W3C

- DRAFT -

Accessibility Guidelines Working Group Teleconference

02 May 2017

See also: IRC log

Attendees

Present
AWK, Jake, ChrisLoiselle, Melanie_Philipp, Wilco, Rachael, JF, LisaSeeman, allanj, jasonjgw, Bruce_Bailey, marcjohlic, Greg_Lowney, Lauriat, MikeGower, kirkwood, Laura, Kathy, KimD, Katie_Haritos-Shea, shwetank, steverep, Pietro, Davidmacdonald
Regrets
EA_Draffan, Crystal_Jones, AlexLi, Stein_Erik_Skotkjerra, Denis_Boudreau, Mike_Elledge, Jim_Smith
Chair
AWK
Scribe
ChrisLoiselle

Contents


<AWK> +AWK

AWK, do you want me to set myself as scribe now? Do you need any other pre-meeting setup?

<scribe> Scribe: ChrisLoiselle

TPAC reminder

<allanj> will there be phone bridge available for those not able to attend?

AWK: Reminder to book hotel in Burlingame, CA.

<Bruce_Bailey> http://www.w3.org/2017/11/TPAC/

<Zakim> Bruce_Bailey, you wanted to say I am thrilled to be going to TPAC

Bruce B: Friday the day for working group?

<JF> CAn somebody re-post the survey URL? Thanks.

AWK mentions that it should read correctly on page for TPAC

Hotel is booked by phone, should be mentioning "CSAIL"

<AWK> CSAIL

<Ryladog> "CSAIL"

<Bruce_Bailey> https://www.w3.org/2002/09/wbs/35422/AGWG_TPAC2017/results

<Bruce_Bailey> my mistake, Friday is listed

Single Key Shortcut Alternative: https://www.w3.org/2002/09/wbs/35422/Top3_18Apr2017/results#xsc

AWK: asks about single key shortcut concerns

<AWK> Current version: https://rawgit.com/w3c/wcag21/single-key-shortcuts_ISSUE-69/guidelines/sc/21/single-key-shortcuts.html

Jason W: concerns are still there. Conversations need to be had with speech technology manufacturers

Speech technology research and development needs to be looked into , if not by WCAG, by public possibly

JF: Is there an example of re-mapping keyboard shortcuts in the wild? Is it implementable today?

Kathy: Gmail does this already, re-mapping is occurring on their apps

<Ryladog> Kathy we cant hear you as someone else is talking

<Greg> As I said on Thursday, I disagree this is a user agent issue: not at all. This is also not an Assistive Technology issue. Jason, this only puts new requirements on those webapp developers who explicitly, really want to use single-key shortcuts, like F. Doing so is significantly less accessible than using combinations like Ctrl+F, which has been the industry standard for decades.

<Zakim> Greg, you wanted to say that I do not agree with Jason's interpretation that this requires changes to most pages, nor do I agree this is the responsibility of assistive technology

AT vendor relationship can be talked to by Kim

<david-macdonald> https://rawgit.com/w3c/wcag21/single-key-shortcuts_ISSUE-69/guidelines/sc/21/single-key-shortcuts.html

<Ryladog> is anybody else hearing two calls?

Greg L: differences are based on interpretation of SC by people within our group

<Greg> Also, this is not just speech, but all AT that simulates keyboard input, and also for users who have trouble telling whether the keyboard focus is still where they think it is.

Greg L: it is a fairly important requirement who aren't using AT...

David M: not access keys related. What SC is requiring that there is a button to deactivate or activate shortcut keys

David M: Supports doing the due diligence, but understandable it may not be fixable with Dragon

Lisa S: open source available to re-write shortcuts

Shawn L: keyboard accessibility related to 2.1.1 , you can turn it off and there is keyboard access to this functionality

<Zakim> AWK, you wanted to clarify what "a mechanism" can mean

<gowerm> Note: Where the single-character keystroke is turned off, the requirments of 2.1.1 continue to be met

<laura> Mechanism definition: https://www.w3.org/TR/2008/REC-WCAG20-20081211/#mechanismdef

AWK: mechanism doesn't need to be created by page author. if browser did this work, it would be less work for developers. author can not use single key shortcuts...

