- From: Joseph Scheuhammer <clown@alum.mit.edu>
- Date: Thu, 23 Apr 2015 14:10:01 -0400
- To: W3C WAI Protocols & Formats <public-pfwg@w3.org>
Link: https://www.w3.org/2015/04/23-aria-minutes.html
Plain text follows:
[1]W3C
[1] http://www.w3.org/
- DRAFT -
Protocols and Formats Working Group Teleconference
23 Apr 2015
See also: [2]IRC log
[2] http://www.w3.org/2015/04/23-aria-irc
Attendees
Present
janina, Michael_Cooper, Tzviya, Joanmarie_Diggs,
+1.703.978.aaaa, Rich_Schwerdtfeger, Joseph_Scheuhammer,
Fred_Esch, James_Nurthen
Regrets
Chair
Rich Schwerdtfeger
Scribe
clown
Contents
* [3]Topics
1. [4]dpub aria module
2. [5]The "table' branch.
* [6]Summary of Action Items
__________________________________________________________
<trackbot> Date: 23 April 2015
<jongund> going to be a few minutes late
<richardschwerdtfeger>
[7]https://lists.w3.org/Archives/Public/public-pfwg/2015Apr/015
6.html
[7] https://lists.w3.org/Archives/Public/public-pfwg/2015Apr/0156.html
can you hear me?
<richardschwerdtfeger> scribenick: clown
dpub aria module
<richardschwerdtfeger>
[8]http://w3c.github.io/aria/aria/dpub.html
[8] http://w3c.github.io/aria/aria/dpub.html
RS: The dpub working group think it is ready for publication.
... Trying to arrange a joint meeting next week.
JS: <discussion of times>
RS: How about the Wed PF meeting? A lot of the PF members will
be there.
JS: Let me see James Craig's availability, from his email.
RS: <checks his availability>
TS: <checks dpub availbility>
<janina> James Craig says: I can do 10am Pacific (1pm Boston)
on either 27 Monday or 28 Tuesday. I can also do 11am Pacific
(2pm Boston) on 29 Wednesday.
<general discussion of when various people are available>
JS: The other request is to log issues.
... People are doing this, which is good.
... It's being done in github.
... Hopefully, the filing of issues will get them into the
draft, and will satisfy people such that we can publish a
draft.
RS: Yes, if we flag it in the spec, that's a good way to deal
with these issues.
... I would like to talk a little bit about the "abstract" role
vs. "abstract roles".
... It's going to be authoring tools that put these in.
TS: Hopefully. Not yet, but that is the goal.
... There is no confusion in the publishing world about what
the "abstract" role is.
RS: I've been on hundreds of ARIA consulting calls, and I have
yet to encounter a problem with abstract roles.
... But if the primary place for this role is publications, I
don't see this as an issue.
... Also, Ivan raise the issue that OWL doesn't have an idea of
an abstract class.
... But, it's a common idea in programming.
<janina> [9]https://en.wikipedia.org/wiki/Abstract_type
[9] https://en.wikipedia.org/wiki/Abstract_type
JS: Sometimes these are called "existential classes".
<laughter>
<fesch> base class
<richardschwerdtfeger>
[10]http://dictionary.reverso.net/english-synonyms/abstract%20c
lass
[10] http://dictionary.reverso.net/english-synonyms/abstract%20class
RS: digital publishing is a big deal, and we want to do our
best for them.
... How about "indefinite" role.
MK: I kind of fall on the side that there will be confusion
between "abstract" role and "abstract roles" is low.
... go forward with it as it is, and look for evidence of
confusion.
... If there is confusion, addresss it in the future.
FE: What about the landmark abstract role? Doesn't that cause
similar confusion?
MK: I don't think so.
RS: I am more than comfortable with that.
... Can you put out a call for consensus on that, Janina?
JS: Sure.
JN: I see it all the time where people try to use abstract
roles, and finding they don't do anything.
... I would like the remove all the abstract roles to avoid the
confusion.
