> -----Original Message-----
> From: Francis Norton [mailto:email@example.com]
> Sent: Wednesday, March 28, 2001 7:52 AM
> To: John Cowan
> Cc: Richard Lanyon; firstname.lastname@example.org;
> Subject: Re: CDATA sections in W3C XML Infoset
> We've all learnt to fear and loath the sight of CDATA sections being
> used to output unbalanced markup, but I would like to suggest
> respectfully that dropping the CDATA start and end markers out of the
> infoset as a way of discouraging this practice is overkill.
[Speaking only for myself, not the DOM WG or my employer ...]
CDATA Sections have been decreed "syntax sugar" by the experts writing the InfoSet spec. The DOM has to be useful by ordinary Joe's in the trenches, and will continue to expose CDATA sections ... for all the reasons noted in this thread. <grin>
The DOM will *support* the infoset by bringing terminology (and perhaps namespace declarations) in line with the InfoSet terminology. This doesn't mean that the DOM must deprecate CDATA Sections, entity references, etc., it simply means that the DOM *extends* the InfoSet, as the InfoSet spec suggests.
So, you will be able to put CDATA sections and entities in XML syntax, manipulate them with the DOM, but not validate them with Schema or query for them with XPath or XQuery or style them with XSL. I don't *think* this is a significant problem for real developers ... In the <script> example, you could query/validate/style with the <script> tag rather than the CDATA syntax sugar, right?