Re: xml-stylesheet base resource identifier

On Mon, Oct 02, 2006 at 04:01:55PM +0100, Henry S. Thompson wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> In this case, however, this is the language which *XML Stylesheet*
> already refers to for its semantics, and how that reference is meant
> to operate is just not clear in this case, at least not to me.  I can
> make credible (to me) arguments both that the stylesheet should be
> fetched from http://www.w3.org/Style/my.css _and_ that it should be
> fetched from http://www.example.org/demo/dummy/my.css, supposing that
> the document itself were being retrieved from
> http://www.example.org/demo/dummy/toughOne.html
> 
> The first is correct, because *XML Stylesheet* says a) Use HTML
> semantics and b) you can turn (xhtml) LINK elements into
> 'xml-stylesheet' PIs -- inverting that gives me
> 
>  <link type='test/css' rel='stylesheet' href='my.css'/>
> 
> where the <base.../> will have the HTML-specified impact.  (Ignoring
> the inversion, we get a _third_ possible interpretation -- throw an
> error, use of relative URI _before_ <base. . ./>)

  This is similar to assuming that a
    <meta http-equiv="Content-type" content="text/html; charset=EUC-JP" />
would override the default character encoding detection of XML-1.0 .
Specs have to be layered, and the generic specs, and hence generic
code need to take precedence if you want to benefit from the standardization.
The analogy with link was I guess about the mechanics of the link not
about the specificities of HTML4. Maybe they used too strong words to
illustrate the similarity to the link rel='stylesheet' construct.

  I doubt any XSLT implementor got it wrong, the good point is that it should
be really easy to set up a test case and run it against XSLT implementations
or browsers implementations, maybe this will prove me wrong but I would be
really surprized !

Daniel

-- 
Daniel Veillard      | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
daniel@veillard.com  | Rpmfind RPM search engine http://rpmfind.net/
http://veillard.com/ | virtualization library  http://libvirt.org/