Re: RDF and CSS - was last calls for css3-background, css3-fonts, css21

At 12:31 PM 2002-08-01, Sean B. Palmer wrote:

>> can you write and attach RDF with a CSS file?
>
>You could, but it would be extremely complex. Although the BNF productions
>for CSS are grouped pretty much according to the semantics of each of the
>parts, you'd have to invent a CSSPath for that to be useful. Chaals's ideas
>of adding a new property ("background-longdesc"? I don't think you want
>people to be using chunks of HTML in CSS) to CSS and deploying decent image
>formats (PNG, SVG, etc.) seems to me to be on the mark.

Oops.

I suspect that Sean may be interpreting Lisa more literally than Lisa 
intended, in terms of "attach RDF with a CSS file."

I believe that the leading-candidate method of applying the RDF which 
associates a description of the background is not by reference via
CSS path into the CSS at all but by the URI of the first-class web
object that the CSS processor binds as background to some base-language 
element.

The association of a base-language element instance with the background
image-resource is implied by the style algorithms and the sheaf of 
pertinent style sheets.

The action of the style processor may be reduced to rule form as a binding
of 
 - the property-value pair (background, someImage)
to
-  a match-pattern generated by selectors in style rules and the action of 
the cascade to isolate a surviving style rule per base language element instance

>As for "associate td.header with a standard  h2", I'm not 100% sure what
>you mean, but I think you're in the realm of CSS pre-processing.

Pre-processing, but not necessarily CSS pre-processing.
Perhaps XSLT pre-processing.  One application is to generate a new 
base-language syntax tree for an NN4-compatible view of the page.  
I don't believe that CSS will do that for you.

This leaves the matter of how to frame some RDF that makes assertions about 
what rhetorical role this element is playing that makes the binding to this
background image appropriate.  I am unclear how far up the generic/specific
or is_a ancestry of the element instance one can climb with RDF assertions.
It would be nice to be able to bind sense-clarifications to match-patterns
and not only to instances.  But that may be a 'nice' that fails to make
the grade in terms of lowest level language, c.f. our experience where only
'headers' is implemented among the cell/header association tools in HTML 4.

 Accessible TABLEs from W3C/WAI (FedStats pitch)
 http://www.w3.org/2002/Talks/06/24-US_FedStatsWorkshop/slide1-0.html

Al


>--
>Kindest Regards,
>Sean B. Palmer
>@prefix : <http://purl.org/net/swn#> .
>:Sean :homepage <http://purl.org/net/sbp/> .

Received on Thursday, 1 August 2002 13:31:50 UTC