W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > July to September 2004

re: Layout versus data tables proposal for null summary attribute

From: Lois Wakeman <lois@lois.co.uk>
Date: Tue, 31 Aug 2004 10:57:22 +0100
Message-ID: <B0059539669@inetc133.inetc.net>
To: "WAI list" <w3c-wai-ig@w3.org>
Message re-posted to list after earlier bounce:

This discussion touches on something I have been thinking about for ages,
but hesitated to bring up as I'd rather not be shot down in flames.

The basic problem seems to be that although it is possible to do really nice
layouts in CSS (as demonstrated by CSS Zen Gardens for example), it is not
that well-supported (by rendering as well as coding tools), nor as easy to
do as by plonking things into a table. By and large, it is not done except
by keen proponents of good design, accessibility and standards, who are few
and far between.

We have already had one workaround proposed to distinguish data tables from
layout tables - but that relies on convention or best practice, rather than
being part of HTML semantics, and seems to be a backwards step, in
overloading the semantics of tables as originally conceived.

It is obvious to me that, despite our best evangelical efforts, most
designers want to be able to use grids on the web just like they do in DTP,
and no amount of preaching will change this soon. And perhaps there is
something inherently attractive about grid layouts that makes pages more
usable or attractive - but that is a wild guess. 

By insisting on CSS and absolute positioning, we make things harder for
non-experts to create - but we can feel very smug because *we* can do it,
no? Using tables is easy, works in 99% of browsers, but is frowned upon by
us white hats.

I have a radical solution (flame throwers ready guys?). Why not propose a
new element for XHTML n.n, called grid? It would be a semantic-free
construct for layout, working in the same way as a simple table without
summaries, headings, column groups etc; I imagine spanning could be allowed.

By default, it linearises in the order the cells are placed in the code,
though there is no reason why there should not be a render-order attribute
to affect this for AT.

Well, why not?

1. Oh no, another element to wrangle over for years in W3C WGs.

2. If accepted, perhaps another 5 years till new browsers even begin to
implement it, even though most of the table render code could be reused.

3. By the time 1 and 2 have happened, all the common tools will have decent
support for table free gridded layouts. And with any luck, I'll have

Lois Wakeman


http://lois.co.uk <http://lois.co.uk/> 

http://siteusability.com <http://siteusability.com/> 

http://communicationarts.co.uk <http://communicationarts.co.uk/> 


Received on Wednesday, 1 September 2004 12:18:08 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 13 October 2015 16:21:29 UTC