Re: The role of the style attribute: HTML WG official position

On Tue, 29 Feb 2000, Steven Pemberton wrote:

> So the best way forward for the Web, should the community feel
> there is a need for mixing document structure with styling, would
> be to have a namespace for each styling language. 

No, not a namespace.  A "definition space", maybe.  I'm aware that the
Namespace shibboleth is part of official W3C canon, but suppression of
the fact that alpha-renaming doesn't solve any real problems need to
stop at some point.
 
> Then you could say
> 
>     <p css:style="..." xsl:style="...">
> 
> and the user agent can choose which it wants to use.

It is *not* that simple.  All such examples sneakily exploit the fact
that human beings are reading them, and to humans, strings such as
"css" and "xsl" already mean something (or suggest something).  If
there were a case for namespaces, it should be made on the basis of
generic examples, so as not to burden the software with a DWIM problem
that has *not* been solved.  Consider the restatement: 

: Then you could say
: 
:     <p foo:style="..." bar:style="...">
: 
: and the user agent can choose which it wants to use.

This is semantically identical, but it raises the real issue of what
'foo:style' could be,to begin with.  A Namespace URI doesn't solve
that problem, unless the URI *itself* is recognized, not merely as
some namespace thingummy or deus ex machina, but as an identifier of a
*stylesheet notation*.  But, once it's recognnized that the *real*
requirement is the recording of the notation, whether one uses
namespaces or anything else is mere detail.


> Styling is not in the purview of the HTML WG, and it would not be
> up to the HTML WG to define those namespaces.

Whew!  That's a relief:)


Arjun

Received on Wednesday, 1 March 2000 19:06:07 UTC