RS: Perhaps that is the solution: keep them in the model, but
remove them from the spec.
TS: When I first read the spec, I was confused by them.
MK: If you remove them, then you can't use them for modelling
since if the superclass is an abstract role — do you break the
chain in the spec?
... The superclass is important for getting the characteristics
of the current role.
JN: I don't think that stuff is important as an author.
... We are not the people who are targettied by the spec. It's
target audience are authors and editors.
RS: Is there a way to conditionally hide content, i.e. via
ReSpec?
MC: Shane proposed something similar, but not sure how to pull
it off.
JS: We could have different views for different audiences.
MK: What do you do with the links to abstract roles in the
characteristics tables?
<tzviya> see the treatment of JSON-LD and TTL in web
annotations:
[11]http://www.w3.org/TR/2014/WD-annotation-model-20141211/
[11] http://www.w3.org/TR/2014/WD-annotation-model-20141211/
<fesch> laughs
MC: we could have different versions of the spec for different
audiences.
TS: I put a link in for the web annotations, where they have
different views for the same content.
<discussion of Web Annotation Data Model>
RS: This means take out the superclass role, subclass roles,
and related concepts.
JES: I would leave in the related concepts, since it provides
an "example" of the aria role.
RS: What about the taxonomy drawing.
MC: We're not changing the taxonomy, just how it appears in the
main spec.
RS: I think that's what we need to do.
... If we do that, is that better for developers, James?
JN: Yes, definitely. The stuff they are not meant to see, they
should not see.
<janina> [12]http://www.w3.org/TR/wcag2ict/
[12] http://www.w3.org/TR/wcag2ict/
MK: We wouldn't show super classes, sub classes, etc. in the
characteristics table, but allow it to be viewable for others?
<janina> The above as an example of what might be done
MC: It could be an expanded version. I need to look at the pub
rules if they are relevant.
... The specifics need to be sorted out.
JS: I put in a link to the ITC report to show what might
happen.
RS: I think this is an editors' issue.
MC: Start with me, and I will enlist Shane's help.
RS: A concern: if an author clicks this link, and it pulls up a
bunch stuff, is that a problem?
JN: Yes.
RS: I think we need to get rid of it.
MC: We may need to put out warnings to indicate who this stuff
is for.
RS: People looking at created taxonomies would look at that
stuff. But authors need not.
MC: We need to determine which version is the "main" version.
s/with verision is the/which version is the/
<jongund> Sorry I am late
<richardschwerdtfeger> ACTION: Michael Work with Shane to
develop publication strategy and solution that would exclude
abstract roles for authors [recorded in
[13]http://www.w3.org/2015/04/23-aria-minutes.html#action01]
[13] http://www.w3.org/2015/04/23-aria-minutes.html#action01]
<trackbot> Created ACTION-1624 - Work with shane to develop
publication strategy and solution that would exclude abstract
roles for authors [on Michael Cooper - due 2015-04-30].
MC: This is not going to be done by the time we publish.
JS: right, but it will be mentioned in the CFC
MC: So the next version will have the abstract roles, with a
note that we will be hiding them in the future.
JN: Should we add clarification to the "abstract" role in the
dpub spec to indicate it is not an "abstract role".
MC: Yes.
JS: The CFC will go thru by Wed noon.
FE: In the dpub document, they talk about the landmarks role —
are they going to take that out?
TS: I think you are looking at an earlier draft.
RS: It's been removed, so we're good.
FE: I"m looking at one from 9 Apr.
<tzviya> [14]https://rawgit.com/w3c/aria/master/aria/dpub.html
[14] https://rawgit.com/w3c/aria/master/aria/dpub.html
MC: There are various drafts: there is the github.io which is
behind rawgit.
<discussion of charters>
MK: There are requirements about landmarks, but there are a lot
of specific landmark roles in the dpub document.
... If we strip out the "abstract" classes, we will loose the
information that a specific role isa landmark.
JN: At the top the spec where we group roles in terms of
landmarks, widgets, etc. We need to bring that down closer to
the roles themselves.
