Lists Home |
Date Index |
Joe English wrote:
> I'm ambivalent about this. QNames in content sound like a
> Really Bad Idea on the face of it, but on the other hand
> of all the RECs to come out of the W3C, XSLT is the only
> one in which namespace issues don't cause a tremendous
> headache to end users. It shifts all the pain onto XSLT
> implementors (thanks, Mike :-).
> I'm not sure how this could be accomplished without QNames
> in content.
The problem isn't QNames in content. It's how you declare
prefix=>namespace mappings. One way to do this is to define
application-specific mapping mechanisms, such as those that allow the
declarations only on root elements or with specific child elements such
<Namespace Prefix="foo" URI="http://www.foo.org/" />
The problem with these is that they inherit the problem that dogged the
original namespace PI: code that is unaware of the namespace declaration
mechanism can't safely copy fragments.