Summary of 20 and 21 March 2004 WCAG WG face-to-face meetings
- Present
- Timeline and Planning
- Conformance
- Transition
- Dependencies
- IG Meeting
- Conformance, continued on Sunday
- JIS Presentation
- End to End Process and Issues
- Plan and Timeline, continued on Sunday
John Slatin, Tuomo Riihomaki, Andy Judson, Doyle Burnett, Ben Caldwell,
Michael Cooper (Saturday only), Jenae Andershonis, Paul Bohman, Katie
Haritos-Shea, Wendy Chisholm, Don Evans, Kerstin Goldsmith, Loretta Guarino
Reid, Marc Walraven, David MacDonald, Charmane K. Corcoran, Shadi Abou-Zahra,
Matt May (Saturday only), Andi Snow-Weaver, Takayuki Watanabe
Techniques
Requirements documents for techniques [1] On the critical path to get WCAG
2.0 Recommendation (i.e., by Last Call), we need to have "complete"
techniques documents for:
- HTML
- CSS
- Script/Web Applications
To transition to Last Call, we should have proof-of-concept drafts of the
following technologies. The more complete a draft the better, but "complete"
drafts of the following technologies do not seem as critical to transition to
Last Call. At a minimum, we need to show that WCAG 2.0 can generally be
applied to these technologies. A clearer requirement and exit criteria will
have to be determined and documented.
- RDF/semantic web
- server-side
- VoiceXML
- SMIL
- SVG
Test Suite
Timeline is TBD. We would like to have done by last call.
Usability
Need people to apply 2.0 to their sites and need a good mixture of different
types of applications, languages, and types of sites.
Question: What are the requirements for call for implementation
participants?
Question: Do they have to conform?
Answer: They should conform to the minimum, one site does not have to
meet all. But we do need examples of conformity from each of the sites.
Basically the sum of the implementation must show at least two
implementations of each success criterion.
Question: How do we take a snapshot of the conformance example? We may not
be able to link to sites that conform, because the conformance could
theoretically change.
Answer: don't know. Open issue.
Possible Action Items:
- Testing all of the techniques
- Redesign Working Group homepage
- Need to have 2.0 link to the techniques
- March 26 -- Need to publish working draft of the Gateway
- Techniques submission form. This allows anyone to submit techniques,
submissions must be vetted by WG before incorporated into a techniques
document.
Question: Have the test cases/suites been started?
Answer: Yes, we have several sources to pull from. Including Chris
Ridpaths work, and other W3C working groups, to name a couple. We are
coordinating with the UAAG, ATAG, and QA Working Groups.
Question: Do we need a Test Case submission form?
Question: How do we get others, for example developers, from outside of
the working group involved?
Action Item: Find developers to work on developing 2.0 conformant
sites.
Action Item: Send a "request for submissions" or "request for
techniques"
Action Item: Find more people to review and comment on drafts and
submitted techniques.
Question: What is the difference between implementation testing versus
test cases/suites are they different?
Answer: Implementation testing generally refers to the process of
testing implementations of technology specifications. This process serves the
dual purpose of verifying that the specification is implementable in
practice, and that implementations conform to the specification. This process
helps to improve the quality and interoperability of implementations.
(taken from [2].)
Request: Could we provide the group with an example of implementation
testing versus test cases?
Question: Are we separating HTML from CSS?
Answer: This is an open issue
Question: Will the checklist be normative or not?
Answer: We need to discuss this more with Gregg present. Maybe tech
checklist should be normative because they are necessary to understand the
guidelines? There are several issues with making technology-specific
information normative.
Question: How similar are HTML Techniques for WCAG 1.0 and HTML Techniques
for WCAG 2.0?
Answser: very similar, yet there are differences. Refer to David's
comparison. @@link
Question: If checklists are normative, what are implications for
technologies, such as PDF and Flash, that don't have normative checklists?
Answer: good question. open issue.
Question: if the techs are normative do they need to go through the entire
recommendation process? [last call to candidate rec to proposed rec to rec].