RS: This is going to be a re-factoring of the document.
<Zakim> Joseph_Scheuhammer, you wanted to suggest a solutions
to Matt's question.
<jnurthen> +1 to bringing inheritence down
JES: Matt had a question about loosing the link to the
superclass role.
<richardschwerdtfeger> +1
JES: I suggest we bring the inheritance information down into
the characteristics so you don't have to click the link to get
at it.
MC: Some of that inheritance information is already presented
locally.
TS: I was going to suggest for dpub to, after the name of the
role, we go state the "superclass". Like "landmark role".
JS: I would find that very useful.
MC: Generally that's good. But there may be some roles where
that might not make sense.
... For example, widget roles.
The "table' branch.
JD: which table branch?
... To review: first we did the properties.
... Then we started talking about table roles.
... So, there are a number of branches.
... Whether we have agreement on the table roles, the
properties still apply to grid.
... So they are separated out.
<joanie>
[15]https://rawgit.com/w3c/aria/table-properties-only/aria/aria
.html
[15] https://rawgit.com/w3c/aria/table-properties-only/aria/aria.html
JD: This is the properties branch.
... If we can reach consensus on these (we almost were
previsously), we can finalize these and then turn to table
roles.
RS: Is this where the grid cell came out?
... there was one thing about rowcount and where it belongs.
... "We want ATs to not have to go back to the containing
Grid/table. For older IE browsers ATs only access content from
the DOM"
... let me add the text.
<richardschwerdtfeger> - Why aria-rowcount was not confined to
the table/grid:
<richardschwerdtfeger>
[16]http://www.w3.org/2015/03/19-aria-minutes.html
[16] http://www.w3.org/2015/03/19-aria-minutes.html
<richardschwerdtfeger> - Summary: We want ATs to not have to go
back to the containing
<richardschwerdtfeger> Grid/table. For older IE browsers ATs
only access content from the DOM
<richardschwerdtfeger> - Treegrids expand rows increasing row
count. ... Again would need to
<richardschwerdtfeger> go back to the container
<richardschwerdtfeger> - The net, net: If we don't want to
support older IE versions (US
<richardschwerdtfeger> Federal, banking, state governments)
then we can limit supporting the row
<richardschwerdtfeger> count to the containing grid/table
<richardschwerdtfeger> It will be up to the user agent to
propagate the row count down to
<richardschwerdtfeger> either the cell or row level. Group
needs to decide. Will need to put ATVs
<richardschwerdtfeger> on notice - Go to AAPIs or performance
<richardschwerdtfeger> will be impacted. Also, will there be an
event notification of
<richardschwerdtfeger> changes in row count?
<joanie>
[17]https://lists.w3.org/Archives/Public/public-pfwg/2015Apr/01
40.html
[17] https://lists.w3.org/Archives/Public/public-pfwg/2015Apr/0140.html
RS: rowcount on cells, or on the grid itself. Which is it?
JD: Alex had concerns.
<joanie> So we have two statements
<joanie> * the browser has to expose aria-rowcount some way
<joanie> * the AT has to support it
<joanie> Then some AT don't want to invest much into
aria-rowcount implementation,
<joanie> and thus they want to see it on a row, and the
proposal is to obligate the
<joanie> author to duplicate aria-rowcount on rows. Is this
observation correct? If
<joanie> it is then I'd say it's more reasonable to have either
the browser or the
<joanie> AT doing that rather than the author. My preference
would be if those old
<joanie> AT would care about new attribute support themselves.
JD: Makes sense:, but does the author do it, or the browser/AT.
RS: We need to make a fundamental decision.
... There are a lot of ATs that only access stuff from the DOM.
... But, CSS allows content to be injected via styles.
... The DOM based ATs miss that.
... Do we continue to coddle these ATs?
JN: If they are going throught the DOM, they can still get it.
RS: It's a performance issue to have to walk the DOM to get at
ti.
JS: Let's just say that ii will continue to be in the DOM for
the time being, the things are changing.
