- From: Charles McCathieNevile <charles@w3.org>
- Date: Sun, 25 Nov 2001 20:58:51 -0500 (EST)
- To: Jim Ley <jim@jibbering.com>
- cc: WAI GL <w3c-wai-gl@w3.org>
On Sun, 25 Nov 2001, Jim Ley wrote: Whether websites are illustrated or not, makes no difference to me, there's no burden on me at all, I'm very much encouraging it, but the abscence of any techniques is currently very limiting especially for my own personal sites where I don't have designers behind me doing the work. CMN Yes, I agree that the lack of techniques, and more particularly of success criteria, is a problem. Working on these seems like a particularly important area, given that we appear to be so far behind at the moment. Kynn: > So what do you propose to solve this supposed problem? Jim: Well, there are a number of ideas all of which I think have problems, this alternate content and on a paragraph basis (single idea per parargraph, all ideas illustrated.) therefore we have the opportunity to add an alt/longdesc type attribute to the P element, I don't like that much, but it's an idea. CMN And an intereating idea. I think that the development of it so far is a bit simplistic - there are some things that are clearly an illustration of a particular idea, and some things that are general illustration of a page, but there are some things that fill both roles at once. In order to use something like classes I think the answer is to use RDF to describe thje relationships. As you know, I think that developing Web browsers in the medium term without RDF parsers is a mistake, although it is possible to connect a browser through a proxy that does this part of the processing to provide what the user needs. This is the basis of the IBM Webi technology, and of most WAP services by the way, so it isn't just vapourware - it has been impemented on very large scale. Jim: Perhaps the simplest approach would be to require that all such illustrations have a class of "alternate" * or something, therefore I could simply hide them (and them alone) by a user stylesheet, something that is already well supported in current browsers. e.g. <p>Chickens like to fly. <img class="alternate" ... ></p> This would also allow page authors to choose to have the alternate content hidden by default on pages (if they wished). There are though still bandwidth issues with such as current browser CSS implementations still downloads images with display:none - but that's still less development work than supporting new attributes. * It would actually need to be more unique than this to avoid potential collisions with authors chosen names, who are not aware of the guidelines. The proposition that I can just disable images, leads me to another area which I haven't seen addressed with this alternate content - what are the alt etc. attributes? if it's a text explanation of how this illustration is trying to get the idea across then I will probably struggle just as much - ie "A man dropping a piece of paper on to the floor with a large red cross through it" - would still make be more confusing than just the paragraphs "don't drop litter" instruction. CMN I think this is the difference between "alt text" and a longdesc - the appropriate text replacement for the image is "don't drop litter" (nor rubbish ;-) and the description of what the icon is should be a longdesc. Cheers Charles
Received on Sunday, 25 November 2001 20:58:55 UTC