Answer: you may revise a recommendation in place as long as you do not add
anything new. Modifications to techniques will likely add new "features"
(techniques) and in that case they will need to go through the whole W3C Rec
process to be updated.
Question: If checklists are normative, then new techniques documents would
effect conformance. What would be required in a conformance claim - wcag
version, level, technology, techniques/checklist version? How would you
compare conformance claims?
Answer: open issue.
History
In April 2002 the requirements document for 2.0 was published. [3] Last
year at CSUN, Gregg presented a new conformance structure, which is basically
what is in the 11 March 2004 Working Draft.
The June 2003 draft grouped checkpoints into two groups: Core and
Extended. In the feedback we received, some preferred, and some said it was
too different.
In the 11 March 2004 draft we have 14 guidelines and 72 success criteria
which is interesting to compare to the 14 guidelines and 65 checkpoints in
WCAG 2.0.
Current open issues related to conformance center around:
- Scope and making claiming conformane claims
- Migration from 1.0 to 2.0
- How to make conformance claim
- Terminology
- How many levels/groups/categories of criteria
- How the levels are defined
Goal of the discussion was to try to determine which parts of conformance
we were near closing and of the aspects that were still contentious,
harvesting the concerns of the people in the room. Here are some of the
questions and issues we gathered:
- If my site has 1 million pages and many of those change 6 times per
hour, do I have to update my conformance claims every hour? Can I
randomly sample a portion of those pages every so often? If so, how many
pages and how often?
- It is important to several people that we define the different levels
very clearly. The current definition of Level 1 in the 11 March 2004
draft can be explained to developers, but is often difficult to explain
the concept of "default presentaiton" to people who are not
developers.
- In WCAG 1.0 we have two concepts: priorities and conformance levels. Is
it possible, no matter what terminology we use, to have just one name?
For example, with WCAG 1.0 could you have had priority 1 checkpoints that
you have to meet in order to claim priority 1 conformance?
- We want to work with people from around the world to harmonize on wcag
2.0. There is concern that the more conformance levels we define, the
more unlikely 2.0 will be widely adopted. The more ways we give people to
permutate, the more likely they will.
- If we allow + levels of conformance (intermediate conformance, granular
claims) will anyone other than "accessibiliy geeks" use them? Is it
useful to do? Will it create more confusion than a useful mechanism?
- Will the minimum level ensure that the most basic accessibility
barriers are addressed?
What documents do we want to provide?
- Mapping 1.0 to 2.0 and 2.0 to 1.0
- FAQ
- If I meet WCAG 1.0 A, how do I meet WCAG 2.0 A
- WCAG is not just for HTML only – anymore
- "who should use" - reaching more technologies
- clarify relationship to proprietary technologies
- How do adopt 2.0 as a policy
Open issues:
- people aware of 1.0 are using html. what about people who are not using
html. they are a whole new audience. a transition task. hand to EOWG?
- when run conferences or workshops, it is clear "who should attend."
similarly we have "who benefits" but we should also have "who should
follow."
- how translatable is 2.0 to other languages? Action Item: Send request
for review of 11 March 2004 draft to offices and ask about
translatability.
- opinions about legal uptake of 2.0. diffs in laws. handle in "benefits
of standards harmonization?"
- policy makers are in audience - so have something for them. but what?
who will write it?
- articulate benefit of shifting from 1.0 to 2.0. if you have set a
standard, here are the benefits of shifting to 2 and why. one reason
(among others): testability.
- translation - more translatable expands the audience.
- don't need to sell it - most people aware of and waiting for.
- any time start "making case" send to eowg.
- a guide for policy makers (ala discussion at venice mtg).
@@add link to IG Notes
Conformance-related Straw Poll Questions
20 people total. Not everyone voted and some people raised their hand for
one of the options, but then also said, "don't care" (thus the "~2").
- 3 levels or 2? no discussion about how the levels are defined
or if using two levels meant that some that are currently level 3 might
end up in level 2 or be deleted. It was just a "what is your general
feeling about how many levels we should aim for"
- 3 levels - 6
- 2 levels - 11
- ~2 people don't really care
- should we create a flexible claim?
