W3C Web Accessibility Initiative

WAI Authoring Tool Guidelines Working Group, WAI Evaluation and Repair tools group

WAI AU / ER Teleconference - 11 July 2000


Chair: Jutta Treviranus

Date: Tuesday 11 July 2000

Time: 2:30pm - 4:00pm Boston time (1830Z - 2000Z)

Phone number: Tobin Bridge, +1 (617) 252 7000


The Latest Draft is the Recommendation dated 3 February, available at http://www.w3.org/TR/2000/REC-ATAG10-20000203. The latest techniques draft is dated 4 May March, available at http://www.w3.org/WAI/AU/WD-ATAG10-TECHS-20000504. The latest draft of the Accessibiltiy Evaluation and Repair Techniques is dated 15 March at http://www.w3.org/WAI/ER/IG/ert-20000315

  1. Review outstanding action items
  2. Other business



Action Items and Resolutions



KD: Has translated for WAI.  concerned about quality of tools and software 
in the world.  new job to be defined, will try to improve implementation of 
recommendations in software and documentation etc.  Prefer to have some 
comments from us, what we think about quality assurance, what's your 
experience, how to achieve.
JT: Karl is new w3 conformance manager

JT: I'm chair authoring tools

MC: I'm project mgr bobby at cast.  mainly with er group at WAI

HB: On number WAI groups, consulting on digital talking books

CR: working on a-prompt, editing AERT

DB program mgr access

HS: member access and disability group at MS

FB: from html writers barnett guild

WL: smith-kettlewell, part of ER and AU

WC: with er, staff contact and editor, also editor GL

BM: cast, Brian Methany bobby

LK: Institute on Disabilities/UAP at Temple, Chair Evaluation and Repair 
Interest group

CM:  Charles McCathieNevile, Staff contact AU

JT: anything to add to agenda?

WC: add to agenda integrate au and er

JT: lets explain  tools for ER


LK: we give up/down or tools or just guidelines?

KD: will not give advice on commercial products at first certainly, cause 
big job, very difficult, and perhaps not best method for first approach. 
Easier to try to convince people on recommendations and 
specs.  Recommendations and specs very important.

Important for WAI, because influences html writing, much larger than 
accessibility, could help convince people to do more job for 
recommendations in general.  Will set up mailing list for all groups to 
have right way to do my job.

JT: will set up listserv?

KD: will try to have mailing list

WL: start at home, best thing is to get w3 and WAI web presence to make 
sure that all of w3 site is conforming.

HB: good learning experience for those responsible for recommendations.

JT: w3c and WAI should be first sites. Start at home

KD: sure

JT: and move to Amaya

LK: include tools like change control tools

KD: also, hearing impairment.  Do you mean tools on site?

CM: yes, we have tools like change access control, mailing list tools, 
other internal tools.  Accessibility is being worked on.  One of my roles 
is to interface to the system team.  So send to me or sysrec@w3c.

A couple of areas where conformance runs beyond w3c pages, we should be 
exemplar, not run behind.  Lot of call for "how do we do this?"  This is 
what we're specifying in contract, how to make sure it's being 
tested?  Can't just look at code with Emacs.  So pressure coming from 
people asking, what's tools, where's the software, how do we know what's 
good?  In au we want to be able to say: here's what we know about the 
various tools?  We've don't by informal assessments.  No checkpoint by 
checkpoint comparisons, haven't said, "doesn't pass".  But we are in 
position to make decisions to particular checkpoints.  We need a 
reproducible process so that I and developer get same answer.  And that 
anyone would have come up with.  People are asking the questions.  We're 
not sure how far we go in answering, but are doing a disservice by not 

JT: in other words, we need conformance objective conformance, for AU 
that's a big body of work?  The amount of work to make tool is beyond what 
we can do.  One of the hopes was that you would be able to help with effort 
or process.  In terms of time you have, how much would you have to devote 
to tool, or architecture of tool?

KD: don't know my time allotments, but will know in a few days or 
weeks.  Will try to be available to all groups and help everyone.

CM: we could look at how many activities across w3c, could expect 4% of 
time.  But what is w3 approach of this?  Has you looked at AU conformance 
work? I.e. our tests.

KD:  No, not yet.  Will look.

LK: can integrate obj sub

KD: be fully conformant, not just with technical specs, need to manage.

CM: can write syntactic html which is bad HTML also.  E.g. can be fonts, 
perturbations, etc, all wrong.  It will all parse, but goes against w3 html 
really says.  So it's wrong. So it's not just in accessibility where you 
can't test accessibility.

KD: So are you saying recommendations need to include WAI?

CM: yes, in part.  E.g. in SVG, halfway approach, conformant SVG must meet 
authoring level A.  That kind of spec in w3 in general  is helpful.   HTML 
is not correct just because it has right pointy brackets.  So it's not just 
accessibility that needs human checking.

