- From: David Woolley <david.woolley@bts.co.uk>
- Date: Sun, 15 Oct 2000 23:35:07 +0100 (BST)
- To: w3c-wai-ig@w3.org
kynn@idyllmtn.com wrote: > > * The creation of an "icon:" pseudo-URI scheme similar to the You mean pseudo URL. They would appear to be real URIs, but URNs, rather than URLs. I'm not sure whether the definition of URN allows different servers to serve the resource in different forms, although I don't see that as an unsurmountable hurdle. More of a problem is that lack of URN support in current browsers. (I think that lynx has open ended scheme proxying, so might cope, but GUI browsers tend not to be so flexible.) However, the definition of URN syntax says that one can use a well known and stable URL as a URN. In the case of trademarks, many trademark owners would probably prefer that people linked to their site rather than duplicating the trademark. This doesn't happen often, which is bad news for an attempt to standardise icons. I still think that characters are less of a hack in terms of the desired semantics, although Unicode seem to have reservations on adding new graphics, and don't want trademarks. Amongst other things, the CSS precedence rules combined with downloadable fonts give more flexibility with regard to control of physical form. > * Alternately (or in addition), extend XHTML entity sets to include > the icon sets (possibly via assigning them to unicode): The trend is towards reducing the number of general entities, basically because you can't rely on browsers actually knowing all the entities in use, an the large number of characters now available to browsers, and which can be addressed using numeric entitities. One could define XML modules, but I think there is some scepticism as to whether the modules concept will work outside of niches. Note that entities are just macros. > > * Or when encountering the following: > > <a href="http://google.com/">Search the Web</a> > > ...n intelligent user agent or server processor could convert that > to iconic markup automatically. Detecting idioms is problematic because most commercial site authors keep changing them (google seems fairly stable, but many search engines keep changing to frustrate attempts to technically bypass adverstising and to frustrate users attempt to shift their gaze directly to the functional part of the page). It's been discussed on the lynx-dev list in relation to coping with people who use javascript: links in contexts which could have had a clean fall back. The problem is that someone has to rapidly respond to new idioms with browser updates.
Received on Monday, 16 October 2000 02:53:18 UTC