SVG A11y TF -- 16 Aug 2016

<scribe> scribenick: LJWatson

Round table

FE: Nothing to report.

News from SVG

BM: No news from SVG.

News from ARIA

FE: Trying to test and close ARIA 1.1.
... Don't think there is anything relevant to SVG specifically.

Publication status

FE: We'll release heartbeats when we get Michael's time.
... Think it will be after the dPub specs go out.

Testing

FE: Talked with Shane about automated testing for ARIA 11.
... We'll use the same format of testable statements that we have in SVG, then run a diff between it and ARIA 1.1
... We'll convert them with a Perl script into the format needed for the Web Driver test suites.
... We need to go through our wiki and make sure that the tests there are still current and valid, and also complete.
... Think Amelia was going to review this, but we should all do this.
... We need to check things relating to <use> and shadow DOM with care.
... but perhaps we could get volunteers to look at different sections.

LW: Should we try to divide it up now?

FE: Let's try it now, yes.

<fesch> https://www.w3.org/wiki/SVG_Accessibility/Testing/Test_Assertions_with_Tables

<fesch> https://www.w3.org/wiki/SVG_Accessibility/Testing/Test_Assertions/Language_Tables

FE: Test info for language/multilingual is linked from the above URL.

LW: So there are 8 groups of tests?

FE: Multilingual is further broken up, so there are 7 groups initially.

LW: When do we need to get this done?

FE: No timetable, but sooner the better.

LW: It's the AB meeting next week so Chaals and i will be regrets next Wednesday.

FE: Two weeks?

LW: WFM.

BM: I can look at the bottom three groups (role="presentation", role="none", and not in a11y tree), by next week.

LW: I'll take the acc name tests.

<scribe> ACTION: Brain to review role="presentation", role="non", and non-a11y tree tests due 2016-08-23 [recorded in http://www.w3.org/2016/08/16-svg-a11y-minutes.html#action01]

<trackbot> Error finding 'Brain'. You can review and register nicknames at <http://www.w3.org/WAI/PF/svg-a11y-tf/track/users>.

<scribe> ACTION: Brian to review role="presentation", role="non", and non-a11y tree tests due 2016-08-23 [recorded in http://www.w3.org/2016/08/16-svg-a11y-minutes.html#action02]

<trackbot> Created ACTION-2028 - Review role="presentation", role="non", and non-a11y tree tests due 2016-08-23 [on Brian McNeily - due 2016-08-23].

<scribe> ACTION: LJWatson to review tests for acc name due 2016-08-30 [recorded in http://www.w3.org/2016/08/16-svg-a11y-minutes.html#action03]

<trackbot> Error finding 'LJWatson'. You can review and register nicknames at <http://www.w3.org/WAI/PF/svg-a11y-tf/track/users>.

<scribe> ACTION: Leonie to review tests for acc name due 2016-08-30 [recorded in http://www.w3.org/2016/08/16-svg-a11y-minutes.html#action04]

<trackbot> Created ACTION-2029 - Review tests for acc name due 2016-08-30 [on Léonie Watson - due 2016-08-23].

FE: May need acc name tests for the <use> element.

<scribe> ACTION: Fred to review acc description and role tests due 2016-08-23 [recorded in http://www.w3.org/2016/08/16-svg-a11y-minutes.html#action05]

<trackbot> Created ACTION-2030 - Review acc description and role tests due 2016-08-23 [on Fred Esch - due 2016-08-23].

FE: If someone creates a new test it needs to use the same syntax as the other tests already in the wiki.
... Comments start with //
... Used to identify what's in the testable statement - in human readable form.
... By tradition the element being tested has id="test"
... testable statements are typically if x then y.
... The results table includes the different APIs, plus notes (for Joannie's use only).
... She uses the notes column for webkit tests only.

ABR: Makes sense to focus on tests other than the language ones, because those features are at risk in SVG.

Amelia's changes

<fesch> https://github.com/w3c/aria/pull/423

ABR: We've discussed the changes in this commit.
... Would like to be sure it meets expectations though.
... Most changes relate to <use> and the shadow tree.
... Have been specific about referring to "use element shadow tree", as opposed to generic "dhadow tree".

FE: Noted one instance where the generic version is still used.

ABR: Which doc?
... Make sure you're looking at the final changes.

FE: Ah. Ok.

ABR: Reference the normative definitions for shadow tree, shadow dom etc.

FE: Isn't there a problem with referencing the Shadow DOM spec?

ABR: That's why I referenced the DOM WHATWG spec.
... Have added more tests.
... for whether something is a link.

FE: As in has anhref?

ABR: Yes pretty much.
... Final change is the section on animations.
... Had an editors note to update SVG2 for the state and property attribs.
... Have cleaned up our spec to match.

FE: Are we ok to accept these changes?

<fesch> +1

FE: Everyone seems happy.

RESOLUTION: To accept Amelia's changes and merge #423 into master.

FE: Does the way <use> is used affect name computation?

ABR: Yes - now it's treated as a container. Same for symbol.
... Will need some more tests for when <use> is an accessible object etc.

FE: Same for description?

ABR: Yes.

Graphics text role

ABR: The text role was added to ARIA 1.1.
... Mainly used for things like icons or emojis, where having it announced as a graphic wasn't deemed necessary.
... SVG text converted to paths is another example.
... It was controversial on the ARIA side.
... People were concerned that role=2text" would be added inappropriately and the graphic info/structure would be lost.
... Also that an aria-label could override the actual text content.
... Also announcing the role as "text" would be disconcerting for screen reader users.
... After a lot of debate it has been removed from 1.1.
... Webkit already has support for it though, so it hascontinued to cause controversy.
... Think this role makes sense in a graphical setting.
... Where getting the right font/presentation is important for example.

LW: I am not keen on this role.
... Losing the knowledge that something is an image is of concern.

ABR: ATs could have different options perhaps - so in some cases the role/graphic would be announced.

BM: Think announcing the text role would be of concern.
... Otherwise think I see benefits and concerns.

FE: You could label the text content perhaps?

ABR: Currently you'd have aria-label (or <title>), but it would be announced as the acc name for the group/container.

LW: What is the use case?

FE: Something like an emoji like "black heart" for example

ABR: The existing solution is to use <span role="img" aria-label="love">emoji</span.

LW: So do we have evidence that this is a problem for users?

ABR: Don't think so for that particular use case.

LW: Worried this might be a solution in search of a problem.

ABR: For SVG with text it might be cnfusing for an image of a word, as opposed to an image of the thing the word describes.
... Having a way to distinguish betwen the two might be helpful.

FW: Or on a map a river might have squiggles.
... To label it you'd put a group around it and label it with role="text2.
... When things are broken into separate random bits it might be helpful.

ABR: Just wanted to throw this out there for thought.
... the discussion will continue in ARIA space.

FE: for SVG the concept of a label means something in structural graphics.
... Text I'm not sure has so much semantic meaning.
... give peole the option to choose text or label, think we'll have lost something.

ABR: That's a bigger worry for me.
... Labels are different to alt texts.
... In ARIA the two are often confused.

LW: ARIA tumps everythng in the acc name computation.
... We're also over the hour.

FE: Ok. Time to leave it here for today.

- DRAFT -

SVG A11y TF

16 Aug 2016

Attendees

Contents