RE: Please clarify ARIA definition of "grid" role

Hi all, 

M understanding is that a role="grid" is different from a role="table"
although for accessibility APIs that do not make this distinction
role="grid" can be mapped to role="table".

For accessibility APis that do make the distinction (UIA has distinct grid
pattern and table pattern) it would not be correct to force the browser to
map the HTML table element to what ARIA grid is mapped to.

 

The current spec states:

"The implicit ARIA semantics defined below must be recognized by
implementations. [ARIAIMPL]
<http://dev.w3.org/html5/spec/Overview.html#refsARIAIMPL> "

 

Which is normative MUST. So if Internet Explorer mapped HTML table to a
table pattern in UIA rather than a grid pattern it would be non conforming
implementation of HTML5.

 

As I said this is my understanding. Is this the case?

 

 

with regards

Steve Faulkner
Technical Director - TPG 


www.paciellogroup.com |  <http://www.twitter.com/stevefaulkner>
http://www.twitter.com/stevefaulkner

HTML5: Techniques for providing useful text alternatives -
<http://dev.w3.org/html5/alt-techniques/>
http://dev.w3.org/html5/alt-techniques/

Web Accessibility Toolbar -
<http://www.paciellogroup.com/resources/wat-ie-about.html>
http://www.paciellogroup.com/resources/wat-ie-about.html

 

 

 

 

The implicit ARIA semantics defined below must be recognized by
implementations. [ARIAIMPL]
<http://dev.w3.org/html5/spec/Overview.html#refsARIAIMPL> From: Maciej
Stachowiak [mailto:mjs@apple.com] 
Sent: 30 August 2010 20:52
To: Richard Schwerdtfeger
Cc: James Craig; public-pfwg-comments@w3.org;
public-pfwg-comments-request@w3.org; Steve Faulkner
Subject: Re: Please clarify ARIA definition of "grid" role

 

 

On Aug 30, 2010, at 12:32 PM, Richard Schwerdtfeger wrote:





The ARIA spec is clear as with the change you suggested it more clearly
states that a grid encompasses interactive, tabular data.

We do not have a table role in aria. So, if we want to have HTML use grid as
the default host language semantics used by ARIA the only appropriate role
would be to use grid. Yet, grid is not used for static tables. 
So, if we simply state somewhere in the HTML 5 spec that a table is a form
of static grid. 

And table refers to a grid in its definition:

The  <http://dev.w3.org/html5/spec/tabular-data.html#the-table-element>
table element  <http://dev.w3.org/html5/spec/rendering.html#represents>
represents data with more than one dimension, in the form of a
<http://dev.w3.org/html5/spec/tabular-data.html#concept-table> table.

The HTML 5 definition of table:

http://dev.w3.org/html5/spec/tabular-data.html#concept-table

A table consists of cells aligned on a two-dimensional grid of slots with
coordinates (x, y). The grid is finite, and is either empty or has one or
more slots. If the grid has one or more slots, then the x coordinates are
always in the range 0 ? x < xwidth, and the y coordinates are always in the
range 0 ? y < yheight. If one or both of xwidth and yheight are zero, then
the table is empty (has no slots). Tables correspond to
<http://dev.w3.org/html5/spec/tabular-data.html#the-table-element> table
elements.

So, if we change the table description in HTML 5:

The  <http://dev.w3.org/html5/spec/tabular-data.html#the-table-element>
table element  <http://dev.w3.org/html5/spec/rendering.html#represents>
represents data with more than one dimension, in the form of a
<http://dev.w3.org/html5/spec/tabular-data.html#concept-table> table.

to 

The  <http://dev.w3.org/html5/spec/tabular-data.html#the-table-element>
table element  <http://dev.w3.org/html5/spec/rendering.html#represents>
represents data with more than one dimension, in the form of a
non-interactive
<http://dev.w3.org/html5/spec/tabular-data.html#concept-table> table.

And then put grid as the native host language semantics for table I think
the HTML spec would be in synch. 

