Re: [xml-dev] RE: James Clark: XML versus the Web

[Doug <doug.duboulay@gmail.com>]

On Tue, 14 Dec 2010, Michael Kay wrote:
> On 13/12/2010 09:25, David Carlisle wrote:
> > On 13/12/2010 03:15, Henri Sivonen wrote:
> >> The script element in thehttp://www.w3.org/1999/xhtml  namespace has
> >> already been implemented. How would a new processing instruction be
> >> better?
> >
> > XML is designed (mostly) not to use fixed element names. If you are
> > styling xhtml, xhtml:script is OK, but if you are styling docbook, or
> > some personal xml vocabulary or anything else other than xhtml, then
> > adding a processing instruction will maintain the validity of your
> > source, but adding an xhtml:script will not.
> 
> Perhaps something in the HTTP header would be even better than a
> processing instruction, as it would avoid disturbing the XML content
> entirely.
> 

You can already get XML+javascript into the browser, using 
the "document reboot" trick, which is to say, add an XSLT stylesheet PI 
to your XML that transforms it into an HTML shell document (with 
embedded onload() javascript elements) that contains your entire 
original document as an "XML data island". This works perfectly well 
in firefox, opera, chrome and safari. Konqueror 4.3.5 and IE 8 (MSXML3), 
of course, cock it right up :-/ 
Konqueror doesn't even attempt to handle the PI and IE mis-interprets 
any XML namespace <title> elements as html title elements. Allegedly 
using javascript in IE it is possible to choose the version of MSXML 
you want to use, but unfortunately the damage is already done by the 
default MSXML3 before any javascript is loaded.