W3C home > Mailing lists > Public > www-style@w3.org > April 1999

RE: CSS and XLink

From: Didier PH Martin <martind@netfolder.com>
Date: Mon, 12 Apr 1999 16:27:46 -0400
To: "Simon St.Laurent" <simonstl@simonstl.com>, <www-style@w3.org>
Message-ID: <000301be8522$ece0f2a0$311fcdcd@total.net>
Hi Simon,

Were your experiments with XSL or CSS?  I seem to think they were XSL, but
I'd still be very glad to work on CSS with the problems you've encountered
in mind.

Some with CSS and most of them with XSL. However we experimented the same
problems or similar problems with documents having vocabulary overlap but no
specific name space. We tried to simulate a real Web situation. With CSS
what happened is that a rule applied to Document A is also applied to
document B even if document B has some vocabulary elements similar but with
different meaning or author meaning intention. We deliberately introduced
inter language confusion like French markup with an English text. Or an
English written document with French markups (this could happen if the
author is French using a French DTD but publish a text in English). So in
one document we got the markup <convention> meaning a collective agreement
and in the other document the markup <convention> meaning seminar, trade
show. In fact, we discovered this by accident when two people used two DTD
with the same vocabulary and two meaning. The CSS in the first document
worked on the second but the layout did not got proper meaning (from a
semiology point of view). We discovered at that time or re-discovered, to be
more specific, that formatting do convey meaning or re-enforce the
document's meaning.

Did you actually deal with inclusion of non-XML materials, like graphics?
Or was it all XML documents/fragments?

Yes but with non XML material its easier to deal with except if they are
tagged differently. For example, we got a French marked document with the
word <image> but with its French meaning "backup". But when both document
where using a HTML vocabulary to identify images we didn't got any problem.
In fact problems occurred with identical vocabulary but with different
meaning or different classification schemas. The problems where most of the
time with the markups. With XML fragment, the problem could even be worse
because the formatting would be totally not matched. But with name space,
the problems where resolved.

Finally, does the W3C's fragment working group have anything going that
might address the problems you encountered with fragments?

I am reading again the document (ref from Daniel Koger) to find if there is
some provisions here. But, most of our problems came from name collision,
association of a particular element meaning and its associated style. I have
now to check if navigation errors could be induced by vocabulary overlap.
Most of style problems occurred with the link inclusion mode. When the whole
document is changed, no problems - it works well. But to include a resource
from an external document is not as easy if the two DTD contain similar
elements with different meaning.

We tried to use Jon Bosak documents with Chinese markup  to see if confusion
may occur there too, but we are not expert in oriental language. Rick
Jeliffe could be in a better position than us to tell if this phenomenon
could happen with oriental languages using chinese markup encoding. Our
conclusion is that for the Web, name space support is mandatory to prevent
style errors. This means that we can reuse documents with potential
vocabulary overlap without problems. In the worse case, the style won't
work. But we can build a style sheet where all elements are defined within
their name space. Now, CSS may have problems with name space. Am I wrong on
this? In  the case of XSL all elements have to be appended with their name
space identifier to resolve ambiguities.

With this experiment we discovered that people would markup their document
with markups in their own language. They could publish in a foreign
language. Thus, as an example, the content may be English but the markups
could be French. If you take French and English the potential for name
collision is greater than for English to Russian. But maybe some collision
could occur with Norwegian,Swedish or Danish marked documents. So, in the
scenario where we want to re-use the content of a document marked with a
different language, problems occur. And even in the same language, content
re-use could be limited with same DTD different meaning. A style sheet
either created in CSS or XSL will have the same problem if the style is
associated to a certain meaning and used to re-enforce that meaning. We got
out of context layout problems, even if the content made sense. I hope I was
enough clear with my examples. If not, just tell me where this need

Didier PH Martin
Received on Monday, 12 April 1999 16:29:19 UTC

This archive was generated by hypermail 2.3.1 : Monday, 2 May 2016 14:26:50 UTC