LK: Can you give examples not having to do with accessibility.

CM: 7 paragraphs with bullets  instead of list is not really meet 
specs.  Or using blockquote to indent.

KD: In addition to accessibility, there's usability for people without 
disability.  maybe we should have recommendations that it be usable.  If 
you make document with white character it's generally inaccessible.

LK: What about white box vs. black box testing?  In white box testing you 
know what code is running and do tests on the code.  In black box testing 
you just have user operate and observe what happens.  So e.g. if you are 
testing the HTML code that's white box texting, if you're just using a 
browser and screenreader that's black box testing.

DB: both done at MICROSOFT.  Hard to do black box testing. depend on the 
assistive technology

CM: most developed while only way to be sure is to complete black box, so 
subset to most efficient point, sometimes can do code test.   Would like to 
automate the test also to read browser and read the results.  Line is 
whether you're looking at underlying code or looking through a 
browser.  usability testing important and expensive.  So how to do things 
from start to see if things are right?  Also, w3 doesn't certify 
anything.  If anyone stands up and claims conformance other people can say 
opposite,  so people are reluctant to say it if they are not supported.  So 
they ask if they can hang back and wait.


JT: so to summarize,

1. would like WAI conformance in ever tying in w3c
2. every evaluation as automatic and human judgment... challenge is how to 
do conformance by machine,
3. and how to do objective, repeatable human judgment.  Part of this is 
compromise, because of all differences in browsers.  And representative 
sampling of white box sampling.

LK:  What do you mean by sampling?  E.g. do you mean instead of looking at 
all alt text just use statistical sample?

JT: or try sample browsers instead of all

KD: Do we need test suites to make better tools.

CM: will start develop test suite for authoring tools.  Fairly large piece 
of material.

Another big part: building processes, the development groups have more 
experience.  Need process: these are steps to go thru so that when they do 
testing they've got it right.  Important here in Australia  Done in ER?

WC: EO has looked into it some.  Have doc on where to begin?  Is it GL or 
ER?  Next version of GL will say that minimum requirements are easily testable.

WL: EO is link to outside for everyone.

CM: what's your experience in developing process where tools come out 
accessible, (question to dick and heather)

Heather: we're learning how to do that properly,  e.g. didn't do first time 
in ie4.  How do we test and agree product is ready to go... is that what 
you're asking: :

CM: no, what process you need to be doing?  E.g. for lots of stuff, like 
all stuff under MSN.

DB: talking about sites?

CM: both.  Easy to see in sites.

DB: process more along for applications, cause we've had own guidelines 
that predate WAI.  Still developing systematic process.  Still pretty young

Heather: agreed.  In application we have check and balance, program 
manager/developer/testers .  We teach all how to conform.

DB: in MSN side, checklist is one of the big things we use.

JT: one of most powerful tools for large site is to create accessible 

DB: Yes, that helps, but much more needed.  We also try to do things for 
things common across site families.

WL: Also, every time you do this stuff, keep in mind everyone who will do 
whatever you want.


JT: so other points

1. how to give guidelines on process
2. how to evaluate when done.

HB: how many other groups have you been getting this info from?

KD: You're the first.  Will be talking about schema later.  Theoretically 
easier cause maybe more automatic.  WAI most difficult because of human 
judgment.  Just starting here at Sophia.

JT: any last thing?

LK: don't put all load on accessibility.

WL: Must be usable to be accessible.

MC: yes, but that's always true.

WL: diverts efforts if trying to make sites usable by all.

DB: if we focus too much on solutions for all takes effort from accessible.

JT: also difference in degree.  usable different, cause less effect

WL: but if I can't order on web, it's a complete barrier.

MC: As definition includes cognitive more they come closer together.


JT: action items

(to CM and WC) talk about database tools

CM: will work on Thursday with system team for database for checking 
tools.  Should be on line pretty soon.  Will then work with Ian on 
techniques unified document for techniques for /* document would come from 
database to produce doc  */
create views whereby user can specify what view they want

So for authoring tool, can create view that gives techniques that I need,

create database so we can create views and collectively do work.

JT: CM... you're working on testing tool and instrument in addition... is 
that true

CM: /* dropped */

JT: Would ER look at things other than content.

LK: resolution was just doing content

JT: are you working on testing instrument, not just database?

/* CM returns */

CM: yes.

LK: say more about testing instrument?

JT: tester for AU tools to make elements of human judgment as objective as 

CM: here's how to do test, is document, in some cases can automate.

JT: it's the AERT for authoring tools.


JT: anything else?

Next combined meeting first Tuesday of august, august 1, 2:30pm Eastern Time

Copyright 2000 W3C (MIT, INRIA, Keio ), All Rights Reserved. W3C liability, trademark, document use and software licensing rules apply. Your interactions with this site are in accordance with our public and Member privacy statements.

Last Modified $Date: 2000/11/08 08:13:13 $