RS: I'm for telling them to move off the DOM
... I think that the only way forward is via AAPIs.
... You agree Cynthia.
CS: Yes, and we have made that pretty clear to AT vendors.
RS: Let's have a resolution on this. Let me propose something.
<richardschwerdtfeger> Proposal: We tell AT vendors that if
they need to access ARIA content it will need to be through
platform accessibility apIs and we are moving away from DOM
considerations
RS: Does that sound reasonable?
+1
JS: works for me.
<richardschwerdtfeger> +1
MK: We is PF?
<joanie> +1
RS: ARIA task force.
CS: we could say the ARIA task force and the industry.
<mattking> +1
RS: Sounds like we have group consensus.
<richardschwerdtfeger> RESOLUTION: We tell AT vendors that if
they need to access ARIA content it will need to be through
platform accessibility apIs and we are moving away from DOM
considerations
JS: I want to flag this for UAG.
<joanie>
[18]https://lists.w3.org/Archives/Public/public-pfwg/2015Apr/00
33.html
[18] https://lists.w3.org/Archives/Public/public-pfwg/2015Apr/0033.html
RS: Getting back to the properties, is the rowcount only on the
grid?
<joanie>
[19]https://lists.w3.org/Archives/Public/public-pfwg/2015Apr/00
37.html
[19] https://lists.w3.org/Archives/Public/public-pfwg/2015Apr/0037.html
JD: I propose rowcount/colcount only apply to tables/grids.
<richardschwerdtfeger> +1
JD: If we all agree to this, I can make the change to the
branch, and we can finish with this.
MK: How can you have rows with different number of columns.
RS: You can have something that spans columns, so there are
fewer columns for that row.
JD: It should give out the span information on a cell, but the
row/col counts for the entire table.
MK: JAWS and NVDA will tell you that you've landed on a row,
reading across, with fewer columns.
... you aren't going to understand why you abruptly at the end
of the row.
JD: It should give you the cell index, and that it spans, say
10 columns.
s/an it spans/and it spans/
MK: You get an index…
JD: There is a colindex and a rowindex.
MK: And, also a colspan?
JD: Yes.
RS: I think this is okay.
<joanie>
[20]https://lists.w3.org/Archives/Public/public-pfwg/2015Apr/00
37.html
[20] https://lists.w3.org/Archives/Public/public-pfwg/2015Apr/0037.html
JN: I think it's okay, but I need to read it closely.
JD: make the row/colcounts apply to grids (the table like
container), and push it to master?
JN: Sounds right for just the counts.
+1
<richardschwerdtfeger> RESOLUTION: aria-rowcount and
aria-colcount will only apply to role “grid”
+1
<joanie> +1
<richardschwerdtfeger> +1
<richardschwerdtfeger> RESOLUTION: aria-rowcount and
aria-colcount will only apply to role “grid” at this time
RS: grids are dynamic and interactive, but tables are just
static.
... Why can't I just re-use this?
... When focus lands on a gridcell, it's important to know that
I have keyboard navigation for the gird.
JN: But you can tell that from looking at the parent.
RS: So you want ATs to walk the a11y tree to find out the
context>
<Stefan> +q
JN: Isn't the purpose of the spec to make things easier for
authors?
RS: Authors and AT vendors.
MK: You would still add a cell role, but it's gridcell in a
grid, but equivalent to a <td> in a table.
JN: Yes.
RS: Still, you make the AT go up the parent chain.
MK: that's okay.
RS: I think that's a lot to ask.
JD: We could propogate an "interactive" property throughout the
grid.
RS: Actually, we have a new property that means "interactive".
<joanie>
[21]http://www.w3.org/2015/04/02-aria-minutes.html#item02
[21] http://www.w3.org/2015/04/02-aria-minutes.html#item02
RS: If the cell is inside a grid, does that propogate in the
a11y mappings down to cell itself.
... How about that Joseph?
JES: If there is a "isInteractive" property, sure.
... That could be propogated.
