W3C home > Mailing lists > Public > public-html-a11y@w3.org > February 2011

Minutes: HTML-A11Y Task Force on 17 February at 16:00Z

From: Martin Kliehm <w3c@kliehm.com>
Date: Thu, 17 Feb 2011 18:36:49 +0100
Message-ID: <4D5D5CB1.3010501@kliehm.com>
To: HTML Accessibility Task Force <public-html-a11y@w3.org>


HTML Accessibility Task Force Teleconference
17 Feb 2011



     Cynthia_Shelly, Gregory_Rosmaita, Janina_Sajka, John_Foliot, 
Léonie_Watson, Michael_Cooper, Mike_Smith, Rich_Schwerdtfeger, 
Martin_Kliehm, Paul_Cotton
     Marco_Ranon, Laura_Carlson


* Topics
     1. Actions Review
     2. Is ARIA Normative for HTML5?
     3. Issue-30 longdesc--Updates?
     4. ISSUE-122

TOPIC: Actions Review

MC: Action-22 Testing sub-team is moving forward in particular in the 
ARIA field

<oedipus> issue-134?

<trackbot> ISSUE-134 does not exist

<oedipus> GJR: keep action open -- will complete today and alert list

<MichaelC> action-101 due 3 March

<trackbot> ACTION-101 - email ARIA Caucus to request a "modal" attribute 
for ARIA due date now 3 March

<janina> http://www.w3.org/Bugs/Public/show_bug.cgi?id=11892

<oedipus> action-104?

<trackbot> ACTION-104 -- Frank Olivier to post question to HTML working 
group on whether to limit HTML elements in non-visible content such as 
the Canvas fallback content -- due 2011-02-14 -- OPEN

<trackbot> http://www.w3.org/WAI/PF/HTML/track/actions/104

TOPIC: Is ARIA Normative for HTML5?

RS/JS: Would inclusion of ARIA in HTML spec inhibit the extensibility of 
the ARIA spec?

RS: I think they should be normative.

<oedipus> propose that we de-couple ARIA qua ARIA from EVERY markup 
language -- the concepts have to sync so that A11y APIs know what to do

<Zakim> oedipus, you wanted to say that HTML5 annotations for a11y 
content should be based on ARIA concepts, but not be bound to a specific 
release of ARIA

MS: Integrating things into HTML requires a binding to the HTML spec. If 
we take SVG and MathML, which are kind of an exception because they also 
have elements, the SVG group had concerns that SVG would only work in 
browsers, but not seperately in an SVG player. Though the player could 
interpret HTML5. ARIA could take a similar approach.

<JF> +1 to oedipus

Gregory: I propose that we decouple ARIA from every languageso that it 
stays extensible. ARIA will become better and better and avoids the trap 
of being stuck in one particular spec.

MS: That would be the ideal solution, the question is whether it's possible.
... There's a formal objection that we need to have the ARIA attributes 
defined, so there's a limit what you can decouple.

<oedipus> personally, i would MUCH rather have the annotations for a11y 
content section removed and repopulated throughout the spec where 
individual features are introduced/defined

<oedipus> agree about @role -- definitely MUST be native to HTML5

RS: There are examples like media queries that are fairly separate, so 
it works. The lexical part could be normative, but others need to be 
separate because they will continue to be developed.
... ARIA supports the accessibility APIs, but doesn't have any impact on 
the standard function. So it's separate.

MS: I don't quite understand the design decisions behind the ARIA spec. 
I had a similar problem with MathML where we had an element and asked 
ourselves how it would be interpreted by ARIA. The answer was that by 
design it was undefined so that other specs would be able to expand.

RS: If aria-grab=false indicates it's grabable, but false means it can't 
be grabbed. If we didn't define it in ARIA, because of the design of 
HTML5 attributes it would be impossible to set it to false. The 
processing should be part of the host language, but it needs to be 
... Anything else like custom strings is not an issue because you could 
just define name-value pairs.

<MikeSmith> http://dev.w3.org/html5/spec/elements.html#global-attributes


<oedipus> http://dev.w3.org/html5/spec/elements.html#global-attributes

<MikeSmith> "Authors may use the ARIA role and aria-* attributes on HTML 

<MikeSmith> "The following attributes are common to and may be specified 
on all HTML elements (even those not defined in this specification):"

MS: If you take Ian's question in the bug report it asks where the 
attributes should be defined.