- make it clear, simple. define x levels (2 or 3) - 15
- however, seemed to be consensus that while we only want to define X
number of conformance levels (2 or 3) we don't want prevent anyone
from detailing which criteria they do and do not meet.
- Some people have said that reusing A/AA/AAA is confusing because it is
the same terminology as WCAG 1.0 but it means something different.
should we use the same labeling scheme as WCAG 1.0 (A/AA/AAA)?
- yes - 12
- no - (use different terminoloogy) - 3
- don't care - ~2
We spent quite a while discussing conformance testing on very large sites
and sites that aggregate content.
- Content aggregators are likely to have policies/standards that content
must meet before it is used on the site. For example, they may reject
content that cannot be downloaded in 10 seconds.
- Do i have to test 100% of my pages? Do I have to test every 10
minutes?
- What about the idea of "conformance profiles" that had discussed
before? One set of requirements for documents compared to a different set
for applications? One set for aggregators and another for people who
create all of their own content? Requirements for templates?
- We should bring to the table people who are creating syndicated
content: advertising agencies (ogilvy), syndicators (cnn, washington
post, reuters, weather feeds, stock tickers).
- Should we address RSS feeds?
- Aggregator terminology: "well" - a well of content is where one
syndicator's content appears on the page. "portal" and "portlet" - a
large portal might contain a portlet within it. for example, the aol
portal may contain a cnn portlet.
- What if syndicators sent metadata with the content? Could an aggregator
aggregate the conformance information of all of the wells and create an
aggregated conformance claim to wcag?
- Are aggregators only responsible for the content they create? In many
cases, this would only be the framework. However, "it's the network that
carries janet that gets sued, not janet." (refering to the recent Super
Bowl half-time show where Janet Jackson's wardrobe malfunction caused CBS
and MTV to be sued for indecency)
- Could aggregators say, "we only purchase accessible content" - similar
to 508?
- Similar issue for educational software that claims they are 508
compliant but the courses created with it are not 508 compliant because
the teachers are creating content and not aware of 508 provisions.
Therefore, the software is often hit with claims that they are not in
compliance, but it is because of how the teachers are using.
Creating a harmonization target
What is needed for WCAG 2.0 to become a convergence target? For WCAG 2.0
to harmonize existing policies?
- Business case for ATAG.
- Ideal if testing for conformance was automatable. At minimum,
everything must be testable.
- Add something in conformance scheme about level of confidence or how
much tested or what kind of testing done?
- Is the definition of conformance in WCAG 1.0 adopted? Yes, where WCAG
1.0 has been adopted the policies say "level {A|AA|AAA} conformance" and
rely on the WCAG 1.0 definition.
- Could someone say "we're 20% conformant to WCAG 2.0"? Is it possible to
have an "A-" level? Here are the guidelines that we
don't meet? Look at UAAG?
- How achievable is the minimum level of conformance?
- Is the problem with evaluation and development tools? If they had
better support would it be less of an issue?
- What are the differences between policy vs conformance claim? State
exceptions in the policy not in our guidelines and definition of
conformance? For example, real-time captioning for EU meetings would be
in 20 languges at a time. expensive.
Looking at policies and conformance from around the world.
U.S. and 508
- VPAT is how claim conformance to the law.
- Purchasers decide which product "best meets the standard" with respect
to 508. Developers have the option of saying "these are the things that
don't meet 508"
- if it's a contract to build a web site the contractor usually says,
"the site will" not that "the web site does" because the site doesn't
exist yet.
- vpat in scope of work of the product. for service - part of
contract.
- does claiming conformance mean filling out a checklist? for 508,
yes.
