W3C home > Mailing lists > Public > www-style@w3.org > October 2003

Re: CSS 4?

From: Ian Hickson <ian@hixie.ch>
Date: Sun, 26 Oct 2003 10:18:58 +0000 (UTC)
To: Dylan Schiemann <dylans@yahoo.com>
Cc: www-style@w3.org
Message-ID: <Pine.LNX.4.58.0310260953110.14732@dhalsim.dreamhost.com>

On Sun, 26 Oct 2003, Dylan Schiemann wrote:
>> BECSS bindings and similar technologies are designed to add stylistic
>> (albeit quite involved) look and feel to documents.
>> They are not designed to add semantics to the document.
> Yes, but the flexibility of XBL seems to encourage it (the script
> element for example).

No more so than the flexibility of class attributes encourages its own
abuse, IMHO.

>> For example, BECSS-like technologies could be used to bind the logic to
>> XForms controls. No new semantics -- the XForms controls already have the
>> XForms semantcs -- but a look and feel (and in this case logic) that
>> implements those semantics.
> Sure, but why do this through css, and not through the dom using
> addBinding()

Because (and this is the litmus test that proves the binding is
presentational and not semantic) you want different bindings depending on
the media type of the document. And even on the same media, you want to be
able to use different bindings for different alternate stylesheets. For
example a "cute" stylesheet could use a binding which showed bouncing
bunnies next to :invalid form fields, and hovering on the bunnies could
cause the bunnies to walk into the form field and "fix" the error; while a
"business" stylesheet could use a binding that should showed a red border
and displayed a dialog box when the user exit the field.

Both cases are conveying the same _semantics_, which were originally set
out in the XForms specification. However, they are giving the user a
different presentation, look, and feel. They are thus stylistic.

> (besides the obvious fact that it is often easier to describe a set of
> elements through css selectors than through DOM, though maybe DOM XPath
> resolves this?  maybe DOM Style/CSS needs to be fixed to be able to
> select the same things as css selectors?  getElementsByCSSSelector ?)

It's also a lot easier to do:

   <style type="text/css">
    @namespace url(http://example.net/);
    section { binding: url(bindings#collapsable-sections); }


   <script type="text/javascript">
    var sections = document.getElementsByTagNameNS('http://example.net/',
    for (var index = 0; index < sections.length; ++index)

> Seems that using uri for 'content' is comparable depending on what is
> allowed in uri (which isn't clear in css2.1 in my opinion).  For
> example, could the uri be a document fragment?  If so, could it contain
> a stylesheet to define style rules for the inserted content?

The result of:

   div { content: url(http://example.com/#test); }

...is _visually_ equivalent to:

   <iframe src="http://example.com/#test"/>

...although of course the semantics are very different.

> By the way, http://www.w3.org/TR/CSS21/generate.html#propdef-content
> does not list <uri> in the condensed Value list (not sure if this was
> noted previously on the list).


> I agree that it is nice to have this.  What happens when bindings are
> added using the addBinding method?  Is this the equivalent of adding a
> statement at the end of the stylesheet?  Can a binding be added/removed
> using the DOM 2 CSS StyleSheet interface's insertRule/deleteRule instead
> of DocumentXBL's addBinding?  I guess my real question here is how do
> DocumentXBL and StyleSheet interact?

The above should all be explained at:


If it isn't, it would be defined before such a spec went to REC.

> Say for the sake of argument that I were to define an xml syntax lxbl
> (lesser xbl) whose sole purpose was to list pairs of event type and
> event handlers (in other words a small subset of xbl).  Would it make
> sense to use css to bind these defintions to selectors?  I would say no,
> that it should be done through the DocumentLXBL interface, or through
> some other css-like document that supports selectors, but that isn't
> css.  At least that would feel "cleaner" to me.

Well, if the language was designed to do stylistic things, then it would
make sense to have CSS bind to it.

For example, a binding which handled onmousemove and changed the style of
the element based on the position of the pointer with respect to the
element, would be purely stylistic, and so would belong in CSS. If, on the
other hand, the binding defined the form validation logic for an onsubmit
event, then it would make sense for it to either be defined directly
within the document, or have the external logic bound to the form using
inline script. (Note that the first example is basically a fancy :hover,
while the second is a fancy onsubmit="" handler. The bindings logically
belong in the same place as their simpler cousins.)

The question is merely "is the binding stylistic". Most, in my experience,
are, and thus IMHO those belong in the stylesheet, with other stylistic

> Whatever, though, I think I'm in the minority on this.

Being in the minority isn't a problem, if you're right! In this particular
case, I'm not convinced that you are, though. :-)

Ian Hickson                                      )\._.,--....,'``.    fL
U+1047E                                         /,   _.. \   _\  ;`._ ,.
http://index.hixie.ch/                         `._.-(,_..'--(,_..'`-.;.'
Received on Sunday, 26 October 2003 05:19:00 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:24 GMT