Re: CSS & DSSSL together (Was: Re: <insert> and external entity

[Chris Lilley, responding to Jay Bazuzi:]

| >  UA's which only want to do HTML wouldn't need DSSSL support 
| I wouldn't go that far. DSSSL could well be useful with HTML too.

I suspect that DSSSL is too complex to find much use with HTML in
practice.  On the other hand, all the proposals I've seen for CSS
evolution are utterly inadequate for the solution of general SGML
formatting (including print formatting, by the way).

The key distinction here (which I think that many people keep losing
track of) is whether the document can contain embedded formatting
information or not.  Attributes that assign formatting-related classes
or style names to particular elements are (from the large-scale,
long-term document management point of view) just as much embedded
formatting information as specific font names or type sizes, because
they prevent one from completely changing the rendition of a document
based purely on its content, structure, and semantic categories.  A
stylesheet language that is suitable for specifying the formatting of
generic SGML documents must be able to completely change the
formatting without requiring any changes to the document.  DSSSL is
designed to provide the mechanisms needed to do this.  For a markup
like HTML that is moving inexorably in the direction of a kind of
abstract RTF, this level of capability is probably not necessary.

People who are interested in DSSSL and dsssl-o (a subset of DSSSL
suitable for generic SGML Web browsers) should check out the



Jon Bosak, SunSoft
2550 Garcia Avenue, Mountain View, CA