization layer. > * People can't remember the extra layer of organization > anyway. > * It feels "cleaner" to have a straight list of guidelines, > numbered 1 - N. > >I thought of this as soon as I saw section 2.2.2.1 (four levels >deep) in the P&O Guidelines. I urge you to simplify the table of >contents by removing some grouping layers. I suspect there will >be resistance to this at first (as there was in each WAI WG), but >my experience is that shallower TOCs are more effective. > >So, below I: > > * suggest a simpler TOC > * suggest that some checkpoints are unnecessary > * reorganize the checkpoints > * use the WAI style of the imperative voice for > guidelines and checkpoints. > >========== >Sample TOC >========== > > 1. Introduction > 2. Guidelines > 1. In charters, describe QA goals, deliverables, resources > 2. Ensure that policies for QA work are clear. > 3. Promote clear communication with other groups > 4. Make test suites/tools part of specification development > 5. Plan for test suite maintenance > 3. Working with the QA Activity [Essentially the current 2.6] > 4. Conformance > 5. References > 6. Acknowledgments > [I moved Acks after References based on IETF RFC model > where the document progresses towards the most informative > (i.e., non-normative) information. This is a new practice > for me.] > >Note: Most of the rest of the prose in the document can be >inserted as notes after checkpoints or guideline introduction >material, or even subsections of "1. Introduction". > >=========== >Checkpoints >=========== > >It is my opinion that each checkpoint should be addressed to a >consistent audience within each Guideline. For instance, the >checkpoints in the first Guideline below (on charters) should >only be addressed to charter creators. Below I use the imperative >as in the WAI guidelines, but I have no problem with making the >subject explicit, as in "The charter shall...". > > ============================================== > 1. In charters, describe QA goals, deliverables, resources > ============================================== > > 1.1 Include a section on test suite/tool plans in the > charter. The plans should be at least a commitment > to "QA level four" (see range 1-11). > > Note: I don't think it's necessary to have a checkpoint > that existing charters should be updated. That requirement > only applies once, when the QA documents are approved > within W3C. > > 1.2 Ensure that each charter makes explicit resource > commitments to QA (e.g., for staffing of test > suite development and maintenance. > > Note: It's not necessary to say too much about > documenting resource commitments in the charter; > that's already required by the Process Document. > But it makes sense to say "Don't forget QA." > > Note: There were several staffing checkpoints in > the document. In my opinion, one checkpoint suffices, > even if the resources are for different types of > work. Just enumerate them, if necessary, in the > checkpoint. > > 1.3 Identify where and how conformance materials will > be produced. > > Note: The checkpoint about w3c providing a secure > and reliable repository can be deleted. > > ============================================== > 2. Ensure that policies for QA work are clear. > ============================================== > > 2.1 Ensure that W3C has ultimate responsibility for > evolution of test suites and tools. > > 2.2 Ensure a clear policy for branding materials > (e.g., conformance logos). > > ============================================== > 3. Promote clear communication with other groups > ============================================== > > 3.1 Identify the communication channel for appeals > of test validity. > > ============================================== > 4. Make test suites/tools part of specification development > ============================================== > > 4.1 Review technical guidelines before starting QA work > > 4.2 Synchronize specification writing and and development > of test materials early in the WD process. Keep > them in sync. > > 4.3 Publish a QA process document that explains: > a) communication mechanisms with the test suite managers, > b) contribution mechanisms, > c) review procedures for contributions, > d) disclaimers regarding certification > > > Note: Provide boilerplate text (informal) for the > disclaimer: passing all of the tests does not guarantee > full compliance of implementation to spec, and failing > the test suites means failing specific tests for specific > targeted features. > > Note: I don't think it's necessary to have a checkpoint > requiring the process document to point to the test > materials; this seems like an ordinary thing one would > do. > > Note: I don't think it's necessary to have a checkpoint > on usability of test materials for two reasons: > (a) people will make use of the most usable instance > available. > (b) unless you explain what makes the test suite > usable, the checkpoint won't be helpful. If you > have points for making a test suite usable, > add a guideline on test suite usability and > include specific checkpoints under it. > > > ============================================== > 5. Plan for test suite maintenance > ============================================== > > 5.1 Designate a task force to maintain the test suite > and maintenance procedures. > > 5.2 Support specification versioning and errata. > > 5.3 Ensure that test materials are documented, > self-contained, and reusable by other parties. > > 5.4 If the WG is to close, ensure that test suite > maintenance is handed off to another entity. > >============== >Other comments >============== > > 1) In section 2.2.1, the range of possible commitments would > be easier for me to compare if each additional point added > just one piece to the previous point. There seem to be > several axes: > > - WG composition > - What's in the spec > - Extent of test materials > > I could see organizing the levels like this: > > WG composition | In spec | Extent of materials > ------------------------------------------------------- > 1: Some for QA | | > 2: | Example | > 3: Some for QA | Example | > 4: .... > > In other words, it would be great if the levels built on each > other in a predictable and incremental way. The order in the > current rating 1 - 11 is hard to follow. > > Note: I think you can delete point 1 or rephrase it > since "promises to think about" is not verifiable. > > 2) Point 11 in the table includes the phrase "complete test > suite." It occurred to me that the I don't know what a > complete test suite is. It would be useful in the Intro > to make very clear: > > a) Objects. What is a test suite? Catalog? Etc. Perhaps these > items will be discussed in the glossary, but it will be > important to know what the deliverables are. > > b) Processes. The reader should have a better understanding > of the processes associated with the objects (and the roles > of people in these processes as well, e.g., "contributor", > "moderator"). This will help people understand the > associated requirements. > > I'm thinking of the diagram for the XSLT test suite, that > shows where information comes from, who manages it, when > approval takes place, etc. Part of this might belong in > another document more about test suites and less about WG > processes. But sections 2.4.5.1 and 2.4.5.4 are about > contribution and review processes. > > In short, I think the reader will have an easier time > if the Intro explains in broad strokes how the pieces > fit together. > > >-- >Ian Jacobs (ij@w3.org) http://www.w3.org/People/Jacobs >Tel: +1 718 260-9447
Received on Thursday, 17 January 2002 11:08:39 UTC