W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > July to September 1998

Re: http://www.w3.org/TR/WD-WAI-PAGEAUTH

From: nir dagan <dagan@upf.es>
Date: Sun Jul 26 11:41:56 1998
Message-Id: <199807261534.RAA19635@sahara.upf.es>
To: EKNUTH@tiny.computing.csbsju.edu
Cc: w3c-wai-gl@w3.org
> Tables:
> 1. I have tables that look and read fine in Lynx 2.4.2 through 
> 2.7.2, that make sense in HelpRead (formerly Read To Me), but 
> which fail the "sheet of paper over graphical browser screen" 
> rule of thumb.  How do screen readers really work?  Straight 
> across, or taking cells and paragraphs into account?

Lynx disregards all the table markup (TABLE,TR,TD,TH...)
So if the pargraphs are ordered logically in the HTML code there 
wouldn't be a problem. 

With screen readers it is a different story. Some of them work very 
similar to the "sheet of paper" test. If one wants to make sure things 
are readable on all screen readers one must avoid placing text blocks 
side by side via HTML tables. 

> 2. Bobby (CAST) implies that a text-only version is required 
> for all tables with more than one cell.  I believe this imposes 
> an undue burden on webmasters.

This recommendation is based on the screen reader issue. The tendency 
in the WAI is to allow writing tables, when the nature of the information 
calls for it. At the same time avoid (or at least minimize the use of) 
tables as a mere layout mechanism.

From a pure theoretical point of view, the screen reader technology
is the wrong tool to use. The theory says: authors write tables when the 
information calls for it, and use the accessibility features of HTML4.0.
User agents (browsers) interpret HTML4.0 well. Authors should layout their
pages with stylesheets, which are in turn well supported by browsers.

In practice authors should be nice to "bacward technology". There is 
a subtle line of how backward compatible you can get, and with what 
technology you are backward compatible. There are tables that are better
than others. The current draft does not refer to that in detail.

> 3. If tables are discouraged, I think conscientious webmasters 
> will make greater use of imagemaps... but I also believe that 
> imagemaps are a greater burden to accessibility than well-designed 
> tables.

I don't think people will use image maps more. Image maps are being used 
less and less over the net due to usability drawbacks in graphical media.
Text links are very uasable: the user distinguishes visited and unvisited 
links, gets feedback via active and hover, and download time is smaller.

Read my anti images-as-links rant at:

Also webmasters who want to put the navigation bar of the left of
the flow of text will still need a table to do it with the image map. 
If they float the image <IMG align=left>, text will wrap under it, 
and that's not what they want. 

I think everybody agrees that well written tables are more usable 
and accessible than image maps. I hope that the guideline don't give 
another impression. However with CSS positioning instead of 
tables usability and accessibility increase further.

Nir Dagan
Received on Sunday, 26 July 1998 11:41:56 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 16 January 2018 15:33:27 UTC