W3C home > Mailing lists > Public > w3c-wai-ua@w3.org > July to September 2000

Raw minutes from 17 August UA Guidelines teleconference

From: Ian Jacobs <ij@w3.org>
Date: Thu, 17 Aug 2000 16:11:22 -0400
Message-ID: <399C46EA.D25F5F86@w3.org>
To: w3c-wai-ua@w3.org
17 August 2000 UA Guidelines Teleconference

 Jon Gunderson (Chair)
 Ian Jacobs (Scribe)
 Mickey Quenzer
 Eric Hansen
 Harvey Bingham
 David Poehlman
 Tim Lacy
 Dick Brown
 Gregory Rosmaita

 Kitch Barnicle 
 Denis Anson
 Jim Allan

 Rich Schwerdtfeger
 Charles McCathieNevile

Next meeting: 24 August
 Regrets for next meeting: Mickey

Minutes of previous meeting:



   1.FTF meetings for WCAG and ATAG working groups in
      Bristol, England in October

    HB: Tentative EO ftf in October (In Ottowa)

   2. Checkpoints resolved unless complaints: 2.3, 8.6, 7.2

   3. Action IJ: Talk to Judy about UA ftf.
    IJ: She's on vacation, so continue this.


    1.Issue 294: Native support and downloadable modules

    Refer to JG proposal:
    Refer to EH comments:

    JG: I spoke to Al Gilman, Denis Anson, Len Kasday about this
    issue. My proposal is a result of that. People agreed that as long
    as modules required to conform were documented, that would 
    be sufficient. Still need to encourage developers to provide
    accessible browsers by default.

    IJ: Model is (a) anyone can claim conformance 
                 (b) encourage accessibility by default.
    GR: I'm concerned that that's not sufficient. Importance
    of accessible installation. I think that developers need to
    provide facilities to install components. You need to ensure
    that the way of getting and installing software is accessible.

    EH: Chicken and egg problem. 

    IJ: I think that assertions about conformance are different
    from operational aspects (installation, privileges, etc.).

    EH: When you combine things into a compound UA (screen reader,
    general purpose UA, etc.), then it follows that the requirements
    of UAAG 1.0 apply.

    IJ: The compound entity conforms as a whole, not independently.

    EH: All the requirements apply to all the parts about which
    the claims are made.

    GR: I think that the means of getting software have to be 
    accessible as well as the software (and it's documentation).
    If Vendor A relies on Vendor B, then it's incumbent on Vendor
    A to ensure that Vendor B's stuff is accessible.

    IJ: No, no dependencies like that! These are just claims.

    /* GR still doesn't have access to the Internet */

    IJ: Why this model promotes accessibility:
      a) Allows developers to improve accessibility of deployed
      b) Allows AT developers to make claims.

    JG: What are the parameters that make something obtainable
       in an accessible manner?

    GR: It needs to be accessible over the Web according to WCAG

    JG: So I hear that at least one form of the software must
    be available in a WCAG-accessible form.

    IJ: Current text in document:

       "In order for people to use the user agent at all, 
       the installation procedure (and any subsequent software 
       update procedures) must be accessible according to the 
       guidelines of this document. For example, the user agent 
       must provide device-independent access and accessible 
       documentation of the installation."

    IJ: Furthermore, "applicability" applies to all of the

    EH: Is it permissible to say "product A works with any
    product that supports X". 

    IJ: I think for simplicity, you need to name by brand and
    version, rather than generic claims.

    IJ: Profiles by name? (e.g., conforms by default).

    JG: I don't think that that's useful at this point.

    EH: I think that composite user agents are the only way
    you'll get conformance for the foreseeable future. 

    TL: I agree with the summary: from Microsoft's perspective,
    we won't guarantee that we'll work with another UA. But we
    do encourage, and see in the market, that certain screen readers
    will say "we're optimized to work with IE 5.5". It's good
    for MS to be able to say "We will work with any AT that works
    with such and such and interface."

    4.Issue 295: Repair functionalities required / relationship to 
      conformance to other specs.

    IJ: Two questions:
      a) Should we delete 2.5, a repair checkpoint? (Priority 2).
      b) If we don't delete 2.5, should we mark it as a repair 

    TL: If you have UAs generating missing markup, you will result in 
    lazy authors. This could perpetuate poor markup.

    JG: But they won't conform to WCAG, which is what is their
    EH: I'm concerned about it from a different standpoint: it's
    presumptous to have a machine not under the Web content developer's
    control trying to guess what the functionality is.

    IJ: Refer to "altifier" in techniques.

    GR: I think that this is mostly to make the user aware that
    an object exists that has not been given an equivalent.

    EH: I'm concerned that we call this a "text equivalent" if 
    it hasn't been provided by the author. Call it something else.
    Also, should we indicate explicitly that this thing has not
    come from the author? (i.e., that this is an annotation from
    an agent that is not the user).

    IJ Proposed:
     a) Leave 2.5 as a repair checkpoint.
     b) Allow configuration to generate or not.
     c) If configured to generate, provide information to identify
        that the content has been generated.

