SVG Working Group Teleconference

12 Nov 2015


See also: IRC log


nikos, ed, heycam, AmeliaBR, fesch, BogdanBrinza, Tav


<trackbot> Date: 12 November 2015

<AmeliaBR> WebEx quick-join URL: https://mit.webex.com/mit/e.php?MTID=m55af864d1884d38ad10291441f98b2dd

<scribe> scribenick: Nikos

<scribe> scribe: Nikos

<scribe> scribenick: nikos

Accessibility Task Force documents

<AmeliaBR> https://lists.w3.org/Archives/Public/www-svg/2015Nov/0022.html

AmeliaBR: There's two docs. One is an updated draft, the ARIA team is hoping to publish that next week
... other is first pass wd - won't be published for a few weeks but if we can sign off on it that would be good

<AmeliaBR> https://rawgit.com/w3c/aria/master/svg-aam/svg-aam.html

AmeliaBR: This is the SVG accessibility api mapping spec
... Purpose is to define how browsers should map svg specific features to the OS accessibility api
... that are then used by screen readers, special input devices, etc
... OS accessibility APIs have a standard way of describing everything and they have to work with the content of the web site as well
... there's a core accessibility API mapping guide that defines how the basic ARIA roles should map
... but that doesn't cover the unique features of a given langauge
... so there'll be a HTML mapping guide, covering form elements and other interactive elements
... then there's this guide which talks about SVG features
... Starts with introduction
... talks about how dom tree maps to simplified accessibility tree that the assisted technologies need to work with
... Important terms has a long list of terms that are common to all ARIA specs
... Then there's a section on keyboard navigation that references the new tabindex requirement from SVG 2
... shouldn't be too controversial - just basic tab index navigation
... Then we get into the specifics. I should mention the entire TOC of this document mirrors the TOC of the core mapping doc
... Many of the sections are very short and say follow core spec
... Where we need SVG specific roles they're described
... core spec has rules for how elements in general are exposed in this accessibility API
... and which elements are exposed
... general approach is to keep it as simple as possible
... not give unimportant info to assistive tools
... e.g. hidden content and things that don't have layout information
... that's where things get trick y with svg because we have a lot of content that isn't rendered directly on screen - filters, gradients, etc
... there's also much more complex rules for pointer events
... something can have pointer events even if invisible
... there's about 6 or 8 options for pointer events
... so we need some svg specific rules that say even if this element does not cause any pixels to change, if it reacts to pointer events then it is still perceivable to a user of a mouse who can see the end result of clicking, therefore it should be accessible to users of assisted tech
... there's a note explaining that, we need to do some work with the core group on making sure the defs in the core are general enough to account for these svg specifics
... other editors note is about how we handle desc element
... svg 1 spec talks about using css to make an alternate presentation of desc
... so you can display html instead of graphcs - but that doesn't work on any ua
... so we allow any html in desc but it's not going to display anywhere
... this is tricky because if something isn't drawn on the screen there's no way for someone to browse to it and read it in a structured way
... we still use the description, but treat it as plain text
... similar to an alt

heycam: is there a reason that can't work?

AmeliaBR: there are practical reasons in that it doesn't work now, there's also the more intentional reason that we don't want to encourage a screen reader experience that is disconnected from the visual experience
... having complex content that doesn't show up on the screen can be problematic and confusing
... could happen in future, we've talked about it in the TF - that's why we have a note
... was suggested that we could have html structured tooltips instead of plain text
... but the tech isn't there yet, and there isn't a framework for creating it

heycam: if the TF has specific suggestions on what should change in SVG 2, then it would be good to hear them

AmeliaBR: right now we want to make sure the text in the SVG 2 spec doesn't imply to authors that they can do things that won't have any effect
... so might want to add a note to the desc element
... so although it's allowed, it's not exposed currently to assistive tech
... I've had complaints from authors about why it doesn't work
... There are some issues on aria roles
... part of the second doc is trying to fix this
... First question is what to do with the use element
... right now we're mapping it as an image
... you can't access the internal content
... the only way we use the source graphic is as a name and description
... so we tell browsers to look at the source graphic and see if it has a title instead
... problem with that is if we end up in SVG 2 moving to a shadow dom based thing, where the contents of teh use element are a complete dom tree that scripts can interact with, then that needs to be reflected in the accessibility tree
... so depends where the svg 2 spec goes with use
... svg 1.1 had the element instance tree that could have conceivably be used, but wasn't implemented
... so there's a note, and we're trying to get feedback from users
... but we need a decision from SVG WG about how use elements are handled wrt shadow dom specs
... is there a desire to keep use elements simple and use custom elements for other things?

shepazu: Just want to say that we're requesting approval for updated publication of the AAM spec and publication of the other spec
... so we have two svg accessibility specs and they're joint deliverables wit the SVG and APA WG
... we need approval from both WGs to move them forward
... It's good that Amelia is giving you details on the spec, but we should also open the floor to questions

AmeliaBR: To be clear, these issues, we're planning to publish with them as notes in the document

heycam: that's fine, from what I can see there's not that many issues

shepazu: There is obviously a need for ongoing co-ordination between SVG and accessibility TF about the issues
... but I don't think these are show stoppers, think we could sort them out in the course of the next publication
... but it is something the svg wg will ultimately have to be responsible for and accept

heycam: So does anyone have any questions about publication of this first document?
... And is everyone in favour of publishing?

nikos: yes

shepazu: +1

<ed> +1

<AmeliaBR> RESOLUTION: SVG WG endorses publication of a new working draft of the SVG Accessibility API Mappings specification.