<Zakim> Greg, you wanted to clarify that even today's accesskey is not single-key activation, because user agents take the key specified by the author and add a modifier such as Alt or Cmd

<Greg> Note that today's accesskey does not even support single-key activation, because every user agents I know of take the key specified by the author and add a modifier such as Alt or Cmd. We could add language to except cases where the user agent makes it single-key when the author did not intend that.

<Greg> (Note that Accesskey was originally added not for single-key activation but to emulate the Common User Access input standard when authoring dialog boxes in HTML. CUA combined access keys with Alt or prefixed them with F10, except when in special modes.)

<Zakim> JF, you wanted to suggest that this is more tightly scoped to *only* developer-scripted shortcuts

JF: Follows up to Greg and David...scope to only keyboard shortcuts that are custom scripted vs. any single key shortcut (user agent) vs. author scripting a widget

<gowerm> just add "without modifier keys"

<david-macdonald> "If <a>single-character key shortcuts</a> are implemented by the web page to activate a control,..."

Kathy: access keys need modifier vs. shortcut keys

<Ryladog> custom behaviors

<gowerm> Just address in the glossary

David M: implemented vs. scripted , implemented by author

JF: main concern is around access key...

Jason White: Speech tech and AT getting involved with this would possibly interfere with keyboard functionality. Mechanism approach may be best bet

Editorial note on what issues are could be useful

James N: Two points. Survey comment: Never seen this as an actual issue with voice recognition. Text area example. Speech recognition character settings need to be fixed. Unless you turn these commands off in settings...can't find an issue

<Greg> ...one non-printing key."

2nd point: take into account multiple key access issues. "G" to go and then a character to go to an element. I.e. unmodified access keys.

Mike ?: keyboards taken over by different apps...thinks there is use for this SC. Talks to modifier key use and internationalization mapping of keys, keys applied in a series,

<laura> G suite has single keyboard shortcuts: https://support.google.com/a/users/answer/163225?hl=en

<Zakim> Ryladog, you wanted to say identify it as 'custom scripted behaviors'

Katie H-S: could you repeat your first point? Second is custom scripted behavior

<Zakim> Bruce_Bailey, you wanted to suggest to DM latest edit to "set by the content" rather than "implemented by the webpage"

<JF> +1 to Custom Scripted behavior

<Zakim> Greg, you wanted to say that the proposed wording is pretty good but the final "two or more characters" should be "include at least one non-character key", as Ctrl+F is fine, as is

Bruce B: keystroke vs. character in terms of words used.

<Greg> I pretty much agree with everything Mike said.

<Greg> James, when you tell tell Kim she does not have evidence that this is a real-world problem, please be aware that you’re telling a person who uses speech input full time and creates commercial tools to support speech users, that what she says is a problem for herself and her users is not a problem.

<Greg> The proposed wording is pretty good but the final "two or more characters" should be "include at least one non-character key", as Ctrl+F is fine, as is PgDn. I would actually prefer "If a key shortcut using only printing characters is implemented by the web page to activate a control, then a mechanism is available to turn them off or remap them to a shortcut that uses at least one non-printing key

<Greg> ."

Greg L: to James, Kim does have ability to talk to the points James made on use case

Greg L: talks to use of the word "Printing key"in terminolgoy

RESOLUTION: Leave Open

Timeouts: https://www.w3.org/2002/09/wbs/35422/minimize-user-errors-13/results

<Greg> "If a keyboard shortcut uses only printing characters is implemented by the web page to activate a control, then a mechanism is available to turn it off or remap it to a shortcut that using at least one non-printing key."

<Bruce_Bailey> s /terminolgoy/terminology/

<Ryladog> My first point was; in my experience it is areal world problem,. I first discovered it when first learning to use voice recognition and trying to say-in a password

<AWK> Current version: https://rawgit.com/w3c/wcag21/minimize-user-errors_ISSUE-13/guidelines/#minimize-user-errors

<Bruce_Bailey> s /areal world/a real world/

James N: address field issue vs. choosing from a list. As long as information is validated

James N: validated to user , picking thing from a list could make it more difficult...

Lisa S: validated by enough is not enough. People may give up if whole form is incorrect and red errors are everywhere. Errors corrected automatically...

James N: type of value vs. pick from a list of integers. Narrow use case.

Lisa S: how to not make mistakes is main point. Don't want people to leave the task

perhaps adding "where it can't be reliably automatically correctly" into the text

James N: the experience could be worse by introducing this, so that is the concern.

