Summary of 24 March 2003 WCAG WG F2F Meeting

We broke into subgroups to work on the following work items

• conformance levels
  • Jason, Marc, Mark, Cynthia, Andi,  Dave
• map HTML techniques to success criteria/checklist
  • Ben, Michael, Paul
• definitions
  • Katie, John, Doyle, Alex
• scoping - create ways to include/exclude pieces of a site
• UAAG conformance profile/baseline profile
  • Loretta, Wendy
• I'm leaving on a jetplane...
  • Shawn

UAAG conformance profile for Checkpoint 5.2/5.3

• Conformance level: A
• Supported conformance profile labels must include all content type labels for content types that you are using. Options include:
  • VisualText,
  • Image,
  • Video,
  • Audio,
  • Animation,
  • Speech [Issue: do we want to require this label?]
• If technology supports events, then you must include the Event label.
• [Issue: Selection label. Not sure how to handle/refer to.]
• [Issue: do we need to specify that Input modality labels are not required or leave it implicit?]
• Applicability: [Issue: checkpoint applicability. UAAG checkpoints have requirements that the UA can't satisfy because of the content (particularly, #4, e.g., scripts). Should we put requirements on the content that alleviate these issues for the UA? @@ask Ian and Matt.]

User agent baseline issues

• How many different user agents must claim conformance? Just one? Just two? Must they be independently implemented?
• Do they need to be for different platform? What counts as a platform? Language (if content is French, then need French technology)?
• How long should that conformant UA have been available to users?
• What if only one UA conforms to UAAG? Are we tying accessibility to that one UA? Does that force users to use only that UA?
• We assume that conformance claims exist for UAs for authors to sort through and refer to AND that for some technology, there are some UAs that exist that conform. If our assumptions are incorrect, then that means that no one can satisfy WCAG.

Issues with 5.3/5.4

• Should these be combined to refer to UAAG profiles? for example, if you choose quicktime is there a plug-in that conforms to UAAG?
• scripting stuff (i.e., "custom user interfaces" covered by 5.1)

Other issues

• Issue: in the UAAG conformance section, at one point it says "This document does not require content focus and user interface focus" while someplace else it says "does."

Discussion

• How does this fit in? It addresses checkpoint 5.2, that says that technologies must be available in at least two independently-developed implementations. This defines what it means to be implemented, that is, a UAAG-compliant user agent is available.
• Do the new levels make that unnecessary? It is claimed that in the new proposal, guideline 5 goes away.

Conformance Levels

• Redefined them, then went through checkpoints and success criteria (through 4.2) to classify them in terms of the 3 categories and where necessary splitting success criteria across levels.
• This means that some checkpoints don't have minimum levels.
• The only part of  5 that would still be  needed is 5.1
• Issue: is it ok for some checkpoints not to have anything at level 1?
• Issue: we could attach levels to checkpoints rather than success criteria.
• GV's reorg had levels within checkpoints (as exists now?)
• It felt much easier to categorize with this approach.
• Action: ASW to write up notes and send to list.
• Action: CS and MU to write proposals for checkpoints
• Upshot: we can do this without requiring a transforming server
• Issue: we probably need a conformance profile mechanism.
• Action: @@ proposal for conformance profile mechanism.

Discussion

• Proposed way to handle - set up a conformance profile that would require reasonable levels of accessibility based on current technologies.
• It is not clear what people think when they see "1, 2, 3" vs "A, AA, AAA". It is valuable to think of  Level A, B, C - you might have to do some of each . People think 1,2, 3 is linear, but we've redesigned it as nonlinear. A good argument is that sometimes you need to do some of  2 and all of 1. (is this a conformance profile?) if so, then we need to use better terminology.
• It would be a good idea to get rid of the word "level"; it implies hierarchy. Things are currently lseparated into categories but they don't have a hierarchy.
• However, the original intent was that they were hierarchical. Should we call them "type" instead of level? might fall into 2.
• We found this morning that we needed 3 groups since there might be some that are mutually exclusive. There are some things that only fit into group 3 because an optimization for one group might be a "de-optimization" for another. Then these are not hierarchical. Someone could do a site that satisfies groups 1 and 3. Is everything in group 1 in the minimum? The use of A, AA, AAA hasn't been discussed.
• AT compatibility would be based on today's comformance profile. In the future, that would change as ATs change. Conformance profiles are separate, normative documents, based on UAAG conformance profiles.
• An idea: conformance profiles live outside standard, but the standard is constrained by them. Conformance profiles could change over time.
• Is the conformance profile something the author does for their site? is it now a spec? The author makes a conformance claim. A conformance profile is the requirements for a claim.
• If every country in the world created a different profile, they would be subsets, but what is the impact? there should be constraints on how to create profiles.

