W3C

- DRAFT -

SVG a11y TF

13 Jul 2016

See also: IRC log

Attendees

Present
chaals, Léonie, Brian, Rich, rich_schwerdtfeger
Regrets
Chair
Fred
Scribe
chaals

Contents


<scribe> Scribe: chaals

FE: We're in a holding pattern waiting for ARIA 1.1 - need feedback.
... Rich will send a CfC tomorrow for ARIA, to produce heartbeats, and a FPWD for the graphics AAM
... I have been trying to recruit people for this group.

RS: Did you ask Elsevier? Oh wait, they aren't W3C members any more.

FE: Also the Diagram Center, at @@tec. I had hoped Jason would champion instructional diagrams but he has been too busy to attend the meetings.
... if you know someone who has an interest, please recruit.

<fesch> s/@@tech/Benetech/

FE: we still have some open action items - please look through them and update / close / do them.

LJW: Followed up with David at Graphical Web conf, and will be there presenting SVG.

[I am presenting SVG accessibility at Accessing Higher Ground later this year]

New time

FE: Doug has a conflict - can people do either Monday or Tuesday?

[I have conflicts on Monday]

LJW: Works for Me.

FE: Let's switch to Tuesdays at the same time

RESOLUTION: SVG a11y TF will have teleconfs on Tuesdays at 2pm Boston time, effective now

post ARIA 1.1 work

CMN: Increased effort on authoring guidance.

FE: Low vision / colour blindness / navigation - to they all seem like reasonable things to work on?
... trying to bring in commercial vendors so people learn to do things the same way, and don't have to switch between different techniques.
... Anyone want to think about affordances in charts?
... anything we should start from?

LJW: What's the difference between that and a role?

FE: An affordance is visible as a rule.

LJW: Isn't role an equivalent for that?

CMN: Think that roles are clearly equivalent to the visual affordances like zoom and pan controls…
... Fred, what do you want to do with this?

FE: want to make it clear that people can move around a diagram to get tooltips, popups, etc - if you are a keyboard user how do you know that there are interactive parts of this.

RS: You can provide information about the chart.
... zoom is a good affordance. Something that is visual, should also be navigable. Imagine navigating along a line to examine the data points. If there is a tooltip how do people know that?
... you can't just put everything on the screen, because it gets in the way.
... we could create best practices for that.

FE: Something to think about. Instructional diagrams are a case for thinking about it.
... How do we handle a more robust taxonomy? Do we need to have separate kinds of graphics or put them all together?

RS: There are different kinds of images - drawings, maps, flowcharts, charts and graphs…
... we may have some abstract roles that can be used in graphics, and then we allow people to create taxonomies around that.
... If the WG is a bottleneck to doing stuff, that's bad.

<Rich> ack

RS: we need to allow people to move faster, if graphics were extensible that would be great.

CMN: Start with a concrete approach to developing stuff and seeing if it is accessible, rather than starting with making taxonomies. Making taxonomies is going to lead to lots of unused and complicated work.

FE: How hard is it for a user to transition from one ??? to another?

RS: It's hard.
... the user needs to get to the original functionality.

LJW: Biggest worry about role descripition is that developers will mistake it for a way to do role extensibility.

RS: Role-description is a new piece of UIA, and in apple platform.
... UIA operates from control patterns.
... "I'm clickable, draggable", and they add localised role-description to it

<LJWatson> @shepazu WFM

RS: Freedom Scientific doesn't look at that. You need to get back to the role type.

FE: When it comes up in ARIA, and people want to describe a pie chart, how do you do that?

RS: Will you have a taxonomy for pizza?

FE: I would have it in the title somewhere...

RS: But now you mix labels with types. That was a mistake MSAA made…

CMN: Work with concrete examples and see if you can do it already, rather than inventing roles that may or may not be useful.

FE: Where do we get examples of instructional diagrams. Assembly diagrams, explanatory diagrams, …

LJW: We need to make sure we keep roles as things that help know what you can *do* with something, not just using them to add labels

FE: So is role about interactivity, or what?

RS: It is the type of the object.

CMN: Roles identify the type of *interactive* object…

RS: We will have to deal with interactive or not down the path…
... we have to say what this is for. There are structures like landmarks or tables.
... if you want to identify something on the screen that behaves in some way you need to be able to provide an association for the role-description.

FE: knowing whether something is a compound object or not, if you can put interactive things inside each other…
... can a tree leaf have a button in it?

RS: We clarified that

FE: Don't think we ended with clarity

RS: If you use ordered lists to make a tree and make things presentational, you get problems.
... In ARIA 2 we want to talk about device independence, and talk to the TAG about this. I think we are eaerly for this - we have a problem where things are not device independent at the moment, but bundling interaction patterns would be good.
... example: you have a keyboard interface, so you have to trap cvarious keys. But you may want to add voice control, and gestures.
... keyboard implementation varies across browsers.
... UIA lets you say that things are scrollable, etc. You could apply that approach to an SVG drawing.

FE: Yes abstracting out the interaction would save a lot of code.

RS: If you're moving a slider, you need to know what the increments are, so there needs to be a way to expose that back to the system.
... have web components looked at anything like that?

LJW: not that I know

RS: If you had data that could be put into a page, e.g. for a range control, you could give commands to the application in a high-level manner. We didn't do that in IndieUI and I think it was a problem.

FE: We did this in a project we wrote in Java and then transpiled. You could bind the commands to anything.

1pm tuesdays?

RS: not sure

LJW: WfM

CMN: WfM

RS: cannot do that time.

FE: We'll stick with the resolution above.

<fesch> rrsagent make minutes

Summary of Action Items

Summary of Resolutions

  1. SVG a11y TF will have teleconfs on Tuesdays at 2pm Boston time, effective now
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.144 (CVS log)
$Date: 2016/07/13 19:00:23 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
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)

FAILED: s/@@tech/Benetech/
Succeeded: s/affordances/affordances like zoom and pan controls…/
Succeeded: s/LJW/LJW:/
Found Scribe: chaals
Inferring ScribeNick: chaals
Present: chaals Léonie Brian Rich rich_schwerdtfeger
Got date from IRC log name: 13 Jul 2016
Guessing minutes URL: http://www.w3.org/2016/07/13-svg-a11y-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]