- From: Eric van der Vlist <vdv@dyomedea.com>
- Date: Fri, 27 Oct 2000 13:30:07 +0200
- To: Ian Hickson <ianh@netscape.com>
- CC: www-style@w3.org
Ian Hickson wrote: > > On Fri, 27 Oct 2000, Eric van der Vlist wrote: > > The 'style' attribute already existing in XHTML, SVG, MathML, ... > > doesn't specify any namespace and has, since the default namespace > > doesn't apply to attributes, no namespace. > > No. It doesn't have no namespace. An attribute without an explicit > namespace is owned by the element -- namespaces are irrelevant. I meant that the attribute has no namespace property. > > IMHO, one should not be recommending to use an attribute without > > namespace to carry a semantical meaning to be used by tools. > > There is no such thing as an attribute with no namespace, merely global > attributes and attributes particular to specific elements. I find the notion of global attributes introduced in the NS spec very confusing (the so called global attributes are still attached to an element per the DOM, XPath and Infoset [4] specifications which does not even mention global attributes). [4] http://www.w3.org/TR/xml-infoset#infoitem.attribute > > In this case, 'style' is a very common word (in several languages) very > > likely to be defined with other meanings by XML vocabularies. > > Each vocabulary that wishes to use the "Syntax of CSS rules in HTML's > "style" attribute" [1] will have to specify what name they wish to give > that attribute. > > > The clean way would be to define a namespace for this purpose. > > Agreed. > > > This namespace could be the XML 1.0 namespace if the XML Core WG aggreed > > that this practice is generic enough and 'style' would then become > > 'xml:style'. > > I personally would think this would be a very bad idea. The style > attribute generally does not belong in content -- the whole _point_ of CSS > is to separate the content from the style (one of the many reasons XSL:FOs > are a bad idea IMHO). I wanted to mention this as an option, though. > > Otherwise, another option could be to create a XML Style specification > > and 'style' would then become 'xstyle:style'. > > Or we could just introduce a namespace for CSS-related attributes and > elements. This idea has been mentioned before. > > <root xmlns:css="some css namespace"> > <foo css:style=" /* some inline style */ "/> > <css:style title="preferred stylesheet"> /* an embedded stylesheet */ </css:style> > <css:style title="alternate stylesheet"> /* another one */ </css:style> > </root> > > > > In both cases, the major drawback would be that it would be incompatible > > with the current XHTML, SVG, MathML specs. > > Just like the lang vs xml:lang thing, or the XLink vs <a> thing. [2] or xml:base :) The problem will probably occur more and more frequently in the future since a (re)modularization of common functions existing in different vocabularies is likely to happen and elements and attributes likely to migrate between namespaces ! Eric > > I think we are, here, facing exactly the same issue that the XML Linking > > WG is trying to solve through its "XLink Markup Name Control" [2] note. > > > > [1] http://www.w3.org/TR/2000/WD-css-style-attr-20001025 > > [2] http://www.w3.org/TR/xlink-naming/ > > -- > Ian Hickson )\ _. - ._.) fL > Netscape, Standards Compliance QA /. `- ' ( `--' > +1 650 937 6593 `- , ) - > ) \ > irc.mozilla.org:Hixie _________________________ (.' \) (.' -' __________ -- ------------------------------------------------------------------------ Eric van der Vlist Dyomedea http://dyomedea.com http://xmlfr.org http://4xt.org http://ducotede.com ------------------------------------------------------------------------
Received on Friday, 27 October 2000 07:31:51 UTC