W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > October to December 2013

Re: Complex table with headings (h3, h4 etc.) to group data rows

From: Joshue O Connor <joshue.oconnor@cfit.ie>
Date: Tue, 29 Oct 2013 15:57:57 +0000
Message-ID: <526FDB05.4000609@cfit.ie>
To: Sailesh Panchang <spanchang02@yahoo.com>
CC: "w3c-wai-gl@w3.org" <w3c-wai-gl@w3.org>, david100@sympatico.ca
Hi Sailesh,

> My main question is:
> how does someone say this technique is invalid / cannot be┬ ┬ ┬ sufficient when
> - An h-tag is valid in HTML 4.01 and in a TD cell of an HTML5 table

<WCAG Chair hat off>

As I mentioned previously, I don't like this method because it doesn't 
sit well with the current user interaction model of being able to browse 
by headings (or indeed any HTML elements) within a page. For example, a 
screen reader user may find themselves in a page with several heading 
elements, some paragraphs and a couple of data tables. Using your method 
(as such) if browsing by headings then the overview of the page would be 
a mix of document headings and then any headings that you have nested 
within your table. Furthermore, they would have no idea which were 
which, without a priori knowledge of the page and its contents.

Secondly, AFAIK, there is no way to associate the document headings with 
and table sub headings, etc. so the relationship between one of these 
headings and its surrounding content is rather loose. This may be ok, in 
some cases as a quick fix for navigation but isn't ideal.

Thirdly, leaving aside that your example is valid HTML4 for a moment. 
According to the HTML5 spec, it states that:

<html5specquote>
ôThe first element of heading content in an element of sectioning 
content represents the heading for that section.ö

ôCertain elements are said to be sectioning roots, including blockquote 
and td elements. These elements can have their own outlines, but the 
sections and headings inside these elements do not contribute to the 
outlines of their ancestors.ö [1]

</html5specquote>

My reading of this is that  a heading used within a data table would NOT 
be able to relate to any following <td> cells? Therefore the 
relationship between a table header and a sub header, or data cell 
cannot be maintained. I may be wrong, and am happy to be proved so.

Finally, your example does show to me that this kind of functionality is 
needed, so I feel taking up this issue within the HTML5 may be a useful 
thing to do, if the language itself is deficient. If nested headers were 
allowed and then appropriate tabular markup could be used, then well and 
good.

<WCAG Chair hat on>
So while the use case is interesting, my current reading of the HTML5 
spec and my previous experience leads me to conclude that this isn't an 
appropriate use of HTML5 markup, though it is indicative of an 
interesting use case that needs further exploration.