- vpat applies to product or service
- 508 doesn't say anything about scope. how to make a conformance claim?
purchase point is a point in time - that's when evlauation occurs
Japan
- conformance is not so well-defined
Europe
- European Commission internal policy: wcag 1.0 level A or AA. EC copied
checklist into communication. policy page states exceptions and
explanation. Pages that fully conform display a logo, for details there
is a checklist attached to "the communication." "In principle, pages are
compliant but not a guarantee" - logo usually only appears on home
page.
- EC cover web applications? not sure.
Canada
- policy is WCAG 1.0 AA. plug into common l&f (covers more than
accessibility). lots of templates and CMS to reach CLF. province of
Ontario: ADA-like, but no teeth.
Test suites, Implementation testing, Candidate Recommendation
Will implementation testers have use of evaluation tools that are
knowledgeable about WCAG 2.0? Currently, we know that Chris Ridpath's OAC
should be inline with 2.0. We could (or help tool makers take) existing xml
output from tools and convert to a WCAG 2.0 report via xslt. Paul Bohman,
Andy Judson, and Michael Cooper are interested in working on this.
Will we test the implementation tests to see if following WCAG 2.0 results
in content that is accessible? John Slatin is interested in this question and
possibly able to do some testing. WWAAC is a potential candidate to test for
intellectual disabilities.
Prof. Takayuki Watanabe presented a comparison.
- All WCAG 2.0 guidelines covered in JIS.
- One JIS guideline not covered in WCAG 2.0 guidelines.
- Six JIS guidelinse have weak correspondance to WCAG 2.0 guidelines.
Determined it would be helpful to compare JIS guidelines to WCAG 2.0
success criteria.
Summary of Japanese-specific issues (many of us saw these at the November
face-to-face, but many have not)
- Unicode ambiguities
- if type "hoshi" in a Japanese word processor, get a number of
characters to choose from. The character is pronounced differently
between screen readers. unicode is ambiguous and complicated.
similar-looking but diff characters are included in unicode. Isn't
this a similar issue to bullets used in creating lists in Word? Isn't
this an issue of marking up lists properly or are the characters used
for more than list bullets? They may also be used for layout
information.
- unicode 2460 is "circled digit one." 2460 is not always displayed
in the same way. sometimes it is 1 in a circle, sometimes not.
- Is unicode ambiguous about what character should be displayed?
- Different applications have different "correspondance tables"
- Font of Kanji characters
- kanji characters that consist of many strokes, need to be large
font to read. might not be a problem if font compensates. issues
w/how font designed as well as used.
- why 5.6.a corresponds to 4.1 - user agent allow to increase font
size. using "em" - don't fix the font size. font is deprecated in
html 4.01. *** an issue to come back to. make sure we're in synch on
this.
- whitespace or linebreak in middle of word
- e.g., t<br/>a<br/>
- no way for screen reader to get meaning
- Characters that look similar with different pronunciations
- diff in zenkaku characters. two characters that look like a dash,
only diff is weight/width. one word means "lead" other means nothing.
or Zenkaku Z and Hankaku Z
- proper nouns written in Kanji sometimes have various
pronunciations.
- similar to what an abbreivation means. how author state
intense.
- ISO Guide 71
- JIS is developed under ISO Guide 71 thus section 6 of JIS Part 3:
general requirements for assurance and improvement of information
accessibility.
- this section: require changes in content or process? way to help
people if all else fails (email and phone)?
- Look at Guide 71. Can WCAG 2.0 be written in accorance with it? If
so, more likely to be adopted by JIS?
- should wcag 2.0 be written in accordance with ISO 71?
- do we need to define accessibility? we're saying "do these things" not
"reach these people"
- straw poll: Do prefer the wcag 1.0 definition of level A or the 2.0
draft definition of level A?
- 17 like the 11 March 2004 WD definition of Level 1/A
- 5 prefer the WCAG 1.0 (1999) definition of priority 1/Level A
- Those who do not wish to use the 11 March 2004 definition of Level
1/A said:
- Are we meeting the target audience? Are we facilitating
adoption? The definition is confusing. It will take a lot of
educating to use the definition.
- I'm too new to have an educated opinion. However, as it is, I
can't sell it to my organization. It is too hard for anyone to
adopt. Part of selling it is automating the testing.