Lisa S: COGA users vs. non-COGA users and experience of lists vs. text edit fields..what choice is easier?

<Bruce_Bailey> i think my proposed edit solves this delima

<allanj> it depends on the person. ALL is ALL. for some it is easy to type some information. for others picking from a list is easy. everyone is different

Jason W: we will have to defend details if it makes to into WCAG 2.1. What constitutes a correct / valid value? What is invalid vs. incorrect? Assessing of user's knowledge.

Mike: list box and keyboard interaction , for numbers it works.

<Zakim> Bruce_Bailey, you wanted to ask how pick lists were required?

<Zakim> Greg, you wanted to say that a drop-down combo-box is the best UI, because it supports both free-form input and choice from a list. I see online advice on how to implement them in

Bruce B: use a list box when you can example, 40 is way too much in terms of list items

<Greg> I agree that selecting from a list can be good or bad, depending on context and user. Drop-down combo-box is the best UI, because it supports both free-form input and choice from a list. I see online advice on how to implement them in HTML without requiring scripting.

<Bruce_Bailey> I still like the gentle edit (to an early version) that I put into the survey:

<Bruce_Bailey> Error Correction: If an input error is automatically detected and the correction is known, then the correction is automatically made, unless it would jeopardize the security or purpose of the content.

Lisa S: moving the number to 31 is the right thing to do.

<Greg> I'm a little concerned about the implication that dots (periods) have to be allowed as separators, when they're also used in numbers themselves.

<Bruce_Bailey> I disagree w/ Lisa that what I proposed is not testable. Here is 3.3.3:

<Bruce_Bailey> 3.3.3 Error Suggestion: If an input error is automatically detected and suggestions for correction are known, then the suggestions are provided to the user, unless it would jeopardize the security or purpose of the content

Lisa S: introducing the text that "free form text edit is not prohibited " could be introduced in the SC

<Zakim> Greg, you wanted to say I'm a little concerned about the implication that dots (periods) have to be allowed as separators, when they're also used in numbers themselves.

<Greg> I'm a little concerned about the implication that dots (periods) have to be allowed as separators, when they're also used in numbers themselves.

<Greg> Also we could add a note to the draft saying that the exact number may be adjusted and asking for feedback on that specifically.

<gowerm> +1 for the second paragraph being problematic. I made a note of that in my comments.

Greg L: dash, dot , numerical input. i.e. what happens with fractions? "." is the separator ...is there an exception for decimal points?

<allanj> and internationalization. in some european languages the comma is the decimal point

<Zakim> steverep, you wanted to comment about keyboard accessibility and other unintended consequences (e.g. splitting fields to avoid the SC altogether)

Marc J: expanding to selected number, is ok. But should this be lane, drive, way? how would this work for all forms of input?

Stephen R: full-time screen reader, auto-complete doesn't always work with custom controls.

Also, loop hole : combining fields.

RESOLUTION: Leave Open

AWK: Links at top, have current version

on Github

AWK: SC managers, please add those links to top of page

<Bruce_Bailey> The older bullet-list version is so much more readable than the current version.

<AWK> Current version: https://rawgit.com/w3c/wcag21/timeouts_ISSUE-14/guidelines/#timeouts

Lisa S: Use case: few items in tax return, press next to go to next page. then you go away for a bit. Logged out, data is lost.

<JF> Have we solicited feedback from the Security Folks at W3C around this?

<Bruce_Bailey> The current version is a long run-on compound sentence.

24 hours is reasonable time for this

Lisa S: not trying to cover everything, but be able to cover this and make an impact. Educating people on what they need to do.

AWK: Goes over James N's points from timeouts guideline

James N: airline shopping, don't want my current info saved , I do want to abandon that data. Give user info that in 24 hours, this ticket may not be available, etc.

<gowerm> Where data can be lost due to timeouts, users are warned at the start of a process about the length of inactivity that generates the timeout.

Mike: on phone , name online is gowerm?

<gowerm> Yes

thanks.

<AWK> from the SC current version: "unless the user can return to the same point in a task without data loss for a minimum of a 24 hours."

<marcjohlic> I think the 24 hours (while making more sense) will cause issues with 2.2.1.. would rather not include a time limit on this one

Mike: read the information, understands what users need to do. Take the timeout, it fits single A.

