Re[2]: comments on 2002-12-12 XHTML 2.0 WD

At 2002-12-30T19:53+0300, Alexander Savenkov wrote:-

> > It is however not entirely clear to me that icons do not belong in
> > style sheets - they are, after all, essentially entirely
> > presentational. On the face of it, icons of this sort do not fit
> > into the CSS framework particularly well, as they seem applicable
> > only to whole pages (or usually collections of pages), whereas CSS
> > would tend to allow an icon to be suggested for any element(s) in a
> > document; OTOH it is certainly not impossible to envisage ways in
> > which icons for elements within a document might be used (e.g. an
> > automatically generated outline).
>
> Have a look at
> http://www.w3.org/TR/2002/WD-css3-ui-20020802#box-model. According to
> the draft there's no need in "icon" value for <link>'s 'rel'
> attribute.

Sadly not: that describes displaying an icon in place of an element's
normal content when rendering the document as usual. The desired
functionality here is to provide an icon to represent a resource
 outside of this context. I suppose it could be shoehorned into this
 model, by saying that the UA is effectively, for example, rendering
 the entire document with a UA style sheet such as

html {
  display: icon !important;
  icon: url(file:///some/default/icon)
}

but that seems horribly cumbersome. It is tempting to extend the icon
property to the effect that the icon(s) can be used in external
 contexts whatever the value of the display property; but I think that
 could interfere with the already proposed usage, especially if such
 behaviour were indeed not restricted to the root element (and I see no
 reason to impose such an arbitrary constraint). I feel a separate
 property would probably be better (though one might also want to
 rename the icon property at the same time to avoid confusion).


Tim Bagot

Received on Thursday, 9 January 2003 18:02:44 UTC