 

1) I don't understand the goal of this proposal. I thought the problem with
using "grid" for table was that "grid" has to be interactive, but <table> by
default is not. If I understand your proposal, you suggest that the table
definition should emphasize that a table is non-interactive, but the ARIA
section should still give it a default role of "grid". I don't see how this
solves the problem.

 

2) tables may not be interactive by default, but they can be made
interactive through script or other means (for example simply by giving each
non-header cell a tabindex, or by making the contents of some cells be
contentEditable). It would be incorrect for the definition of table to
define tables as non-interactive.

 

Regards,

Maciej









Rich Schwerdtfeger
CTO Accessibility Software Group

<graycol.gif>Maciej Stachowiak ---08/30/2010 01:35:18 PM---On Aug 30, 2010,
at 11:26 AM, Richard Schwerdtfeger wrote: > James,

From: Maciej Stachowiak <mjs@apple.com>
To: Richard Schwerdtfeger/Austin/IBM@IBMUS
Cc: James Craig <jcraig@apple.com>, public-pfwg-comments@w3.org,
public-pfwg-comments-request@w3.org, Steve Faulkner
<sfaulkner@paciellogroup.com>
Date: 08/30/2010 01:35 PM
Subject: Re: Please clarify ARIA definition of "grid" role

  _____  





On Aug 30, 2010, at 11:26 AM, Richard Schwerdtfeger wrote:

James, 

The reason we created grid was to turn a table into one that is interactive,
however, some component libraries like Dojo do not use table to create the
actual grid. They use divs and spans to lay out the structure. 
I think if we were to do what you suggest HTML should modify its definition
of table to be a non-interactive grid. 

As for a grid having to be in inside an element with role="application" that
is inaccurate. Windows assistive technologies now recognize elements with an
aria role of grid and pass keyboard navigation off to the widget as they
know they are interactive. No role="application" on the body tag is
required.

Maciej,

Would HTML be willing state that a table is an non-interactive grid in its
definition or simply state that <table> is non-interactive?

I'm not sure what that would entail. Would a table have role=grid in that
case? Would additional notes in the ARIA mapping table or elsewhere be
required?

I personally have no strong feelings on whether the grid role needs to be
interactive. I would be happy either way. However:

a) The ARIA spec should be exceeding clear one way or the other on this
issue, 
b) It would be good to have a clear decision soon, since this will impact
HTML5 and it would be better not to thrash.

I can ask the editor to hold off on fielding the bug that asks for the grid
role to be removed from table, but in general we're trying to get these
HTML5 bugs resolved as quickly as possible.

Regards,
Maciej 





Rich Schwerdtfeger
CTO Accessibility Software Group

<graycol.gif>James Craig ---08/29/2010 02:17:35 PM---Added the following
note to ISSUE-421. I'd support allowing the HTML role mapping of table to
grid b

From: James Craig <jcraig@apple.com>
To: Richard Schwerdtfeger/Austin/IBM@IBMUS
Cc: Maciej Stachowiak <mjs@apple.com>, public-pfwg-comments@w3.org,
public-pfwg-comments-request@w3.org, Steve Faulkner
<sfaulkner@paciellogroup.com>
Date: 08/29/2010 02:17 PM
Subject: Re: Please clarify ARIA definition of "grid" role
Sent by: public-pfwg-comments-request@w3.org

 

  _____  




Added the following note to ISSUE-421.

I'd support allowing the HTML role mapping of table to grid because the
distinction you're making applies to behavior, not role. A table is just a
readonly grid that is not focus-managed by the author. If we limit grids to
interactive controls and explicitly disallow the role on tables, we won't
have role symmetry with HTML, and we'd need to define a few more roles. We
specifically added rowgroup to account for thead, tbody, and tfoot, so we've
already implied that this is a suitable role for HTML tables. 

I believe the reason you're making the distinction is due to the Windows
screen readers behavior of key events inside elements with role application.
What if we added the prose to indicate that grids inside
[role="application"] where interactive, author-managed-focus controls, and
grids inside [role="document"] were readonly, UA-managed-focus tables?