AWK: Advanced notification of timeout, then 2.2.1. would kick in. Mike and AWK, perhaps AA vs. A.

<marcjohlic> +1 to mgowers wording

<Bruce_Bailey> +1 to MG edit

<Ryladog> +1

<Rachael> +1

<AWK> +1

<LisaSeeman> greg and me prefer the old wording

<JF> +1

AWK: Providing advanced information is key aspect

<Greg> I was at https://rawgit.com/w3c/wcag21/timeouts_ISSUE-14/guidelines/#timeouts; is that not correct?

Greg L : more concerned on current version of text is current version. AWK provides link to group

<Bruce_Bailey> I think this is current version:

<Bruce_Bailey> Where a session can time out and form information that has been submitted by the user can be lost before the task is complete, the user is informed of the length of time that the data is preserved and the length of inactivity that generates the time out, at the start of the task, unless the user can return to the same point in a task without data loss for a minimum of a 24 hours.

AWK link was part of survey.

<Greg> Is this the correct text? "Where a session can time out and form information that has been submitted by the user can be lost before the task is complete, the user is informed of the length of time that the data is preserved and the length of inactivity that generates the time out, at the start of the task, unless the user can return to the same point in a task without data loss for a minimum...

<Greg> ...of a 24 hours."

<Bruce_Bailey> MG: Where data can be lost due to timeouts, users are warned at the start of a process about the length of inactivity that generates the timeout.

<alastairc> q

Mike: Data retention vs. timeouts

Lisa S: 20 hours vs. 24 hours

<david-macdonald> Where data can be lost due to timeouts more than 24 hours, users are warned at the start of a process about the length of inactivity that generates the timeout.

<Ryladog> +1 to Lisa's thought that 24 hours is better

<marcjohlic> 24 is a better number but it will confuse matters with 2.2.1

<david-macdonald> Where data can be lost due to timeouts that are less than 24 hours, users are warned at the start of a process about the length of inactivity that generates the timeout.

<gowerm> +1 to David's

<alastairc> +1

<Bruce_Bailey> +1 to DM edit

<gowerm> you'll also need langauge around "intentainl" timeouts. obviously a dev can't do anything about network timeouts

<david-macdonald> 20 hours in timing adjustable came from the idea that everyone needs at least 4 hours sleep in 24 hours

<Greg> David, make that "about the length of time and/or inactivity" that generates the timeout.

trackbot, end meeting

Summary of Action Items

Summary of Resolutions

  1. Leave Open
  2. Leave Open
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.152 (CVS log)
$Date: 2017/05/02 16:33:32 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.152  of Date: 2017/02/06 11:04:15  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00)

Succeeded: s/uses/using/
Succeeded: s/"Minimize user errors: https://www.w3.org/2002/09/wbs/35422/Timeouts_Issue14/results"/"Timeouts: https://www.w3.org/2002/09/wbs/35422/Timeouts_Issue14/results"/
Default Present: AWK, Jake, ChrisLoiselle, Melanie_Philipp, Wilco, Rachael, JF, LisaSeeman, allanj, jasonjgw, Bruce_Bailey, marcjohlic, Greg_Lowney, Lauriat, MikeGower, kirkwood, Laura, Kathy, KimD, Katie_Haritos-Shea, shwetank, steverep, Pietro, Davidmacdonald

WARNING: Replacing previous Present list. (Old list: (no, one))
Use 'Present+ ... ' if you meant to add people without replacing the list,
such as: <dbooth> Present+ AWK

Present: AWK Jake ChrisLoiselle Melanie_Philipp Wilco Rachael JF LisaSeeman allanj jasonjgw Bruce_Bailey marcjohlic Greg_Lowney Lauriat MikeGower kirkwood Laura Kathy KimD Katie_Haritos-Shea shwetank steverep Pietro Davidmacdonald
Regrets: EA_Draffan Crystal_Jones AlexLi Stein_Erik_Skotkjerra Denis_Boudreau Mike_Elledge Jim_Smith
Found Scribe: ChrisLoiselle
Inferring ScribeNick: ChrisLoiselle
Found Date: 02 May 2017
Guessing minutes URL: http://www.w3.org/2017/05/02-ag-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.


[End of scribe.perl diagnostic output]