Re: DESC and TABLE: descriptions for groups of elements

I originally sent this message to the HC list instead of the IG list. This
was an error, and I apologize to those who will now be receiving the
message for a second time.

 ---------- Forwarded message ----------
Date: Thu, 23 Oct 1997 12:27:58 +1000 (AEST)
From: Jason White <jasonw@ariel.ucs.unimelb.EDU.AU>
Subject: Re: DESC and TABLE: descriptions for groups of elements

Al's META and LINK enhancements are intended to enable the logical order
of a document's contents to be preserved in circumstances in which the
successive blocks of text are not contiguous in the HTML file. This should
allow the correct reading order to be marked up, even if the author has
used tables to control the visual layout of the document. Of course, such
abuse of table markup should be discontinued in favour of style sheets,
which offer a much better means of specifying visual presentation, and are
in complete conformity with the semantics of HTML.

Al's linking proposal also enables external resources, including
descriptive text, to be associated with tables and other elements of the
document.

I doubt that the TITLE attribute of the DIV element is likely to be
misused. Consequently there is no need for ALT (or a comparable attribute)
to be added to DIV, and the same reasoning applies to LONGDESC, given Al's
linking proposal as already discussed.

I agree with Dave Raggett's observation that screen readers are becoming
increasingly anachronistic and should not be considered as the paradigm of
accessibility on the basis of which HTML design decisions are to be made.
Backward compatibility is no doubt desirable, but not at the expense of an
efficient, long-lasting design. The WAI should not introduce features into
HTML which are likely soon to be rendered obsolete as screen readers are
dispensed with in favour of genuine braille and audio formatting software,
which has access to both markup and style sheets.

I think that the current proposals achieve an appropriate balance between
backward compatibility and future-oriented thinking. LONGDESC is probably
the best example of this approach, as is Al's linking proposal to the
extent that it is intended to remedy the access problems which result from
the abuse of table markup.

Received on Thursday, 23 October 1997 01:16:21 UTC