On Aug 28, 2010, at 9:13 PM, Richard Schwerdtfeger wrote: 

Maciej,

That's fair. I have opened issue 421 to address this change with your
suggested text.  <http://www.w3.org/WAI/PF/Group/track/issues/421>
http://www.w3.org/WAI/PF/Group/track/issues/421

Thank you,
Rich


Rich Schwerdtfeger
CTO Accessibility Software Group

<graycol.gif>Maciej Stachowiak ---08/28/2010 08:45:00 PM---On Aug 26, 2010,
at 7:51 PM, Richard Schwerdtfeger wrote: > Tables do NOT allow users to move
focus

From: Maciej Stachowiak < <mailto:mjs@apple.com> mjs@apple.com>
To: Richard Schwerdtfeger/Austin/IBM@IBMUS
Cc:  <mailto:public-pfwg-comments@w3.org> public-pfwg-comments@w3.org,
<mailto:public-pfwg-comments-request@w3.org>
public-pfwg-comments-request@w3.org, Steve Faulkner <
<mailto:sfaulkner@paciellogroup.com> sfaulkner@paciellogroup.com>
Date: 08/28/2010 08:45 PM
Subject: Re: Please clarify ARIA definition of "grid" role 

  _____  





On Aug 26, 2010, at 7:51 PM, Richard Schwerdtfeger wrote:

> Tables do NOT allow users to move focus between cells using two
dimensional navigation. 
> 
My reading of that sentence was that it was a statement of fact about what
would happen if you apply the grid role (that the assistive technology would
provide two-dimentional navigation), not a limitation on when it may be
applied. Since multiple people misunderstood this definition on first
reading, I think a clarification would still be helpful, even if you think
it is not strictly necessary. For example:

"A grid is an interactive control which contains cells of tabular data
arranged in rows and columns, like a table."

I think that simple change to ARIA would have prevented all confusion.

Regards,
Maciej

> 
> 
> Rich Schwerdtfeger
> CTO Accessibility Software Group
> 
> <graycol.gif>Maciej Stachowiak ---08/26/2010 05:03:10 AM---ARIA has the
following definition of the "grid" role: ------------
> 
> From: Maciej Stachowiak < <mailto:mjs@apple.com> mjs@apple.com>
> To:  <mailto:public-pfwg-comments@w3.org> public-pfwg-comments@w3.org
> Cc: Steve Faulkner < <mailto:sfaulkner@paciellogroup.com>
sfaulkner@paciellogroup.com>
> Date: 08/26/2010 05:03 AM
> Subject: Please clarify ARIA definition of "grid" role
> Sent by:  <mailto:public-pfwg-comments-request@w3.org>
public-pfwg-comments-request@w3.org
> 
> 
> 
> 
> ARIA has the following definition of the "grid" role:
> 
> ------------
> < <http://www.w3.org/TR/wai-aria/roles#grid>
http://www.w3.org/TR/wai-aria/roles#grid>
> A grid contains cells of tabular data arranged in rows and columns, like a
table.
> 
> Grids do not necessarily imply presentation. The grid construct describes
relationships between data such that it may be used for different
presentations. Grids allow the user to move focus between cells using two
dimensional navigation. For example, grid might be used as the invisible
data model (hidden with CSS but still operable by assistive technologies)
for a presentational chart.
> ------------
> 
> This seems like an exact match for the HTML <table> element, which
contains cells of tabular data arranged in rows and columns, like a table.
According to Steve Faulkner, however, this role is only supposed to be used
for *interactive* presentations of tabular data arranged in rows and
columns. Please update the ARIA draft to make this clear. Or, if the grid
role is in fact supposed to be applicable even to non-interactive
presentations of tabular data, please reply with this information ASAP,
since it is relevant to the ARIA role mapping in the HTML5 spec.
> 
> Regards,
> Maciej
> 
> 
>