- I am conflicted. I agree that we need to figure something out
and meet our schedule. I agree that all checkpoints are
important. 508 doesn't create a minimum. You declare what you
achieve and people compare what you have done against your
competitors. The minimum is the whole set, but you can make claim
that is less than minimum. My concern with saying that there is
not minimum is that there are relationships between requirements.
You can't create a VPAT for every page (too difficult, too much
work). Thus, in 508, no minimum.
- I don't understand "not effecting the default presentation."
The definition is difficult to understand.
- I like the old approach, p1 is a ramp (there aren't stairs, so
with the help of one person I can get into the building. I don't
need two body-buiders to lift me up a set of stairs), p2 is a
usable ramp (i can roll myself up), p3 no elevation (no need for
ramp).
- issue with tying conformance to testability: it is evolving and will
continue to do so. what is not testable today might be testable
tomorrow.
Looked at conformance in UAAG
- uaag talks about applicability vs haven't done.
- document vs application
- aggregator vs content creator
- types of technologies: voice or not
- base issue: what are the minimum requirements and that they almost all
require human intervention.
- reason to do something like uaag: use diff minimum for diff groups.
person claim conformance for cms system but not content created with
it?
Definitions
- accessible: which groups do we mean? if we don't follow the principles,
many people not use. (in intro purpose) we say what happens when the
prinicples are ignored but not what happens if follow.
- web content: uri? what about web applications?
Defining the minimum level
- EOWG argued for 3 levels instead of 2. the simpler we can make it for
end-users, the easier.
- Is our minimum level a ramp? It's not clear. @@map wcag 1.0 checkpoints
to 2.0 success criteria: gottfried already has start on (from 14 february
draft). ben will complete and andi offered to help.
- when we look at success criteria, we ask, "should this be in minimum"
without looking at the criteria. so, still influence of 1.0 idea of
"important"
- not all guidelines have to do with presentation.
- Should there be a difference betwen how we define minimum level and how
we determine what goes into level 1?
- interpretation of the 3 priority levels in wcag 1.0 - 1: ramp (they are
not steps, but someone, if needed, could get pushed up it rather than
carried up steps), 2: usable ramp (can use myself), 3: no grade
- ideally: p1 are those things that are widely applicable across types of
content and audiences (higher levels are more specializes and less widely
applicable). for ourselves, help decide where put things by how effect
presentation. can we be open and recognize that we are not able to meet
the needs of everyone, particularly intellectual disabilities? Note that
many of the things we do address issues for people with intellectual
disabilities.
- guide 71: cog (intellect, memory)
- wwaac - alt representation, summary of home page (in meta).
- can't write court brief in lesser style, can't make "reader's digest
version." do need to state that we've addressed, but are not in a
position to tell authors.
- guidelines are looking at the things that create barriers. principles:
don't create this kind of barrier. guidelines: how can you avoid this
kind of barrier. priorities - techniques in avoiding these kinds of
barriers, what are teh exceptions we allow. priorities associated with
greatest barriers?
- how diff from must should may? it's the same...level 1 that crfeate
most barreirs.
- include specify cognitive? don't, sake of durability of document.
technologies will develop.
- important to discuss limitations of the document?
Updating Requirements for WCAG 2.0
We went through all of the stateemnts in "Requirements for WCAG 2.0" to
determine where we needed to update, clarify, or change statements. In some
cases, we no longer have consensus on consensus statements, but others are
still solid. Wendy will work with Gregg to update and publish a revised draft
to the group.
#2
- dynamic content - what mean? do we do? mean - user profile? application
function is to look up things in database...
- relationship user agent support - clarify. keyboard requirement.
- mention QA in beginnin gparagraph? @@
- @@?? minimum level address largest barriers to accessibility
- testability??
#3
#5
- speech? needed for voice input apps
- other? is it reallyl everything?
- language symbols - people using symbolic language - are we addressing?
success criteria that support concept codes? think we've got it
covered.