    DP: "c" may be more than what users actually need. It doesn't
        matter to me that I know whether the author provided it or not.
    IJ: Is it important to accessibility to have "c", or just

    JG: It is important to the user to go back to the author
        and say "I need this fixed."  

    DP: I don't think notification of generation is an accessibility

    /* DP leaves the call, but not in anger <grin> */

    GR: Lynx has a universal setting for images:
        Render generic placeholder. Or, you can ask for the filename.
        I don't care that whether something was generated if I
        get a page of filenames.

    MQ: I don't need to know that it's generated.
    IJ: At least three users have stated that identification of
    something having been generated is irrelevant.

     a) Leave 2.5 as a repair checkpoint.
     b) Allow configuration for generation or not.
        DP: I want to turn off generated alt text when I evaluate
     c) Don't call the generated thing a text equivalent.
        EH: Just say "supply the URI, element name, etc.)"
     d) If object not included by reference, supply element name
        as a place-holder.
     e) Add a TECHNIQUE:
        If configured to generate, provide information to identify
        that the content has been generated, not provided by the 

    TL: Is the generated thing the value of "alt" in HTML, for

    IJ: I don't think so, it's not what the author provided. 
        But how does the AT get it?

    JG: The AT may generate whatever it wants based on missing
        markup. The AT makes the same conclusion as the conforming

    GR, IJ, TL, JG, DP: Yes.

    6.Issue 302: Definition of "recognize" needs review.

    IJ: Recognize should probably include something about recognizing
    what a script does. A UA should be responsibile for knowing that
    a script is calculating a factorial, for example.

    IJ: Does anyone want to take a stab at rewriting "recognize"?

    IJ: What I mean is "you aren't responsible for worlds you can't
    know about."

    Recognize is:
     a) Know something according to spec.
     b) Know something according to volition by the UA 
        (e.g,. MARQUEE). 

    IJ: Note that "recognize" is part of 2.5 (and should be part of
    It's part of applicability to factor it out of checkpoints.

    IJ: "Recognize" is really about "the UA has been programmed to
    do this." You need to be able to identify text equivalents,
    captions, etc.

    JG: Important here is the part about the UA's <em>response</em> to

    EH: The value of the term "recognize" is that it's a signal
    to the reader. Not every text equivalent is properly identified.

    IJ: Subtle point is scripts: you "know" javascript but not the
    function of the code.

    EH: Some concepts like "recognized by a machine" or 
    "understandable by humans" are very tricky since there are 
    varying degrees (is artificial intelligence required? no...).
    We expect machines to be able to make some choices, e.g.,
    generate missing text, you could even have a UA be able to
    recognize a trend in a table. I don't want to rule out someone
    doing something smart with complex data. Does all responsibility
    lie with the Web Content author? Can we avoid saying in the 
    Guidelines exactly the limits of "recognize"?

    EH: Recognition doesn't require deep inferences when evaluating
    a script.

    Action IJ: Repropose a definition of "recognize". Include some
    information about scripting (distinguish opening window from
    calculating a factorial).

    3.Issue 293: Is the UA responsible for control of timed
      created by scripts?
    JG: In the past, we've said that the UA is not responsible for 
    knowing everything about the results of a script.

    /* JG leaves the call */

    IJ: Looking at current 2.2: 

         2.2 For a presentation that requires user input within 
         a specified time interval, allow the user to configure
         the user agent to pause the presentation automatically 
         and await user input before proceeding. 

    EH: Think of a testing application: you have three minutes to
    answer a question, otherwise you must move on.

    IJ: The question is, does 2.2 include presentations created
    through scripts?

    DP: It may be easier to say what is included in this checkpoint
    rather than not included.

    IJ: So what types of presentations do we mean?

    DP: How are they generated?

    /* DB leaves the call */

    IJ: Note that there's an author responsibility: provide
    content you can get out without relying on time. But the UA
    could not, I think, in the general case automatically
    provide an equivalent for all presentations that involve 
    EH: Please look at my latest memo on the topic of applicability.

    No Resolution.

Open Action Items

   1. IJ: Talk to Judy about UA ftf.

Completed Action Items

   2.IJ: Propose two checkpoints to replace 4.16, they would have the 
         same techniques

   3.IJ: Summarize the discussion of conformance issue from 8/10/00
     telecon (Based on JG's proposal)

   5.HB: Review note for checkpoint 8.1

Dropped Action Items

    1.IJ: Draft a preliminary executive summary/mini-FAQ for developers. 
  (Will be done at a later date)

    4.GR: Re-examine the orientation checkpoints and see whether they
be clarified to account for control of rendering of audio (and possibly 
other content) on load.

Ian Jacobs (jacobs@w3.org)   http://www.w3.org/People/Jacobs
Tel:                         +1 831 457-2842
Cell:                        +1 917 450-8783
Received on Thursday, 17 August 2000 16:11:26 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:49:27 UTC