See also: IRC log
<scribe> scribe: clapierre
title: Accessibility DPUB Task Force Weekly Meeting
1. Discuss Jean’s edits to the Note which have now been merged with the main branch which can be found here: http://w3c.github.io/dpub-accessibility/
2. Update on edits Charles was to make in re-oranizing various sections of the Note, as well as changes to more generic GitHub repositories instead of specific email addresses for the various contact information for some of the sections.
3. Address the GitHub Issues currently filed against the Note
span and word breaks #12
The section on drop caps implies that AT sees spans as word breaks. I've been discussing this with some members of the CSSWG, and they were very surprised by this behavior. Why does AT do this? Span has no intrinsic meaning, and visual renderers do not supply space around spans.
Deborah, I know that spans around first character … that is still true.
Mia, people put things in spans (link/button) then screen reader will say this is a descrete something in this span and spearate it out on a different line.
George, if you have emphasis text it iwll break it up.
Mia, I just saw that. Th and at on a separate line.
George, this is just handled in css and not needed to put spans around this.
Mia, there are key commands to go next span/div for JAWS.
George, we made a decision to drop dropcap since CSS has a solution and if AT hasn't caught up with this. we are agree.
Deborah, I thought we agreed that this is an education.
Tzviya, we agreed to take it out.
Deborah, there are is a browser and screen reader task to reach out after this note.. maybe our own todo list of action items.
Tzviya, then the action here is to go back to CSS and AT reads them as chunking . ARIA and HTML and spans have no semantic value. I didn't know that JAWS will chunk them. that is a NEWS FLASH. So our response to dave that they can be used as navigational elements.
George: i would bet that if its dealt with CSS then AT wouldn't pay any attention to it. SO maybe authoring ?
Tzviya: this is not our problem
Deborah: just response in Issue that CSS group needs to know how screen readers work with spans. I will respond to him.
Section 2.9 on semantic list heads has this paragraph before and after the example:
It is tempting to use HTML5 h1 – h6 elements but these elements may contain the ARIA role="heading" attribute, offering unwanted default navigation. A potentially useful option would be to include a role value as part of an element called "listtitle" (or something similar). The option to include a semantic title in a list requires support in visual as well as aural presentations.
What does the first sentence mean, though? The "heading" role is the implied semantic for h1-h6.
Also, what does "requies support in visual ... presentations" mean? Is it something beyond CSS?
Tzviya: SO just delete the duplicated paragraph.
George: I thought I saw two identical table of contents.
Charles: we need to clarify: Also, what does "requies support in visual ... presentations" mean? Is it something beyond CSS?
Tzviya: I think we can delete
that sentence.
... , we call out ARIA WG.
Deborah, right we were told not to do that.
Tzviya, I don't think we want aria-labelby exposed to everyone. Its proposing a specific solution, which may be a problem.
Deborah, Yeah we should take out the suggestion.
Tzviya, I have to leave. and will be out all next week.
Deborah, Sure I will take that out.
no mention of ssml? #15
Section 2.7 is pretty hard on PLS lexicons, dismissing it in one sentence:
The Pronunciation Lexicon (PLS) exists, but it is exceptionally difficult to use and doesn't scale well to a large publisher's use cases.
George: This sounds like Tzviya. so I am not sure about.
Deborah, I wrote this.
George, I know folks in Netherlands wrote a PLS lexicon editor. and would prompt user for phonetic pronunciation. building a large lexicon to put in all their books. Not sure if its a tools or specs issue.
Deborah, could be just a limitation of knowledge we had. SSML we didn't even talk about. maybe we should say is maybe it doesn't belong in the note, not a W3C issue, their aren't tools or a standard .
George: is PLS a W3C standard?
Deborah: yes it is. in the absense of tools, its not something a publisher will use. it really is a tool gap.
George: we got it built into the DAISY pipline when it generates mp3's but I understand that publishers are not using that tool. But it isn't implemented in a reading systems. that are generating prerecorded audio. don't know where its used in a reading system that AT would use.
Deborah: if its a W3C standard does it belong in our Note to say that there are no tools that support it.
Mia: we need to know more about the tools that can be used. Not sure how to say that.
