See also: IRC log
<fesch> https://www.w3.org/wiki/ARIA_roles_for_graphics
<scribe> scribenick: LJWatson
ABR: There is currently only one
role for SVG graphics, and it assumes all child components are
presentational.
... At the least we need to propose a role that represents
structured SVG images.
... My basic proposal was a "graphic" role, that could be
applied to the whole or part of a structured image.
<fesch> https://www.w3.org/wiki/ARIA_roles_for_graphics
ABR: I also suggested "figure" to
represent a figure in the sense of a technical document
figure.
... Another suggestion is the "icon" role, with dual roles as a
stand-alone icon in the page, and icons as symbols within a
chart or other complex SVG.
... Final suggestion was a "shape" role, to represent simple
geometric shapes.
... The "shape" role would need an attribute to enable the type
of shape to be conveyed.
<fesch> https://www.w3.org/wiki/ARIA_roles_for_graphics
<shepazu> (note that "figure" may be confused with HTML <figure> unless they have the same semantics
DS: Wouldn't the <title> be the way to indicate shape?
LW: Tend to agree with Doug. The <title> would be a good way to provide that information, at least as a first step.
FE: +1
ABR: Happy to go along with the
idea that it's up to the author to express what sort of shape
it is.
... If the author doesn't, should the browser be able to infer
a name based on the actual geometry though?
FE: Amelia, you're happy if we don't include a shape role right now?
RESOLUTION: We will not propose the shape role for ARIA 1.1.
ABR: We can't change the existing
img role. We can chip away at it by excluding certain things
from it.
... Specifically that children are presentational.
LW: If we do change the role's characteristics, do we need to indicate it's specifity to SVG?
DS: We're addressing different use cases.
LW: So we need a new role?
DS: Yes.
ABR: Key question is what we call it.
LW: A role called "graphic" seems too general.
ABR: Concern with including "img"
in the name, for example "structuredimg", is that it's
confusing.
... I want something flexible, that can cover different kinds
of structured graphic.
LW: I liked Chaals suggestion of "compoundImage".
ABR: My concern is whether people will recognise that it could be nested groups?
FE: Thinks a separate role for nested images would be good.
DS: Will people understand "compound image" in the real world?
FE: I don't think so.
LW: I think "compound" is familiar enough in the English language to be understood.
ABR: We could use "complex".
FE: But "complex" also means hard to understand.
LW: These images need not be complex either.
DS: I don't see why some of the
examples on the wiki couldn't just be an image?
... The house example... what is the benefit of defining it as
a compound?
ABR: The limitation that if you define it as an "img" you cannot explore its child components.
DS: Léonie, you say "graphic" is
too ambiguous?
... Isn't that the intention of this proposed role?
LW: To the average person a graphic and an image are the same thing.
FE: If we add more specific roles later, those would be specific, no?
LW: Yes, but that wouldn't help me understand the difference between a graphic and an image.
ABR: The problem is you need to understand the difference without reading the spec?
LW: Yes.
ABR: We need to indicate the explorability of the image.
FE: So "chart", "map"...
ABR: We need something more generic.
LW: If someone said something was a "compound image", what would you interpret it to mean?
FE: A collage or something artistic.
ABR: If you use complex markup to represent a graphic, but it only makes sense as a composite unit, you want a way to convey that.
JW: So let's do that with an attribute instead.
DS: Another problem is that we can't change image.
ABR: Right, it's already out there.
JW: Not concerned about that. It's never used correctly.
DS: There will be lots of
different opinions about that.
... Suggest we set this aside for now, have more discussion on
the mailing list.
+1
ABR: Need a way to identify a
graphic, and at the same time its position within the larger
document.
... Concerns are that it will get confused with HTML
figure.
LW: Is the conceptual idea the same as with HTML?
ABR: Yes, but the HTML
interpretation is broader.
... One option would be to match the HTML figure role.
FE: So it would be used for something like a plan view or an overhead view?
DS: No.
ABR: In HTML <figure> allows for non-graphical content.
DS: HTML treats figure as a point of reference... "See fig. 3" for example.
<AmeliaBR> The HTML5 figure: http://www.w3.org/TR/html5/grouping-content.html#the-figure-element
DS: Think ARIA should have this
role. It might not encompass everything we want to do with SVG,
but that's not the point.
... Agree with Léonie, we should leave the HTML meaning of
figure as the default meaning.
ABR reads figure definition from HTML spec.
DS: Right, all of which can be done in SVG. So the figure role "as is" is totally appropriate.
LW: +1
FE: Thought point was if you wanted to focus someone's view on a figure?
ABR: Mostly it's how some SVG relates to the larger document in which it sits.
FE: Would the role go on the <svg>?
ABR: Yes.
FE: What about inside an SVG?
ABR: If you had a multi-part graph for example, yes.
FE: Like a 3D technical drawing, you could use figures to indicate different views of the schematic?
ABR: Yes.
DS: Wouldn't try to yoke it to
the idea of a "view". A "figure" is more generic.
... It's a generic and long standing concept. We shouldn't try
to change it.
JW: What does it map to for the <figure> element?
LW: Believe it's group.
<AmeliaBR> http://www.w3.org/TR/html-aam-1.0/#html-element-role-mappings
RESOLUTION: We propose a figure role based on the HTML5 concept of figure.
<fesch> +1
ABR: Should anyone follow up with the group responsible for the HTML acc roles?
FE: We'll discuss later.
ABR: Key question is whether icon and symbol should be different roles?
DS: Think it's useful to have both.
+1
<fesch> +1
DS: An icon usually represents a point of interaction, or a stand-in for something else.
FE: It's atomic.
ABR: We could use img as the 1.0
fallback.
... We need to think about icons used in a widget role. Don't
think we could specify something was both a button and an icon
for example?
... There are other icon use cases though.
... Like a weather status for example.
FE: I'd call that a symbol.
ABR: If it was part of a set of icons in a web page, no.
LW: Does anyone disagree with having an icon role?
RESOLUTION: We will propose an icon role for ARIA 1.1.
DS: If it's atomic does it mean that the <title> and <desc> would not both be read?
FE: No, only that there are no child components to explore.
ABR: Similar to an icon, but
typically used within a more complex graphic - like symbols to
represent categories in a scatter graph.
... Is there agreement this is a useful role to have?
+1
DS: Yes. Want to share the difference between an icon and symbol...
<shepazu> http://www.differencebetween.com/difference-between-icon-and-vs-symbol/
DS: an icon looks like the thing it represents, a symbol does not.
ABR: So a symbol is more abstract.
DS: Think the key differentiation must be that a symbol is a reference to something else within a larger graphic.
ABR: When the distinction is based on context, what happens if the role is used outside of that context?
FE: We could limit the roles scope.
DS: Seems functional.
LW: I hesitate to add restrictions.
ABR: Yes, you could have a complete SVG UI that included icons for example.
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/respresents/represents/ Succeeded: s/RESOLUTION: We propose a figure role HTML5 concept of figure./RESOLUTION: We propose a figure role based on the HTML5 concept of figure./ Found ScribeNick: LJWatson Inferring Scribes: LJWatson Default Present: +1.512.286.aaaa Present: +1.512.286.aaaa Fred Esch Doug Schepers LJWatson Amelia BR Jason White Got date from IRC log name: 29 May 2015 Guessing minutes URL: http://www.w3.org/2015/05/29-svg-a11y-minutes.html People with action items:[End of scribe.perl diagnostic output]