Thanks

Josh

[1] http://www.w3.org/TR/html5/tabular-data.html#the-th-element


> - It helps users and developers and is AT supported.
>
> If I interpret David MacDonald's emails and notings,┬  I believe he seems to┬  see the value in it too
> http://davidmacd.com/test/sailesh-table-test.html
> I believe there are more who consider this to be a very pragmatic approach.
>
> The WCAG-WG inserts a sentence┬  saying 'even though the absence of an alt attribute on an image is invalid, we deem it is alright if one wants to use an aria-labelledby on a static image to convey text alternative'. Absence of an alt attribute as a failure as per WCAG1, S508 and WCAG2 F38 / F65. All tools and accessibility practitioners call this out as a failure.
> A static image is not a rich element for which ARIA is intended.
> The use of a title attribute is deemed sufficient despite┬  user agent issues noted and lack of keyboard support or other techniques are deemed sufficient even┬  when a single AT like JAWS supports it.
> And the techniques doc documents that┬  an SC can be met by other means┬  too ... I assume basically because all techniques cannot be documented for practical reasons.
> I did see the discussion and notings in minutes of  Oct 22
> http://www.w3.org/2013/10/22-wai-wcag-minutes.html
> The minutes disregarded  the  fact that this complex table has a summary attribute  which explains the  ttable's structure which I highlighted is past emails is a critical part of the suggested technique.
> So I am really┬  really lost in┬  trying to understand the motivation behind the reasoning for not considering the use of h-tags as a sufficient in certain situations as documented.
> Code that is not valid gets a special note, saying, it is not important but we bless this technique as valid. And things that is valid, works with AT, helps users and developers and promotes accessibility is not deemed sufficient?
> So please document  how you decide something is sufficient even though not valid. Without this, the  response as documented in the minutes is incomplete and unacceptable.
> There is discussion and documentation that Joshue sent me  which indicates that many disagree with HTML5 requirement that headers attribute of a data cell should only reference TH cells and not TD cells when HTML4.01 recognizes this reality.
> HTML5 table specs itself clearly states that user agents should render  the non-conforming (as per HTML5) summary attribute.
> As of today, HTML5 is not a recommendation.
> WCAG2 SC 4.1.1  does not fail use of  deprecated attributes.
>
> Thanks,
> Sailesh Panchang
> ┬
> --------------------------------------------
> On Wed, 10/9/13, Joshue O Connor<joshue.oconnor@cfit.ie>  wrote:
>
>   Subject: Re: Complex table with headings (h3, h4 etc.) to group data rows
>   To: "Sailesh Panchang"<spanchang02@yahoo.com>
>   Cc: "w3c-wai-gl@w3.org"<w3c-wai-gl@w3.org>
>   Date: Wednesday, October 9, 2013, 6:22 AM
>
>   Sailesh Panchang wrote:
>   >  Dear Andrew and WCAG-WG,
>   [...]
>   >  Surely all complex tables can be broken into simpler
>   tables but that is seldom possible in real life.
>   >  And as a user, I├óÔéČÔäód rather have a complex table
>   with the alternative markup that makes it accessible than no
>   accessibility markup on grounds of undue burden.
>
>   Thanks for that Sailesh, while I agree with this mood and
>   spirit, I'm
>   reluctant to codify this even as a remedial approach.
>   However, I will
>   test it a little more and bring feedback to next weeks
>   call.
>
>   >  Thanks for your time and pardon my persistence.
>
>   Not at all, thanks for your persistence!
>
>   Josh
>
>   >  Regards,
>   >  Sailesh
>   >  [1] http://mars.dequecloud.com/demo/Census_2013.htm
>   >
>   >  --------------------------------------------
>   >  On Tue, 10/8/13, Andrew Kirkpatrick<akirkpat@adobe.com>┬