George: blindness community this is a common issue. soon as TTS is used for main-stream this issue will be front an center. People will demand correct pronunciation.
Mia: its an a11y that will impact mainstream. more and more.
George: call for implementation for authoring tools and RS to handle this.
Mia: implementing is hard because of lack of tools or do tools exist and are reasonable to use on large content.
Deborah: there are multiple problems, someone using IPA is a skilled intellectual job. and time consuming. unless there are tools to make it easier people won't use it. so we agree this is a growing problem. so question is is there anything the W3C can do anything about
George: yes I agree. Correcting Obama's pronunciation has to do once. then a tool that would pass throught the lexicon and put it in. the problem then becomes the proper nouns that may need corrections in your book.
Mia: that is going to be an issue by language and by dialect. US / UK english etc.
Deborah: the larger question is a
shared Lexicon and tools outside of a11y space…. Point of View:
if something doesn't have a measurable ROI then we won't do it.
These are the publishers who don't have usable tools.
... , Is this a problem for the W3C?
George: No W3C does not implement
tools.
... , if PLS is a W3C recomendation there should be two
implementations of this. It just hasn't been important to
people in the past.
... , NOA radio has fixed this problem
... , I think we bring this to peoples attension.. we have math
in our note and is a lack of consistency.
Deborah, I will leave it in the note and will rephrase it.
ARIA does have heading guidance #16
I'm not saying the guidance is clear and easy to find, but there is an example of creating an "h7" in the ARIA techniques:
https://www.w3.org/WAI/GL/WCAG20-TECHS/ARIA12.html
What it might be good to point out is conflicts between this model, h1-h6 (can you or can't you change the level of an heading element to bump it down?), and the document outline (if you use div+aria-heading, would that affect future implementation of the algorithm, or is the algorithm dead?).
George: this is on reference
materials and deep nesting.
... , not sure about this issue, section that could nest. 1-6
headings is all fine. but once you pass heading level 6 how do
you show proper structure and text. DAISY suggests that after
that you flatten you structure at that point. But that is a
produciton issue that we made, and not something we put in a
standard. Maybe we can use the ARIA technic to go beyond
heading level 7 And I doubt AT would work.
... we could use the aria techniques and flatten it past
heading level 6. You use ARIa techniques for higher level
elements but put it under H6… which would it put it in the DOM
but would announce. So pressing 6 would give you all H6's but
would at least be accessible by AT be it not perfect.
... we are just pointing out the problem we don't have to solve
it.
... , once you get past heading level 3 that many libraries
flatten past 3.
... , but that is a production decision
... , maybe this is really resolved in these techniques but I
will check it out and provide details.
I saw this sentence and wasn't sure exactly what was being asked:
Further, the concept of skippability, in general, should not be confined to users of AT.
It was repeated again for escapability.
The first time I read it I thought the implication that these features must be available natively in browsers, but I'm assuming that the intent is that it is practical for anyone using SMIL playback, which isn't specifically an assistive technology. If so, it would be good to give more context, such as specifically mentioning the read-aloud feature of ebooks.
George: escapability is the function of AT, complex table and with JAWS you have a key sequence to get to the end of the table.
Deborah: Screen Readers have a way to get out of an element but you can't do that without AT.
George: this is main-stream to get out to the next block element. You can't get to the next paragraph in ibooks . but VS made it work somehow. used Rotar to go to next container.
Charles: maybe we just need to add more examples for on small device screen sizes. etc.
Mia: for skipability sure I can write that up.
Charles: so for escapability could we use the same thing?
Deborah: I think we can.
Mia: next week I am travelling, and then won't be around for over a month after that.
George: I am traveling multiple days in may.
Charles: so we won't change the meeting times for now and will look at this again once the Note is finished.
This is scribe.perl Revision: 1.144 of Date: 2015/11/17 08:39:34 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Found Scribe: clapierre Inferring ScribeNick: clapierre Present: Deborah_Kaplan Tzviya Charles George Mia WARNING: No meeting title found! You should specify the meeting title like this: <dbooth> Meeting: Weekly Baking Club Meeting Got date from IRC log name: 08 Apr 2016 Guessing minutes URL: http://www.w3.org/2016/04/08-DPUB-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]