RS: If external specs conflict with the host language like in the 
example above that needs to be solved. That's a lexical processing. That 
should allow ARIA to expand. If we introduce new ARIA roles in ARIA 2.0 
we could define a "detail" role to reflect the new element in HTML5.

MS: I understand what Ian is asking, I'll respond.

<oedipus> should i file a bug against the spec to get 
into the spec?

<oedipus> Laura's Longdesc Reinstation Change Proposal: 

<oedipus> Verbose Descriptor Requirements: 

<oedipus> HTML WG Bug 10853 - HTML5 lacks a verbose description 
mechanism: http://www.w3.org/Bugs/Public/show_bug.cgi?id=10853

<oedipus> timeline to last call: 

<paulc> See 

TOPIC: Issue-30 longdesc--Updates?

PC: Sam Ruby has talked to Laura. The problem is the deadline was 
January, so the chance of re-opening was missed.

<Zakim> oedipus, you wanted to say this has been an "ISSUE" since the 
chairs' decision last year

JF: There's a formal objection regarding longdesc. So before we clear 
last call this needs to be addressed.

<JF> +q

MS: Confirming there's a formal objection.

<Zakim> oedipus, you wanted to day don't want to have to file formal 
objection -- would rather work with the WG on getting the lack of 
verbose descriptor addressed rather than push off to

JF: It's up to the chairs to decide about the procedure. In seeking 
consensus I think addressing the issue now is easier.

GR: We tried to avoid a formal objection. The discussion has been going 
on, a solution is at hand.

<JF> +1 to0 Janina's point

<paulc> There are already Last Call issues that are on the page that 
lists all issues:

<paulc> http://dev.w3.org/html5/status/issue-status.html

<JF> +q

JS: In the past when a solution was at hand I understood that time 
constraints were secondary.
... It's on the WAI agenda for the next meeting as well.

<oedipus> it isn't a matter of "principle" it is a matter of a HOLE in 
the spec that has been repeatedly pointed out and has been the subject 
of Change Proposals

<oedipus> this decision was draconian to remove longdesc without 
equivalent or replacement

<paulc> See 

PC: The Chairs have to be fairly principal because of the April/May time 
table. We will have time constraints soon, so we don't want this to be a 
precedent. We need to be fairly strict in the future.

<oedipus> we have followed the "lead" of the chairs and staff members in 
addressing this issue -- it is not like others

GR: I disagree because it's not a matter of principal, but it's a hole 
in the spec. It has been in HTML4 and has been removed. Something has to 
be done to resolve that.

JS: The problem is a political problem. The issue has not been rejected, 
it's been postponed.

<JF> +q

<oedipus> paul, if you go ahead with last call without LONGDESC you KNOW 
you are going to get Formal Objections -- why not nip this in the bud by 
addressing an actual need (articulated by WAI, HTML A11y TF, and EPUB/IDPF)

<oedipus> that's just semantic jive

PC: If Laura or other contributors were to send the material they have 
today we believe we have enough evidence to re-open the issue. But it 
would be a post-last call issue.

<oedipus> semantic jive

<JF> +q

PC: It's not unusual to draw the attention on sections of the spec in 
Last Call so that they can be improved.

JS: The message being sent by publishing HTML5 without a prior solution 
for the longdesc issue would be devastating.

<oedipus> timeline to last call: 

<JF> http://lists.w3.org/Archives/Public/public-html/2010Sep/0074.html

JF: The escalation after the deadline has happened in the form of the 
formal objection.

<oedipus> there is an issue, a bug, and change proposals

<oedipus> paulc, the point is that we have tried in good faith to avoid 
a formal objection by fixing what is broken (the lack of a verbose 
description mechanism)

PC: I have discussed this with Sam whether the deadline was met because 
of the issue being escalated before January 27, and we denied that. 
Though we haven't spoken with Maciej yet.

<oedipus> we have tried to avoid formal objections on the ADVICE and 
COUNSEL of the chairs

<oedipus> http://www.w3.org/2005/10/Process-20051014/

<JF> W3C Process: 

PC: I assume we can address that issue long before Last Call, it could 
be as early as June.

<oedipus> the earlier the better...

JS: I believe we need to explain the process in the preamble so that we 
do not alienate accessibility people.

<oedipus> verbose descriptor mechanism also needs to be addressed for 
EPUB 3.0 in order for it to be compatible with HTML5

JF: There are two formal abjections...

JS: But that doesn't mean they will be addressed by the director before 
Last Call.