#6
- providing transition materials
- if we can map
- doesn't say that if a site conforms to 1.0 it can conform to 2.0. "not
completey" is subjecctive
- did we define "accessible content?"
- @@explanation that can be added based on wac's presentations
R
- put part of in audience?
- liink to benefits of standards harmonization?
- link to other requirements document - where?
- put in something about it becoming a convergence target? our desire for
doing so?
N
- normative and not automatic testing...don would object.
- ease of testing? it must be practical to test it.
- n9 delete
C
- C1 means "betwen levels" - not sure? probably not. no consensus.
- C2 - opened that up and this meeting.
- C3 s/accessiblilty/conformance, s/checkpoints/success criteria. "here
is complete set, which have you done?"
- C5 - redundant w/benefits requirement.
- C8 - conflict with C1?
M
- M2 - sure but we should not require
- M4 - however, we put exceptions in checkpoints. either conflicts with
"no conformance claim" if don't meet minimum, or interesting sidenote for
metadata. don't have consensus. in checklists, have ideas of "notes" - "I
implemented this technique that is not in the checklist...via someones
authority b/c...." there are things have changed in the world but 1.0
hasn't changed. e.g. space in form fields. we assert conformance, but
don't need to meet that. as author, i have a legit argument.
- use cases for metadata: 1. negotiation between client and web page 2.
legal issues - i've conformed but i've interpreted the guidelines in this
way. is it metadata or site internal doc or policy page? don't put legal
in metadata (in elements).
S
- S1 - more of a success criteria or techniques? is this an equivalent
facilitation clause? includes transformations and web services. mostly -
server-side techniques. at guidelines say "or provide equiv
alternative."
B1 - is for techniques requriements
G
- G1 - let's meet our own guidelines
- G2 - we're not categorizing as "accessiblity" and "uisability." we're
not trying to make a distinction. leave it. give to john slatin.
- G3 - we're doing for X is it harmful for Y. this is relevant to success
criteria.
A1 - good for introduction.
these need to move up to requirements. link to technqieus requirements.
mention other technologies - like to see a statement, "these intended for
all technologies on the web, even though we only create techniques for w3c
techs"
- what is at risk - how things are dependent on each other
- mock-up of how to connect 1.1 all the way through to html techniques,
then techs and test cases woudl be done. to complete this, team people
from the techniques task force with people from larger group.
- action items: each take a success criteria (or group) and find web
sites that conform and not. connect to techniques and test cases.
- issues with publicly stating sites that do not. what doing with
them?
- way-back machine? implementation reports: snippets of code?
- conformance evaluations....only those sites that really conform...talk
with developers.
- do we need checklists or evaluation forms?
- each person takes x success criteria, here are the existing techniques
that seem to map to it and fulfill it (or i have technique w/out success
criteira, or SC w/out techniques), that helps us find techniques to do.
- vetting techniques real-live examples
- upcoming plan:
- next month (whiloe people review 11 march draft), try to tie
techniques with success cfriteria and find real-world examples
- get through in next month, then begin processing review
comments.
- anonymize the sites that we find
- create mock-ups later
action item: take one criterion, map to techniques, think about other
technologies. look at html techniques to see if it is in there. look for real
world sites that do and don't meet. test cases that used to determine.
june f2f?
AOL/Amsterdam or Paris, IBM/Paris??
RESNA 18-22 orlando
v2 plenary 14-18 nist, d.c.
UPA??
2nd week of june??
Group photo at the end of Sunday - front row
left to right: Kerstin, Jenae, Katie, Shadi, and Charmane. Back row left to
right: Don, Doyle, XX, Takayuki, David, Andi, Andy, Marc, Loretta, Paul, Ben,
Wendy
[1]http://www.w3.org/TR/wcag2-tech-req/
[2]http://www.w3.org/WAI/GL/implementation-testing/Overview.html
[3]http://www.w3.org/TR/wcag2-req/
$Date: 2004/04/01 01:39:33 $ Jenae Andershonis, Wendy Chisholm