RE: Table Techniques - Summary

Machine tests could just as easily query the author about the nature of a
table based on the existence of the <TH> element. When a machine test
encounters <TH>, it would then require that a valid summary (not a null
summary) also be present in the same way that Lee describes a machine test
needing to do the reverse based on the presence of summary. 

I don't think this trivializes the importance of summary for authors.
Instead, it reinforces the idea that using technologies according to
specification (in this case, including structural attributes only when they
have meaning and purpose that relates directly to the content) is an
important part of the process. From an authors perspective, guidelines that
would require an author who uses tables exclusively for layout to add a
semantically meaningless attribute to each and every table on their site is
far more problematic than asking them to understand the difference between
layout and data tables and include appropriate markup.

Though there is a good deal of room for interpretation, the techniques from
WCAG 1.0 [1] seem to support an approach that would not require include
summary in layout tables:

WCAG 1.0 Checkpoint 5.5 says, "Provide summaries for tables." However, the
examples and prose in the techniques document for this checkpoint provides
examples that relate only to data tables.[2]

WCAG 1.0 Checkpoint 5.4 says, "If a table is used for layout, do not use any
structural markup for the purpose of visual formatting." Prose for this
checkpoint implies that the inclusion of structural markup like TH can be
misleading when used in layout tables.[3]

While table summaries don't have an impact on visual formatting, they do
provide structure to tables and, in my opinion, have the same potential to
be as misleading as misuse of the TH element. 

In the WCAG 2.0 Techniques, I think the message about summary on tables
should be:

	1. Provide summaries for data tables.
	2. If a table is used for layout, do not use any structural markup.

-Ben

[1] http://www.w3.org/TR/WCAG10-HTML-TECHS/
[2] http://www.w3.org/TR/WCAG10-HTML-TECHS/#table-summary-info 
[3] http://www.w3.org/TR/WCAG10-HTML-TECHS/#tables-layout

> -----Original Message-----
> From: w3c-wai-gl-request@w3.org [mailto:w3c-wai-gl-request@w3.org] On
> Behalf Of Roberto Scano - IWA/HWG
> Sent: Friday, August 15, 2003 9:14 AM
> To: Lee Roberts; 'John M Slatin'; 'Matt May'; 'Chris Ridpath'
> Cc: 'WAI WCAG List'
> Subject: Re: Table Techniques - Summary
> 
> 
> ----- Original Message -----
> From: "Lee Roberts" <leeroberts@roserockdesign.com>
> To: "'Roberto Scano - IWA/HWG'" <rscano@iwa-italy.org>; "'John M
> Slatin'" <john_slatin@austin.utexas.edu>; "'Matt May'" <mcmay@w3.org>;
> "'Chris Ridpath'" <chris.ridpath@utoronto.ca>
> Cc: "'WAI WCAG List'" <w3c-wai-gl@w3.org>
> Sent: Friday, August 15, 2003 4:00 PM
> Subject: RE: Table Techniques - Summary
> 
> 
> >If we allow the non-use of the summary attribute we are, IMHO,
> informing
> >designers that it is not important.  The summary is just as important
> as
> >the alt.
> 
> I could agree with you but as in the normative documents is wrote, "alt"
> must be present (also null, as explained also in some interesting
> articles [1]. Using an empty summary, referring to the normative
> guidelines, is not needed.
> 
> >As I've pointed out many times previously, a machine tester can
> validate
> >a null summary and ensure that there are no captions and th elements
> >used.  The non-use of the summary attribute then makes the machine test
> >invalid because the machine would have no idea whether the table is for
> >layout or tabular information.
> 
> This means that this would be a machine-tester problem? Must the
> guidelines adapted to machine tester esigence or to the W3C guidelines
> and to the usability of the people with disability?
> 
> Btw, we could add a techniques that "suggest" to use empty summary for
> layout tables but we cannot said: layout tables must have empty summary
> because:
> 
> 1- some people could find useful use summary for describe the layout
> table (eg: "this table is formed by three columns: one for the menu, one
> for the contents and one for the useful links")
> 2- the guidelines for HTML / XHTML, if we ask for a *must be present*,
> need to be upgraded (adding in the HTML / XHTML Errata?)
> 
> >For example, in English the term layout would have to translated and
> >understood by the machine tester if we allowed content in the summary
> >attribute.  How many languages are there?  Would the machine tester
> have
> >to know each translation?
> 
> As suggested by John M. Slatin, the use of new attributes could be good
> but these will be applied only since XHTML 2.0 and newer
> raccomandations... for the *old* HTML 4.xx and XHTML 1.x this, i think,
> could be a problem...
> 
> >Therefore, it is much easier to machine test for a null summary than
> >content that would otherwise indicate the table is used for a layout.
> 
> Remember that the guidelines must be *testable* that this means not only
> machine-testable... :)
> 
> >And again, the lack of a summary would lead designers to think that the
> >summary attribute is not important.  Hence, the null summary provides
> >valuable information.
> >My recommendation is to require a null summary for layout tables.
> 
> What valuable information are given by an empty attribute for what the
> guidelines don't have thinked to put it as *required* ?
> 
> 
> Roberto Scano
> ---
> [1] http://ppewww.ph.gla.ac.uk/~flavell/alt/alt-text.html

Received on Friday, 15 August 2003 12:18:26 UTC