Doyle: Could be more people from Wells Fargo participating in WAI.
Shawn: Status on meeting in Canada?
Judy: Waiting for confirmation on room block at hotel.
Judy: Status is pretty certain, but maybe wait until early next week to reserve
tickets.
Judy: Why are people not posting comments to list?
Alan: Wanted to keep the noise level down.
Judy: Want noise level up! Unless comment is private, post it to the
list.
Judy: We're working on format for change logs.
Alan: Yes, they're confusing.
Chuck: Big issue is resolving wording for identifying scope of the site.
Judy: Yes, that's most difficult piece of document. We not have come to a
conclusion.
Chuck: Number of confusions. Semantic confusion using term "expanded". Default
should be entire site, not subset. The order of point, looks like small set
is first.
Judy: Original discussion yielded two things. Whole site or reduced set.
Chuck: Generates confusion.
Andrew: Agreed.
Chuck: Maybe we need more explanatory text.
Judy: Conformance, one would select a sampling of pages the represent different
parts of site and run exhaustive tests on it. Then, take entire site and
run automated tests across entire site. If entire site is in range of million
of pages, instead of doing entire site, one could pick representative sampling
of pages that was largest than small set and test that and disclose that
not every page on the site was evaluated.
Harvey: dynamically generated pages need to be accounted for.
Judy: Identify two sets. First set, sample with representative page types.
Throw everything at it. Second set supped to be entire site, but if not realistic,
then could be replaced by "expanded" page selection upon which automatic
test would be run and disclosed.
Andrew: Becomes clear in point 2, need to put it in point 1.
Judy: Terms are confusing. Set A, Set B.
Shawn: Requires more thinking.
Helle: Put something in introduction that describes these selections according
to manual or semi-automatic?
Andrew: You still need to make a page selection.
Judy: Assumption checking. Working on this in-depth 3/4 year ago. Thought
then it was unrealistic to run tools across entire site. Still valid assumption?
Are there tools that do massive-scale evaluation. Sense talking to commercial
developers is that a lot has changed. Could we now way there are tests you
run on selection and tests you run on entire site.
Natasha: agreed. There are commercial tools available. Need to mention that.
Is some people can test with commercial tools then we should guide them.
Judy: Put a lot of emphasis on page selection. There are tools where you
can do millions of pages.
Shawn: Agree that things have changed. Good to have in there that if you
don't do the entire site then you need to clearly explain that.
Judy: Remove A-B version references and say page selection and entire Web
site. If for some reason you're unable to, then do an expanded page selection
and disclose. Note that goes somewhere in the document.
Andrew: Then we need to mention commercial products.
Judy: We can do that on tools page. Note there which can run across large
sites. How many can run with WCAG conformance as opposed to 508 conformance.
Judy: Bobby can do millions. Crunchy, Page Screamer?
Natasha: They do 508.
Judy: Need to do research on this.
Helle: What about LIFT?
Judy: Retrofitting tool.
Helle: You can test your site or retrofit.
Judy: Remove reference to expanded page selection, demote to exception note
only.
All: agreed.
Judy: Add more information to eval tools list about large volume evaluators.
Andrew: Need to identify when was update of notes.
Chuck: Still confused.
Helle: What do you do when you test millions of pages?
Natasha: Break them into smaller sections and test them, run the entire sites.
Helle: Maybe you just say you need to do this on the entire site.
Judy: Move Identify targeted conformance level. Identify page selection.
Identify entire site. Note then, if entire site is not realistic, expanded
page selection (see below). Remove referenced to expanded page selection.
Chuck: Okay.
Shawn: Add Identify page selection for manual testing.
Judy: Leave ordering, add in rationale.
Andrew: Manual and user testing.
Chuck: People don't understand "disclose".
Judy: Agreement that we needed to be clearer about disclosure.
Natasha: Not clear whether it should be on Web site, on paper?
Judy: Two stages, one while testing is going on, identification has to be
stated internally. After done, if making public statement, claim should be
accompanied by disclosure that entire site was not evaluated.
Andrew: In notes.
Judy: [Reviewed her changelog notes]
Judy: Adding issue numbering to changelog format.
Chuck: Change freeze to capture. Capture test pages as rendered in browser.
Helle: We use this in both prelim and conformance, in prelim we need to be
sure people know what they need to do, how to capture pages.
Judy: Is anyone evaluating dynamic pages?
Judy: Natasha and Doyle, please check with technicians. What are the 2-3
most important questions, large sites of dynamically generated pages. Can
test templates or capture generated pages.
Harvey: Does page content always have images or multiple images and are the
ALT-texted?
Andrew: If you have image database, what image are you storing with images?
Natasha: Send questions.
Natasha: More detail needed. Should be addressed.
Judy: Want to note location.
Shawn: 2.1 Limit the discussion in main part of document, refer to section
on this issue in specific contexts.
Natasha: Relevant level checkpoint, what does that mean? Give them option
to do that in semi-automatic and automatic fashion. Uncertain what we mean
by relevant.
Chuck: Only check ones that apply to your site.
Helle: Also conformance level.
Natasha: Need to clarify.
Andrew: Don't emphasize conformance level.
Judy: Use checkpoints that are applicable to your site.
Natasha: Would like clarification of "manual".
Andrew: Things you cannot do automatically.
Judy: Next section, usability evaluation.
Chuck: Many of the manual checklist checks can be done automatically.
Judy: Can you really do things on checklist automatically?
Doyle: If you're new, you cannot recognize that these things cannot always
do things properly.
Natasha: Run tool, when it flags things you need to check things manually.
Helle: Sometimes you need to do things manually that are not flagged.
Judy: We have noted in this document that things might not make most sense
to carry out in the order specified. Let's not get bogged down talking about
the order. Not hearing clear direction about what to do with this section.
Natasha: No tool can provide 100% guarantee. Will send suggestions for clarifying
automatic and manual checks.
Helle: Translating document, ran into trouble 2.3. Doesn't make sense. Open
page in HPR, close eyes and listen, open and look at page. Is information
equivalent? First, would never close eyes. Trying to compare incomparable
things, text browsers and graphic browser.
Judy: To do comparison, need things running synchronously. Also, information
might not come the same way. Easy for people who aren't familiar to, in looking
and listening, to fill in things that are missing from audio output. Seen
this technique used, had it used, brain fills things in.
Chuck: Good suggestion for people who are not accustomed to this.
Chuck: Need to separate text browser and voice browser.
All: Good.
Judy: Doyle, Natasha, Andrew, Shawn will look at Selecting Software.
Judy: Developing Org Policies, adding in some changes, needs to be looked
at. Started updating deliverables. Matt May will come on and brainstorm about
WAI page designs.
Judy: Editors stay on, g'bye everyone else.
Last revised June 26, 2002 by Judy Brewer