Re: [CSS21] Please endorse xml:id

----- Original Message ----- 
From: "Ian Hickson" <ian@hixie.ch>
To: "Norman Walsh" <Norman.Walsh@Sun.COM>
Cc: <www-style@w3.org>
Sent: Friday, June 24, 2005 5:37 PM
Subject: Re: [CSS21] Please endorse xml:id


|
| On Fri, 24 Jun 2005, Norman Walsh wrote:
| >
| > I think it would be of benefit to the community if the CSS 2.1 draft
| > suggested that implementations of CSS 2.1 applied to XML documents
| > <rfc2119>should</rfc2119> support the xml:id specification.
|
| It's not appropriate for a spec to take positions on other technologies.
| Technologies should succeed or fail on their own merits, not because they
| were dragged kicking and screaming into implementations by virtue of other
| specs requiring them.
|
| CSS doesn't require particular graphics, video, or audio formats; it
| doesn't require the DOM or any particular scripting environment; it
| doesn't require the CSSOM; it doesn't even require XML or HTML. Which is
| perfectly sensible, since CSS is orthogonal from both and can be applied
| to any tree-based system. I believe the only technologies that CSS
| requires are UTF-8 and the Unicode Bidi algorithm -- the former is
| required because you have to represent CSS in _some_ common form if you
| want interoperability, and the second is required because CSS supports
| bidi layout and thus has to reference some bidi algorithm (or invent its
| own, which seems suboptimal).
|
| Given the above, it would be very strange indeed for xml:id support to be
| required. Why not, say, xml:base? Or XInclude? Or XHTML, XForms, XFrames?
|
| In any case, it makes no difference what we require. Implementators ignore
| this kind of requirement if it isn't in line with what they want to
| implement. To exit CR we need to show two interoperable implementations;
| we'd just end up dropping any requirement like this that wasn't met. And
| that begs the question: why have such requirements in the first place?
|

Ian, having these two statements:

"CSS is orthogonal from both and can be applied to any tree-based system."
and "To exit CR we need to show two interoperable implementations"

Suppose I have implementation of system of UI widgets configurable by
It is a) tree based b) uses CSS level 2.1 but not in full just because
some attributes do not fit in the model.

This system cannot be considered as "interoperable implementation" just
because it cannot demonstrate CSS in full.

Therefore "interoperable implementation" means implementation of HTML
renderer with CSS 2.1. Too many features there are HTML specific or with
HTML only in mind.

Am I right?

Andrew Fedoniouk.
http://terrainformatica.com

Received on Sunday, 26 June 2005 01:59:56 UTC