<Stefan> -q
MK: Is on this table discussion, was there consensus that we
would add a table role and cell role, but if they were
interactive that would be "grid" and "cell" by implication?
RS: Yes.
MK: And it's UA issue to make this implication in the mappings.
RS: You could also have a grid that is not interactive, but it
would have an implied role of "table".
MK: That might enhance the confusion around read-only on grids.
... the issue is called "managesFocus" or "managedFocus"
JN: Caution: sometimes are rows are interactive or
non-interactive on their own.
MK: isn't that a property of a cell?
JN: Think of a mail client where the rows are interactive, but
not the individual cells.
RS: how about we make this an override?
JN: Yes, that should be possible.
MK: no problem with looking at a page in the virtual buffer
when not interacting with it.
JN: Agree, but getting the page to scroll at the right time is
still an issue.
MK: Doesn't that property have to exist in the ARIA spec?
<joanie> issue-633
<trackbot> issue-633 -- listbox and tree may contain only
static items; badly need interactive widgets that can contain
interactive typed items -- raised
<trackbot> [22]https://www.w3.org/WAI/PF/Group/track/issues/633
[22] https://www.w3.org/WAI/PF/Group/track/issues/633
<richardschwerdtfeger> ACTION: Joseph add interactive property
to Core Mappings that propogates from grid or table to all grid
ARIA grid or table structural elements. The interactive
property should have same mapping as defined in Issue 633
[recorded in
[23]http://www.w3.org/2015/04/23-aria-minutes.html#action02]
[23] http://www.w3.org/2015/04/23-aria-minutes.html#action02]
<trackbot> Created ACTION-1625 - Add interactive property to
core mappings that propogates from grid or table to all grid
aria grid or table structural elements. the interactive
property should have same mapping as defined in issue 633 [on
Joseph Scheuhammer - due 2015-04-30].
action-1625?
<trackbot> action-1625 -- Joseph Scheuhammer to Add interactive
property to core mappings that propogates from grid or table to
all grid aria grid or table structural elements. the
interactive property should have same mapping as defined in
issue 633 -- due 2015-04-30 -- OPEN
<trackbot>
[24]https://www.w3.org/WAI/PF/Group/track/actions/1625
[24] https://www.w3.org/WAI/PF/Group/track/actions/1625
RS: It sounds like you can finish with cell and table roles and
your edits.
JD: Do we need a new cell role if we have an isInteractive
property?
RS: Yes, because of backward compatibility issues.
JD: The property defaults to "true" for backward compatibility.
RS: Any objection to Joanie adding the states and properties in
grid.
... Hearing none, let's make a resolution.
<richardschwerdtfeger> RESOLUTION: Add aria-rowcount,
aria-colcount, aria-rowindex, aria-colindex, aria-rowspan,
aria-colspan to grid structural elements
RS: can you close the apprpriate issues/actions with those?
JD: Yes.
RS: Done, end of meeting.
Summary of Action Items
[NEW] ACTION: Joseph add interactive property to Core Mappings
that propogates from grid or table to all grid ARIA grid or
table structural elements. The interactive property should have
same mapping as defined in Issue 633 [recorded in
[25]http://www.w3.org/2015/04/23-aria-minutes.html#action02]
[NEW] ACTION: Michael Work with Shane to develop publication
strategy and solution that would exclude abstract roles for
authors [recorded in
[26]http://www.w3.org/2015/04/23-aria-minutes.html#action01]
[25] http://www.w3.org/2015/04/23-aria-minutes.html#action02
[26] http://www.w3.org/2015/04/23-aria-minutes.html#action01
[End of minutes]
__________________________________________________________
Minutes formatted by David Booth's [27]scribe.perl version
1.140 ([28]CVS log)
$Date: 2015/04/23 18:05:51 $
__________________________________________________________
[27] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
[28] http://dev.w3.org/cvsweb/2002/scribe/
[31] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
--
;;;;joseph.
'Array(16).join("wat" - 1) + " Batman!"'
- G. Bernhardt -
Received on Thursday, 23 April 2015 18:10:35 UTC