W3C home > Mailing lists > Public > wai-xtech@w3.org > December 2009

Re: ARIA setsize attribute

From: Alexander Surkov <surkov.alexander@gmail.com>
Date: Sat, 12 Dec 2009 11:08:35 +0800
Message-ID: <faf3cb2b0912111908u30f69c8by2d3f8f8c703aa6dc@mail.gmail.com>
To: James Craig <jcraig@apple.com>
Cc: Hans Hillen <hhillen@paciellogroup.com>, wai-xtech@w3.org, David Bolter <dbolter@mozilla.com>, Richard Schwerdtfeger <schwer@us.ibm.com>
I think aria-setsize can be calculated easy enough from aria-level
attribute for the flat trees like

<div role="tree">
  <div role="treeitem" aria-level="1">item1</div>
  <div role="treeitem" aria-level="2">item1.1</div>
  <div role="treeitem" aria-level="2">item1.2</div>
 <div role="treeitem" aria-level="1">item2</div>
</div>

Obviously the setsize for the item1.1 is 2. If the author provides
aria-setsize on the item1 then the user agent could find get it for
item1.1 and item1.2. It's similar to how it happens in the case of
node_child_of relation.

Alex.

On Sat, Dec 12, 2009 at 10:55 AM, James Craig <jcraig@apple.com> wrote:
> In light of Hans' reminder, I now recall the original reason for this
> requirement was for the nested groups in trees and treegrids. Although the
> setsize could logically be set at a tree or group node, it's possible the
> visible container may only be displaying a subset items within one nested
> set, so it's not necessary or appropriate to require the author to render
> the group/set node in the DOM.  We could get away with using setsize on the
> node in single-level widgets (lists, grids, etc) but would then be defining
> it differently than is required for nested widgets (trees, treegrids).
> I'm going to correct the new example rather than propose changing the
> group-approved text. I'm also dropping the newly raised issue.
> There's been a deluge of spec edits over the past few months, so there are
> bound to be some mistakes. Thanks again for the quick catch.
>
> On Dec 11, 2009, at 6:13 PM, Hans Hillen wrote:
>
> From what I understand, aria-setsize is intended to fix dom structures where
> the set size can't be determined automatically from the DOM, e.g. a tree
> widget that is marked up as a flat html table and every row has a role of
> treeitem. In this case case the treeitems are not grouped by branch in the
> DOM structure, so you would need to apply aria-setsize aria-posinset to and
> aria-level  to every treeitem to fix this lack of structure.
> Wouldn't setting aria-setsize on a container be a bit pointless, because:
>
> If the container truly represents the set, then you wouldn't need the
> aria-setsize attribute anymore because the setsize can be determined from
> the DOM structure. If it can't be (for example because certain child roles
> are not DOM children of the container) then aria-owns should be used and the
> user agent should add it to the automatically calculated set size.
> If the container does not reflect the actual set (as in my tree example
> above), then aria-setsize wouldn't be applicable on this node.
>
> Unless I'm missing something?
> Regards,
>
> Hans Hillen
> TPG Europe
>
>
>
>
>
> On Dec 12, 2009, at 2:29 PM, Alexander Surkov wrote:
>
> Hi.
>
> The ARIA specification says "This property is marked on the members of
> a set, not the container element that collects the members of the set.
> " (see http://www.w3.org/WAI/PF/aria/states_and_properties#aria-setsize).
> However the same time it provides an example where the aria-setsize is
> used on the container element:
>
> <ul role="listbox" aria-setsize="16" aria-labelledby="label_fruit">
>  <li role="option" aria-posinset="1"> apples </li>
>
> I find the idea to define aria-setsize on the container element
> reasonable and useful and I would happy if the user agents would take
> into account the aria-setsize on the container only. But I'm fine with
> either way because currently aria-setsize is supposed to be used on
> the item.
>
> Can the spec be fixed?
>
> Thank you.
> Alex.
>
>
>
>
Received on Saturday, 12 December 2009 03:09:15 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:25:28 UTC