Minutes, 16 February 2016, SVG Accessibility Task Force

Minutes for the teleconference are pasted below and attached as an HTML
file.

I'm very sorry but they are not also available in their correct place on
the W3C server. I missed the important first step of scriping, to tell
RRSAgent/trackbot to start recording the meeting. Very, very sorry. And not
only because I just spent 20 minutes formatting them in Microsoft Word.


~Amelia

_________________________________________________________________________________________

SVG Accessibility Task Force Teleconference

16 February 2016

Contents

TOPIC: Open Actions
<file:///C:/Users/Dell/Desktop/SVG-a11y-minutes-2016-02-16.htm#_Toc443388955>

TOPIC: Graphics Module AAM
<file:///C:/Users/Dell/Desktop/SVG-a11y-minutes-2016-02-16.htm#_Toc443388956>

Topic: SVG Spec Major issues
<file:///C:/Users/Dell/Desktop/SVG-a11y-minutes-2016-02-16.htm#_Toc443388957>

TOPIC: SVG spec clarifications, re empty attributes & empty title/desc
<file:///C:/Users/Dell/Desktop/SVG-a11y-minutes-2016-02-16.htm#_Toc443388958>

Summary
<file:///C:/Users/Dell/Desktop/SVG-a11y-minutes-2016-02-16.htm#_Toc443388959>

Action Items
<file:///C:/Users/Dell/Desktop/SVG-a11y-minutes-2016-02-16.htm#_Toc443388960>

Resolutions
<file:///C:/Users/Dell/Desktop/SVG-a11y-minutes-2016-02-16.htm#_Toc443388961>





chair: fred

scribe: AmeliaBR

present+ AmeliaBR

present+ fesch, MarkS, shepazu, richardschwerdtfeger


TOPIC: Open Actions

<fesch> https://www.w3.org/WAI/PF/svg-a11y-tf/track/

Fred: Most of current open actions belong to Amelia.

AmeliaBR: Most of these come down to editorial clean up & clarification in
SVG-AAM. We will need to discuss the issue of empty aria attributes,
separately.

<fesch> https://rawgit.com/w3c/aria/master/graphics-aam/graphics-aam.html
TOPIC: Graphics Module AAM

Rich: I've been talking with Joanie & will also have to talk with people
from Linux & other platforms, about what platform API roles could be used
for the new graphic roles.

Fred: Any other comments? Do people want a chance to look it over.

Amelia: Yes, need to read. Wasn't expecting a whole new spec, separate from
Graphics module, but that will help in publishing each spec when it's ready.

Fred: OK, we'll get back to it next week.
Topic: SVG Spec Major issues

Fred: In SVG spec, events and tabindex can be assigned to any element, even
if it can never be in the accessibility tree & doesn't make sense.   How
are we going to handle it? Just state that some things are never affected,
regardless?   People will want to know what to do if these receive events.

Amelia: Most of the elements will never actually receive user events,
because there is nothing for users to interact with.

Rich: Where in the spec does it say that?

Amelia: In SVG 2? No where explicitly. But the standard language for these
elements is that they "are never directly rendered", regardless of display
property.   So for SVG AAM, we should reference that "never directly
rendered" language as a general definition to create a special class of
elements that are never in accessibility tree, regardless of other
factors.   for SVG 2, we need to make sure this wording is used clearly &
consistently  That way, the definition in SVG AAM is more general &
extensible, versus giving an explicit, finite list of elements.

Fred: So both specs need to change?

Rich: Should this all be in SVG? Because we're really talking about
something fundamental about how SVG works?

Amelia: We'll need some changes in SVG-AAM, as part of my general editorial
clean-up.

Fred: We need something in SVG-AAM, so implementers have it all there &
don't need to look at the second spec.

Amelia: We should have a clear definition in SVG 2 of this class of
element, and reference it from SVG-AAM.

Rich: We already have a note for each element in the SVG-AAM mapping table,
so that's where it would be. Those elements are not rendered,
ever.   Except we need to change <mesh>, if that is rendered.

Amelia: Sorry for not catching that earlier. A mesh is rendered by default,
unlike the other gradients.

Rich: what role would it have?

Amelia: Should be the same as <path> because it would just appear as a
fully-painted shape.

Rich: Should I make that change or will you do it with other edits?

action: Rich to change mesh in AAM to map to image

* trackbot is creating a new ACTION.

<trackbot> Created ACTION-2012 - Change mesh in aam to map to image [on
Richard Schwerdtfeger - due 2016-02-23].

Amelia: If you want to push that change quickly, then I won't miss it.

Rich: In both SVG and SVG-AAM, we have a list of allowed roles per element.
Those don't yet reference the new Graphics Module.  Should I add those in?

Fred: Can we have a clear resolution that elements like gradients are never
in the accessibility tree? That's an important decision.

<richardschwerdtfeger>
https://svgwg.org/svg2-draft/struct.html#roleattribute

Amelia: It's not really a change to how we're handling the accessibility
tree, but important change to how we define it.

ACTION: Rich to update allowed roles in SVG-AAM & SVG 2 to include new
Graphics module roles

* trackbot is creating a new ACTION.