AmeliaBR: We'll try to publish that next week - so be quick if you have questions or concerns
... but it is just an updated WD which we'll continue working on over the next few months

ARIA Graphics module

<AmeliaBR> http://rawgit.com/w3c/aria/master/aria/graphics.html

AmeliaBR: one of the issues we came up with in the mapping guide is there's very few roles for graphics
... image was defined so that all child dom nodes would be ignored
... that's not useful for SVG where you want people to explore the sub components that might have their own titles and descriptions and event handlers
... so we need a more nuanced approach to graphics
... Long term goal is to create an ARIA model for describing charts and graphs
... where assistive tech can add extra understanding
... we're not there yet
... this is a basic set of roles for describing structured graphics
... where components have titles and description

<AmeliaBR> http://rawgit.com/w3c/aria/master/aria/graphics.html#roles

AmeliaBR: That section defines the new roles
... Graphics document is the default role for the svg element
... Difference from exiting image is that graphics-doc has meaningful child content

heycam: What would be the difference between the experience if you have inline SVG that does have graphics-doc and one that doesn't?

AmeliaBR: Assuming the alternative is what browsers currently do - they map to a grouping role or an existing graphics role that doesn't have an ARIA equivalent
... they wouldn't see a lot - in future, new tools might allow arrow key navigation instead of dom order navigation
... or other things, if you have a braille doc it could be aware you're dealing with graphics content

heycam: so it's an indication that there's 2d presented information that isn't some hierarchically ordered document?

AmeliaBR: the assumption is with a plain text doc that there's a single reading order
... with graphics that doesn't always work
... so the new role is a signal to tech that they can apply different heuristics
... based on 2d layout
... we're not defining what they would be yet, just enabling that

shepazu: So what we're defining is a framework for future work
... want to allow accessible visualisations and allow screen readers to explore them in novel ways

AmeliaBR: there'll either be separate domain specific specs or a level 2 of this document

heycam: so we have the graphics-doc role that says the whole document is an explorable graphic

AmeliaBR: graphics-object is an alternative to group
... it's adding semantics so distinctions between groups in a document can be described
... final role is graphics-symbol
... for things that are conceptually a categorical item - e.g. data points on scatter plot or astrological male and female symbols
... you don't want to delve inside these objects
... this is the role we will propose as the default role for basic shapes in SVG
... The idea is that these roles will become default roles for SVG. We haven't updated to the other spec to refer to this one yet as we won't be publishing this one until December. But there's notes currently.

ed: anyone opposed to publishing this document?

BogdanBrinza: no

RESOLUTION: SVG WG approves publication of WAI-ARIA Graphics Module draft

AmeliaBR: if any of you have time to have a look at these specs, and especially the editor's note
... we are looking for examples of use

new TR stylesheets


shepazu: You can see an example
... in 2016, all published specs will need to use these style sheets
... so we need to make sure our specs will work with this style

<shepazu> http://fantasai.inkedblade.net/style/design/w3c-restyle/2016/sample

shepazu: Have seen some problems when there's a large table or if there's a large image. Because the spec space is now narrower, that can be a problem

But starting in Jan we'll be publishing with these styles, so we might need to do some changes to the spec generation scripts

scribe: This is almost all CSS - there's very little change to the markup
... we haven't updated our style for 15 years. Future revisions may include script libraries and other things to improve it
... but for the moment, it's almost all just style sheet changes

AmeliaBR: Have you tried to put the style sheet on our current specs?

shepazu: not yet, but we should

AmeliaBR: maybe a branch on github

Tav: annotations aren't included in this

shepazu: Still working on that
... One of the points of this is that we want to start encouraging a common style for all W3C specs
... We'll try to find the best practices and apply them to all specs
... each group may still need variations of course

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.140 (CVS log)
$Date: 2015/11/12 21:23:56 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.140  of Date: 2014-11-06 18:16:30  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/graphics-doc is an alternative/graphics-object is an alternative/
Found ScribeNick: Nikos
Found Scribe: Nikos
Inferring ScribeNick: nikos
Found ScribeNick: nikos
Present: nikos ed heycam AmeliaBR fesch BogdanBrinza Tav
Agenda: https://lists.w3.org/Archives/Public/www-svg/2015Nov/0023.html
Found Date: 12 Nov 2015
Guessing minutes URL: http://www.w3.org/2015/11/12-svg-minutes.html
People with action items: 

[End of scribe.perl diagnostic output]