Re: why split? [was RE: XSL intent survey]

Subject: Re: why split? [was RE: XSL intent survey]
From: Håkon Wium Lie <howcome@xxxxxx>
Date: Thu, 26 Nov 1998 01:26:18 +0100 (MET)
Chris Maden wrote:

 > High-quality formatting is an iterative process; consider the case of
 > footnote placement.  Sometimes placing a footnote on a page causes the
 > text to which it is attached to flow to the next page.  The formatter
 > must have enough information to deal with this.  Page number
 > references are similar, and consider conditional chapter openers that
 > vary whether they are on recto or verso pages.
 > The CSS formatting model is good for what it addresses.  It could be
 > extended for intelligent formatting objects like these, but I think
 > its fundamental model of kind-of-thing { property: value } would
 > either make that extension awkward, or would have to be removed.

But XSL uses exactly the same model. In XSL the "kind-of-thing" is a
formatting object, while in CSS the "kind-of-thing" is most often an
element. It can also be a page, pseudo-class or pseudo-element -- the
first-line pseudo-element in CSS is an example setting style based on
the layout of the content, not on its structure. The point I try to
make is that "intelligent" formatting is orthogonal to the
property/value axis.

 > So the output of a transformation from semantic to styled language
 > needs to be intelligent formatting objects, capable of expressing
 > restraints and relationships; why not use XML?  The formatters can
 > then leverage all the work done in XML processors to their advantage.

The issue of syntax is very important while at the same time almost
irrelevant. It's irrelevant since parsers are trivial to write
compared to the pains implementors go through to get their display
engines right. It's very important since people will be editing style
sheets by hand for the forseeable future.

 > So why not split the language into two specifications, XTL and XFM?
 > It's been discussed, but there's a fear, justified I think by recent
 > vendor actions, that the transformation will be implemented and not
 > the formatting model, trapping us all in this single-scroll ghetto
 > forever.

The work on XSL has increased awareness of ensuring good printouts
from the Web, and I don't think users will tolerate for long that
their browsers print lousily. One aspect of the recent debate on this
list is whether the noble goal of good printing justifies joining two
technically independent specifications. Personally, I prefer technical
arguments to political ones when setting the scope of a specification.



H   å   k   o   n      W   i   u   m       L   i   e
World     W      i     d     e       Web  Consortium

 XSL-List info and archive:

Current Thread