<oedipus> qck oe

<Zakim> oedipus, you wanted to ask about the "single" chance to change 
the chairs' decision

JF. We've been announcing this for a while, so it could have been on the 
radar that there is a proposal in the making. It has been explained in 
mails that even though the deadline has passed it could be reconsidered.

<oedipus> paulc was the first on the bridge

<oedipus> MikeSmith, you've been the one reminding us of a "single" 
chance to have chairs' decision reopened -- can you clarify?

PC: In the W3C culture there's no "single chance." The process is if new 
compelling evidence is put on the table issues can be re-opened or 
amended anytime. This has happened many times in the past.

MS: I agree with Paul. There's nothing in the process that limits a 
proposal to one shot.

<oedipus> we DO NOT WANT to have to file a formal objection -- we are 
trying to work with the WG and the chairs and the editor

MS: Also the director can push it back to the WG.

<oedipus> this was a BAD decision, taken too hastily

<JF> +q

MS: I support whatever the chairs decide because I trust them to make 
decisions in good faith. We need to keep a positive and constructive dialog.

JF: It's frustrating to find the deadline of January 27 blocking a 
solution at this time because it hasn't been communicated clearly.

<oedipus> we've been trying to "advance" this issue under the guidance 
of the chairs and facilitators -- the decision made by the chairs was 
faulty because it removed something added to HTML4 SPECIFICALLY to 
address a need without recognizing that a hole was thereby created in 
HTML that must be fixed by an equivalent or superior mechanism

<oedipus> that is a bad decision, regardless of one's opinion of "policy"

MS: When the chairs choose to allow issues to be re-opened and 
re-escalated to the group it's their decision. The chairs will continue 
to be under pressure to meet the Last Call deadline, so we need to 
respect the decision.

GR: This is different because longdesc has been added to HTML4 by 
request of the director and has been arbitrarily removed by the editor.

MS: The technical issue is out of question, it's a procedural question.

JF: The decision is going to anger many people and I'd preferred to draw 
it back.

<oedipus> we dont want to wait 2 years -- we've already waited too long 
to address these issues and their importance

JS: Do not over-estimate the power of a formal objection. It doesn't 
mean that the issue will be addressed soon, it could be in two years. 
We're not talking about going to CR yet.

<oedipus> plus 1 to JF

In my personal opinion it won't be the only Last Call anyway.

<oedipus> closed with EXTREME prejudice

<paulc> Agree, I expect as 2 Last Calls

JF: [repeating the above]

<paulc> Agree, I expect at least 2 Last Calls

<oedipus> HTML5 is being implemented by fiat -- that is the source of 

JS: If we rely on a formal objection, why do we discuss about it today?
... Going to CR is far in the future.

JF: The longdesc issue was raised in 2008. The chairs told us the issue 
could be re-opened. A lot of work and effort has been going into it. Now 
we're being told it's too late. That's disappointing.


<oedipus> Verbose Descriptor Requirements: 

<oedipus> Laura's Longdesc Reinstation Change Proposal: 

<oedipus> HTML WG Bug 10853 - HTML5 lacks a verbose description 
mechanism: http://www.w3.org/Bugs/Public/show_bug.cgi?id=10853

PC: The chairs didn't say when the issue would be re-opened. It will be 
addressed. I admit the January date could have been communicated better. 
I understand your frustration. When we are going to re-open the issue we 
still want the Accessibility TF to spend their resources on the other, 
pre-Last Call issues.

JS: The problem is how it will be interpreted in the accessibility 
community. I don't think that it's good to postpone the issue, so I 
would like to emphasize the requirement to communicate the basis for the 
decision clearly.

<Zakim> oedipus, you wanted to ask if we can talk a bit about ISSUE-122 
before we part?



<oedipus> At the HTML WG's TPAC 2010 Face2Face meeting, agreement was 
reached that the first two paragraphs of the definition of img should be:

<oedipus> sentence 1: An img element represents an image.

<oedipus> sentence 2: The image given by the src attribute is the 
embedded content; the value of the alt attribute provides equivalent 
content for those who cannot process images or who have image loading 

GR. There are two issues. At TPAC we decided to start with the smaller 
issue, changing the wording above.

<JF> "We stated that the issue can be reopened if one or more of these 
conditions are met."

<JF> Sam Ruby - Jan 6, 2001

PC: If there are more concerns I'm open to discuss longdesc in personal 
Received on Thursday, 17 February 2011 17:37:34 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:05:18 UTC