W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > July to September 2005

RE: [techs] Table Summary Tests (111, 112, 113, 114, 203)

From: Mirabella, Mathew J <Mathew.Mirabella@team.telstra.com>
Date: Thu, 18 Aug 2005 16:35:19 +1000
Message-ID: <73388857A695D31197EF00508B08F29816655521@ntmsg0131.corpmail.telstra.com.au>
To: "w3c-wai-gl list" <w3c-wai-gl@w3.org>

Yes.  Wcag  guidelines are for all disabilities and are not and should
not be specific to user agents or assistive technologies.  I am just
showing/illustrating here an example of one thing I think Jaws is doing
right in terms of how it allows users to skip between tables and

If, however, you are looking for ways to link a table to a heading or
caption, then suggesting that it is ok to just have a heading right
above a table is not really sufficient because the heading could be for
anything else.  In that case, how can you expect a user agent or
assistive technology to do anything useful in terms of knowing that this
hx is meant to be part of this table etc.

So Re your other email, yes the user agent should identify the table
presence, and it does, but how can it identify the hx as intimately
related to the table if the hx is outside the table?

So, linking it like having the hx inside caption seems to be one good
way to programmatically include a caption that really is for the table,
but to provide a heading at the same time, if a heading is what you
want, without redundant repetitions.  You do not of course have to have
hx inside the table caption... It all depends on what structures you
want in there. But... For provision of a bit of text that describes the
table, caption is really the way to do it, and my example shows a
possible way to serve a couple of purposes at once.

What user agents do with all this mark-up may not be a specific wcag
issue, but if you have wcag techniques that promote the kind of explicit
coded associations I am talking about, it then at least allows user
agents to have a hope of doing something useful other than just one
thing comes after the next on a page.  (C.f. alt text for images are in
an alt attribute, not just a bit of text next to the image on the page),
now user agents may do different things with that, but at least they can
do something with it that is useful re telling users that this text is
really for this image).

BTW:  I also think summary attributes are the way to go to provide extra
useful information about the structure of a table etc.  i.e. telling
users of screen readers that this table has many columns and some cells
that span over a couple of columns etc. so they know what to expect when
they start moving around the table.

-----Original Message-----
From: Roberto Scano (IWA/HWG) [mailto:rscano@iwa-italy.org] 
Sent: Thursday, 18 August 2005 4:07 PM
To: Mirabella, Mathew J; w3c-wai-gl@w3.org
Subject: RE: [techs] Table Summary Tests (111, 112, 113, 114, 203)

We need to develop wcag that are for *all* disabilities, and not be
vendor-specific. If it's a Jaws problem, ask to fix it to them, and not
that w3c should fix wcag guidelines.
(as for valid code...)
----- Messaggio originale -----
    Da: "Mirabella, Mathew J"<Mathew.Mirabella@team.telstra.com>
    Inviato: 18/08/05 4.40.34
    A: "w3c-wai-gl list"<w3c-wai-gl@w3.org>
    Oggetto: [techs] Table Summary Tests (111, 112, 113, 114, 203)
    >> but with JAWS the user can jump directly to any table on the page
    >> pressing the letter "t".  If the table has a <caption> element,
    >> speaks the caption; it will also speak the summary if it's
    >> Pressing the "t" would cause JAWS to bypass the H<x> element in
    >> example above.
    > Thank you for documenting another way in which Jaws is a broken
    > agent. I'm sure the Working Group will custom-craft its techniques
    > accommodate exactly what Jaws does in this instance.
    I agree that Jaws is far from perfect, especially considering that
    every so called improved new version of this same program seems
    to change the way it behaves with web content, and as such causes
    problems, for 
    Real people etc.
    However, in an example where you have an hx element above a table,
    a caption
    in a table, What is inherently wrong with Jaws doing something like
    moving to the
    table (thus skipping the hx) when the user presses t?  h/shift-h in
    Between the headings, t/shift-t moves between the tables.  Why
    jaws consider
    An hx to be any part of the table just because it appears above the
    For all we know, the hx could be a section heading in the page that
    happens to be
    followed by a table.
    If we want to suggest that an hx just above a table is suficient to
    actually link  that
    Table to that hx as explicitly as, for example, label with form
    then we are
    starting to expect user agents and assistive technologies to make
    decision about the linkages
    Between different things on a page due to nothing more than screen
    code proximity.  I hope
    People can see the problems inherent in this.
    So how can you link a table to a header that is above it... Is there
    for/id linkage that can be
    made?  No... So my example shows the inclusion of an hx inside the
    caption.... Which is not
    Given just for jaws compatibility, but because it seems to be a
    way to resolve
    The linkage issue in an explicit way in the code... Which is what we
    So in the example where the hx is above the table with caption, Jaws
    probably not really
    Doing the wrong thing.  That's not to say that Jaws never does the
    thing by the way.

[Messaggio troncato. Toccare Modifica->Segna per il download per
recuperare la restante parte.]
Received on Thursday, 18 August 2005 06:40:26 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 16 January 2018 15:33:55 UTC