<trackbot> Created ACTION-2013 - Update allowed roles in svg-aam & svg 2 to
include new graphics module roles [on Richard Schwerdtfeger - due
2016-02-23].

* AmeliaBR draft RESOLUTION: (1) Clearly define in SVG 2 an extensible
category of elements that are never directly rendered & therefore cannot be
interactive. (2) Reference this category in SVG-AAM, to make it clear these
elements are never mapped to accessibility tree, regardless of event
handlers, tabindex, etc.

Rich: So you're an editor on SVG 2 and can make these changes?

Amelia: Yes, may need to discuss with SVG WG about where best to put it,
but I can start drafting both new texts at once.

RESOLUTION: (1) Clearly define in SVG 2 an extensible category of elements
that are never directly rendered & therefore cannot be interactive. (2)
Reference this category in SVG-AAM, to make it clear these elements are
never mapped to accessibility tree, regardless of event handlers, tabindex,
etc.
TOPIC: SVG spec clarifications, re empty attributes & empty title/desc

<richardschwerdtfeger>
https://svgwg.org/svg2-draft/struct.html#implicit-aria-semantics

Rich: In SVG-AAM, we say most elements won't get mapped unless they have
ARIA attributes. But we don't clearly define what happens if the attribute
is there, but empty. Does it not count?

Fred: Is it true that, if it's empty it doesn't count?

Rich: I think if it's empty, that's a mistake, and we should try to work
with that.

Fred: And since whitespace gets trimmed off, all whitespace has same effect
as empty.

Rich: But also, if after all different elements & attributes are considered
for the name computation, final result is empty, then that should be
treated as no name.   Maybe there should be a name for that, invalid name
computation.

Fred: So if the final computation returns an empty string, then it doesn't
count.

Amelia: This isn't just an SVG issue, it needs to be part of the whole name
& description computation.

ACTION: Rich to discuss empty name/description attributes & elements with
relevant spec editors

* trackbot is creating a new ACTION.

<trackbot> Created ACTION-2014 - Discuss empty name/description attributes
& elements with relevant spec editors [on Richard Schwerdtfeger - due
2016-02-23].

Fred: A related issue, is what to do with empty lang attributes.

Fred: And when you have multiple title/desc, with language attributes to
switch between them, but some of those are empty, do we include that
element in the accessibility tree or not?

Doug: This isn't something that's been tested in the wild much. I think we
can cover it with authoring practices.

Amelia: With <switch>, if the author hasn't made an explicit default,
nothing displays at all.

Doug: That's definitely not ideal, but is there for historical reasons.
Browsers also don't deal well when users have secondary language choices.

Fred: So, should we make a recommendation to SVG WG about how to clarify
this?

Doug: Yes, but it should be a recommendation to do something very simple.
We shouldn't be trying to guess what authors want. It's too complicated and
browsers won't implement it.

Fred: So if an element has an empty title/desc for the current language,
but could have a title/desc for another language, we're still going to
exclude it from the accessibility tree.

Amelia: I think that makes sense. This isn't going to be dynamically
changed. If user changes their language preference, they are effectively
re-loading the page.   SVG-AAM should only consider the title/desc element
that is selected based on the language rule.

Rich: Should we ask for clarification from SVG WG? What do browsers do?

Amelia: No one has implemented title language switching yet. But I expect
they'd probably just show an empty tooltip. It's not really an issue for
that, it's only an issue for us because we want to fall back to other title/
desccomputation steps.

Rich: If they're going to show an empty tooltip, we should do something
similar. I have no problem with returning error behavior for author errors.

Fred: Should we confirm with browsers that's how they'd deal with it?

Amelia: Do you want to write an email to www-svg?

Doug: I don't want to introduce special rules for this. This isn't even an
"error" just an authoring bad practice.

Amelia: Rich, you already have an action to follow up on empty title/desc;
language switching is an extra complication to that.

Fred: What about language unknown, as indicated by lang=""?

Amelia: What does it say re the language switching? That's all that matters
for us. Does it become the first-choice match, default fallback match, or
no match at all?

Fred: But when would that be used, would it even be inherited?

Amelia: It should. It would make sense for an entirely pictoral branch of
the tree. Maybe using an emoji as alt text or something, so that could be
read in multiple languages.
SummaryAction Items

·         ACTION-2012 - Change mesh in aam to map to image [on Richard
Schwerdtfeger - due 2016-02-23].

·         ACTION-2013 - Update allowed roles in svg-aam & svg 2 to include
new graphics module roles [on Richard Schwerdtfeger - due 2016-02-23].

·         ACTION-2014 - Discuss empty name/description attributes &
elements with relevant spec editors [on Richard Schwerdtfeger - due
2016-02-23].
Resolutions

·         RESOLUTION: (1) Clearly define in SVG 2 an extensible category of
elements that are never directly rendered & therefore cannot be
interactive. (2) Reference this category in SVG-AAM, to make it clear these
elements are never mapped to accessibility tree, regardless of event
handlers, tabindex, etc.

Received on Tuesday, 16 February 2016 19:39:13 UTC