Re: text transformations in CSS?

Glenn Adams (
Mon, 11 Dec 95 10:34:24 -0500

Date: Sat, 9 Dec 1995 21:57:53 -0600
From: (Scott E. Preece)

I guess I'm disappointed that the combined HTML+stylesheet
doesn't aim higher, for all that it's a useful advance over
HTML alone.

What you are looking for isn't unreasonable, however, it isn't
necessarily reasonable to merge the specification of style (i.e.,
presentation semantics) with the specification of transformations
(i.e., general operations general content/structure).

Merging the two specifications makes for an extremely complicated
set of functionality that would have little hope of seeing acceptance
or, even more importantly, interoperability.

Before you go to far on this topic, you need to read ISO/IEC 10179
Document Style, Semantics, and Specification Language (DSSSL), which
aims at providing precisely what you are seeking. However, it does
so by employing three different languages built upon a common
expression language (which is based on a side-effect free subset of
Scheme). The three languages being a Query Language, a Transformation
Language, and a Style Language.

An effor is currently underway to specify an implementation subset
profile of DSSSL known as DSSSL Online. What you are looking for is
more adequately addressed by this effort than CSS. Furthermore, you
should not encourage CSS to attempt to deliver such functionality,
that is, unless you are willing to wait a year or two to see anything.
This is not to say that DSSSL Online will take a year or two, since
it is already starting with a very well defined specification and a
set of vendors who are cooperating to implement an interoperable
subset of DSSSL.

One of my greatest concerns regarding CSS is that people are asking
for too much too quickly and that in an effort to satisfy the "cool
factor" and whims of marketing and users, vendors are going to produce
a completely inconsistent set of CSS functionality thus yielding
non-interoperable documents.

The authors of CSS and the W3C style community had better begin to
think very hard about limiting CSS1 functionality to the minimum
set of agreed interoperable features if CSS is going to take off.

Glenn Adams