Techniques/Checklists

• We mapped techniques to 2.0 checkpoints.
• We found several techniques that could easily map to existing guidelines,
• We found some important techniques for which it was clear there was no checkpoint,
• Some techniques are not likely to move forward to future versions,
• We logged several issues, such as:
  • some techniques were more clearly UAAG techniques
  • some were usability issues rather than accessibility
  • customization and personalization is important and not well-represented in current draft (of checkpoints/guidelines)
• We need a way to describe how to use built-in accessibility hooks (e.g., labels for form controls)
• We have dealt w/special case issues (e.g., multimedia) but never said "all content needs to be exposed." e.g., many of the frames techniques do not clearly map back to checkpoints. content in this case = titles, table headers, etc. "hidden" content that needs to be exposed. Use those elements so that the UA can expose them.
• Action: MC and BC summarize notes and send to list.
• Action: MC/WAC put this on the agenda for Wednesday techniques calls.

Definitions

• defn of content: @@notes from Katie. "anything that is rendered by a UA. examples are..." followed by the UAAG defn.
• defn of user agent: @@
• documents/web services/web apps, etc. propose calling these things "web resources."
• defn of web resources: @@ points to UAAG defn.
• defn of web site: @@
• action: KHS to summarize proposal and send to the list.

discussion

• content management system? isn't that for intranets only?

CMS can apply to more things and not only intranets.
• User Agents: Real Player vs ActiveX Control? the last couple of bullets seem to make it less clear rather than more.
• Content vs User Agent: does the software do the processing or does it call on something else to do it. brokering and rendering. if doing that, it is not a user agent, but relying on a user agent.
• If we define content as something being rendered by UA and then say that UAs are part of content, the definition is circular. It is important to distinguish between who has the responsibility to do what, i.e. content author vs user agent developer.
• We should be mindful of circularity, but it may be necessary. We may be getting into trouble if we try to enforce a binary distinction when there is a dual character.
• The role of the author is to: either select a technology/language that can be rendered in an accessible user agent or select a technology/language that can be rendered in an accessible plug-in in the user agent. Then use that technology in such a way that the UA/plug-in can take advantage of the accessibility features of your content.
• Proposal: content is the payload of the http response - the stuff that comes across the wire.
• It's the use, not what you are doing. It's not a what, it's a why.  If you are using information (an http get), it is the stuff that is designed to be used by the UA. The content is info meant to be used by a UA.
• Is content defined by transmission? What if I have a local file? what if I am using FTP? what about PDF and Word? those are passed on local file systems, email systems.
• Cynthia's issue: it should only apply to content via the web. http-get allows for plug-ins and user interface elements. Binary files include images, audio files, etc. Were does the blog fit in?
• The technical accessibility issues (whether the file in local or not) are similar. Is it a practical distinction? perhaps for static files, not for web application.
• An author can embed media players in a way that the controls are not available. This is an author and UA issue. The author owns the tag; what the tag does is load the UA. Thus, the part that is content is the object tag and its parameters. "put buttons on embedded controls" sounds like a technique. However, this is only true for Quicktime (there are some players where this is not available to the author).
• "embedded components of user agents..." makes some people feel more comfortable.
• "content is anything that is intended to be rendered by a user agent." If create a local file, I am intending it to be used by user agent. (has support). However, from the user's perspective there are things that show up as content that were not intended to be there (e.g., battles between scripts and JAWS). There are techniques that an author can use to avoid.


What do you mean by intent?

• We need a tight definitionn of user agent. That could help with the circularity.
• consensus is key for the success of this doc.
• In user agent, don't mention any web services? user agents? user services?

• hear more about browser conformance.

Next Face to Face

• end of June (Cynthia, Andi, Marc, Michael
• early July (Cynthia, Andi, Ben, Marc, Michael, Katie
• Alex - later in July the better
• Jason not available in middle of July
• look for events to piggyback off of.
• agenda: techniques?

• HCI conference 23-27 (Crete),
• resna: 19-23 of June (Atlanta),
• v2: 17/18 June
• week of June 9 is out for Gregg
• WAI steering F2F? beginning of June or July?