>   wrote:
>   >
>   >┬ ┬ ┬ Subject: RE: Complex table with
>   headings (h3, h4 etc.) to group data rows
>   >┬ ┬ ┬ To: "Sailesh Panchang"<spanchang02@yahoo.com>,
>   "w3c-wai-gl@w3.org"<w3c-wai-gl@w3.org>
>   >┬ ┬ ┬ Date: Tuesday, October 8, 2013, 2:07
>   PM
>   >
>   >┬ ┬ ┬ Sailesh,
>   >┬ ┬ ┬ My issue with this is that there is
>   nothing in the spec that
>   >┬ ┬ ┬ says that headings (e.g. h2) should be
>   used in this way, and
>   >┬ ┬ ┬ the user agent support for identifying
>   table cells isn't
>   >   there yet.  When I use JAWS on
>   your table and jump to
>   >┬ ┬ ┬ the last cell or move to an arbitrary
>   cell in the table and
>   >┬ ┬ ┬ request the header information to
>   establish the context it
>   >┬ ┬ ┬ only reads headers that are defined
>   with the standard table
>   >   markup.  A user wouldn't know
>   when or if there were
>   >┬ ┬ ┬ headings marked up in this alternative
>   way, so to be certain
>   >┬ ┬ ┬ a user would need to hit JAWS+T every
>   time they thought that
>   >┬ ┬ ┬ there might be an additional heading
>   defined in this
>   >   non-standard way.  I don't see
>   this as very workable
>   >┬ ┬ ┬ for end users, even if it has the
>   potential to save some
>   >┬ ┬ ┬ developer time.
>   >
>   >┬ ┬ ┬ Thanks,
>   >┬ ┬ ┬ AWK
>   >
>   >┬ ┬ ┬ Andrew Kirkpatrick
>   >┬ ┬ ┬ Group Product Manager, Accessibility
>   >┬ ┬ ┬ Adobe Systems
>   >
>   >┬ ┬ ┬ akirkpat@adobe.com
>   >┬ ┬ ┬ http://twitter.com/awkawk
>   >┬ ┬ ┬ http://blogs.adobe.com/accessibility
>   >
>   >
>   >┬ ┬ ┬ -----Original Message-----
>   >┬ ┬ ┬ From: Sailesh Panchang [mailto:spanchang02@yahoo.com]
>   >
>   >┬ ┬ ┬ Sent: Tuesday, October 08, 2013 1:08
>   PM
>   >┬ ┬ ┬ To: w3c-wai-gl@w3.org
>   >┬ ┬ ┬ Subject: Complex table with headings
>   (h3, h4 etc.) to group
>   >┬ ┬ ┬ data rows
>   >
>   >┬ ┬ ┬ Refer to table#2 on this page:
>   >┬ ┬ ┬ http://mars.dequecloud.com/demo/Census_2013.htm
>   >
>   >┬ ┬ ┬ Well if one is arrowing down a column
>   # 2or subsequent
>   >┬ ┬ ┬ column, the screen reader does
>   announce the group header
>   >┬ ┬ ┬ because it is a TH cell./ (this is in
>   response to Josh's
>   >┬ ┬ ┬ question during the call).
>   >┬ ┬ ┬ Use of h<n>┬  tags in a data
>   table is valid code, it
>   >┬ ┬ ┬ exposes structure of the grouping
>   within the table, allows
>   >┬ ┬ ┬ users to jump to the section of
>   interest and then navigate
>   >┬ ┬ ┬ the table and have only row
>   header  to its left and
>   >┬ ┬ ┬ column headers announced. This reduces
>   verbosity. Yet one
>   >┬ ┬ ┬ can press JAWS+T and have nearest
>   heading announced ... a
>   >┬ ┬ ┬ technique that is accepted for
>   determining context of a
>   >┬ ┬ ┬ link. So one can use the same
>   technique to understandÂ
>   >┬ ┬ ┬ context in terms of group headers.
>   (Only JAWS  supports
>   >┬ ┬ ┬ H80) Also one navigates a page and
>   determines structure by
>   >┬ ┬ ┬ multiple methods: landmarks, headings,
>   tables, lists, forms
>   >┬ ┬ ┬ etc. and not just heading navigation.
>   So if this method is
>   >┬ ┬ ┬ used to code complex tables, users
>   will be able to figure
>   >┬ ┬ ┬ that out.
>   >┬ ┬ ┬ Also a single level of headings
>   (styled as needed)Â
>   >┬ ┬ ┬ should be used in the table ... the
>   method may not be
>   >┬ ┬ ┬ appropriate if one needs to use more
>   than one level of
>   >   headings.Â
>   >┬ ┬ ┬ It is a non "standard" method that is
>   valid code and AT
>   >┬ ┬ ┬ supported.
>   >┬ ┬ ┬ I believe it is an alternative
>   technique to headers-id when
>   >┬ ┬ ┬ one chooses not to use headers-id for
>   whatever reason.
>   >
>   >┬ ┬ ┬ Thanks and regards,
>   >┬ ┬ ┬ Sailesh Panchang
>   >
>   >
>   >
>   >
>   >
>   >
>   >
>   >
>   >
>   >
>
>
>
>
>
>
Received on Tuesday, 29 October 2013 15:58:31 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 20:32:54 UTC