Last call, XSL 1.0

This message is sent in reply to the last call for comments for XSL
1.0 and represents Opera Software's views.

This is the second specification coming out of the XSL WG. The first
one, a transformation language known as XSLT, has already become a W3C
Recommendation and now the group proposes a syntax for "formatting
objects" written in XML (called XSL-FO in this message).

The formatting objects contain information about how a document is to
be presented. Also, they contain hyperlink information. A document
written in XSL-FO is therefore at the same level of abstraction as an
HTML document consisting of presentational elements (e.g. FONT and BR)
and A elements.

Indeed, many pages on the web only consist of presentational elements
and links. These pages have been considered a threat to accessibility
and device-independence in the past, and W3C has spent considerable
efforts trying to educate UA implementors and content providers of the
benefits of style sheets and semantic markup. In HTML 4.0,
presentational elements were deprecated in favor of style sheets.
Recommending the use of XSL-FO on the Web means a reversal of W3C's
position on these matters; XSL-FO is not a style sheet language, it's
exactly the opposite. It should be a warning signal to W3C when
content providers intend to use one of their specifications as a
"semantic firewall" [1].

Without a syntax to express formatting objects in, XSL-FO would not be
a thret to accessibility and device-independence on the web. We
therefore propose to express formatting objects in something other
than XML. The only technical reason why formatting objects are
expressed in XML today is that XSLT can only output XML. By extending
XTL to express FOs in some abstract, non-syntaxed manner the problem
can be avoided, and the valuable work that the XSL FO subgroup has
been doing on formatting will be able to move forward.

This way, the XSL group would fulfill its charter to "specify a style
sheet language for XML and other structured markup languages". As the
current proposal stand, we believe the group has gone outside its
charter and that the result will be a web with less style sheets and
more presentational markup.

For an extended argument, see [2].

Besides our problem with having a syntax for formatting objects, we
would like to note the following:

 - from our own experience we find it hard to evaluate a specification
   on formatting without having a conformance/test suite and several
   interoperable implementations. We suggest that the specification
   does not move forward until interoperable implementations exist. 

 - several of the formatting properties described redefines the
   semantics as described by CSS2. We assume the XSL FO subgroup think
   their solution is better, and that may very well be the case.
   However, it will be hard for formatting engines to support two
   slightly different interpretations. In other words, reuse of code
   will be difficult as long as there are differences. Also, this is
   not in the spirit of the charter:

    "The intent is also that within XSL, the formatting properties and
     values of CSS1/CSS2 can be used with their current meaning."

 - the shorthand properties have been given their own conformance
   level. It seems as if the specification encourages implementors not
   to support shorthand properties. Again, this is not in the spirit
   of the charter.

 - making XML attributes out of CSS properties has also been done by
   another W3C WG. MathML 1.0 contains a set of "CSS-compatible
   attributes". However, their solution differs from the XSL solution
   in subtle ways that damages interoperability and questions W3C's
   ability to coordinate.


[1] http://www.mulberrytech.com/xsl/xsl-list/archive/msg02343.html
[2] http://www.opera.com/people/howcome/1999/foch.html

Regards,

-h&kon

Chief Technology Officer                                Opera Software
Håkon Wium Lie                     http://www.opera.com/people/howcome
howcome@opera.com                                gets you there faster

Received on Sunday, 30 April 2000 09:02:16 UTC