W3C home > Mailing lists > Public > public-html@w3.org > December 2008

Re:Comparison of Smart Headers and HTML5 (ACTION-85)

From: Joshue O Connor <joshue.oconnor@cfit.ie>
Date: Sat, 06 Dec 2008 09:51:59 +0000
Message-ID: <493A4B3F.8010402@cfit.ie>
To: Ben Millard <cerbera@projectcerbera.com>
Cc: Aaron M Leventhal <aleventh@us.ibm.com>, HTMLWG <public-html@w3.org>, WAI-XTECH <wai-xtech@w3.org>

Ben Millard wrote:
> 1. The current algorithm runs through the <table> once to find all the
> associations.
> 2. The UA stores them in the relevant accessibility API properties.
> 3. A user interacts with a cell via an AT.
> 4. The AT uses what has been stored in the API for that cell.
> 5. The AT may apply verbosity control, announcements, re-ordering and
> other adaptations to improve the user experience.
> 
> In this way, the relationships are stored rather than queried on-demand.
> Do you think that's a good thing? Avoiding repetitive queries between AT
> and UA seems like a good thing to me, since a user can move between
> table cells rapidly.

I am not sure I understand this part. With a screen reader for example,
many use the OSM or virtual buffer and take a snapshot of the DOM
contents in order to update it when a page loads. This would have to be
updated fairly regularly if there are persistent content changes on a
page. With a screen reader like JAWS a user used to have to manually
update this buffer using the INSERT+ESC keys, but later versions of JAWS
now do this pretty much automatically.

The user then interacts with or interrogates this buffered content
rather than interact directly with the screen, hence the Off Screen
Model. It's only when the user has to enter information into a form such
as text, or use other form controls, that the UA switches into 'Forms
Mode' - this mode of interaction is actually the only time the UA
interacts directly with the browser and not the buffered content. So the
user querying this content does not mean any kind of excess HTTP traffic
as it is all done on the client side until the next browser refresh
(AJAX type functionality aside for the moment). Also the screen reader
often does not interact with the DOM directly, currently most AT uses
this model AFAIK.

So how is this model of storing relationships different from the
explicit semantics already used by the markup language, that are queried
when needed? Are these relationships stored in the DOM? In a sense they
are already 'stored' and called upon when needed as the user interacts
with the table, please do enlighten me if I am missing something.

Cheers

Josh
Received on Saturday, 6 December 2008 09:52:49 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 May 2012 00:16:27 GMT