- From: Jason White <jasonw@ariel.ucs.unimelb.EDU.AU>
- Date: Wed, 24 Sep 1997 09:43:44 +1000 (AEST)
- To: WAI HC Working Group <w3c-wai-hc@w3.org>
Clearly it would be best to build on the solution that has already been developed in CSS 2. I share the concern expressed in Daniel's mailing list thread, that authors might try to convey significant information in the image which is to be used as a list bullet. A means of associating a textual label or description with the image, for use in non-graphical media, would therefore seem to be warranted. This is another example of a situation in which a metadata approach based on HTTP would offer the best solution. However, there is a strong tendency toward reliance on markup as the means by which to convey the textual alternative to a graphical resource. Indeed, given the ALT attribute, the design of OBJECT and the LONGDESC proposal, there is hardly any need for a metadata approach, except in two situations: (1) frames; and (2) the list image bullet in CSS. It would seem reasonable to suppose that HTML authors, who would be responsible for taking advantage of ALT and OBJECT, might not be willing to make the extra effort of preparing metadata so as to provide textual substitutes for certain images, merely for the sake of the anomalous circumstances in which the HTML markup and CSS do not offer a solution. I have no objection to pursuing the metadata strategy, thereby making available two alternative methods of associating text with an image. Nevertheless, it appears inconsistent to rely on markup for most purposes but to insist on the implementation of a metadata scheme in a limited range of circumstances in which HTML and CSS 2 currently fail to give scope for associating text with a graphical resource.
Received on Tuesday, 23 September 1997 19:44:05 UTC