Re:

> That's not true. Without a global metod for asserting it,
> a private communication (maybe built into th code)
> might let it know of the equivalence and therefore dispatch
> a Y document to an X processor.  You can't stop people doing
> that, even though the specs we are talking about don't say how to do it.

In general, at the level of the namespace spec, obviously you can do
that, you can write a conforming namespace application that treats all
namespaces equally if you so wish. But higher level protocols can ban
that. For example an XSL system that allowed you to match an element
in one namespace with an element in the other would not be conforming
to the XSLT recommendation. There is no way you can take a conforming
XSL system (even using an extension namespace) and say that it should
treat elements in namespace X as if they were in namespace Y.
Of course, you can take the code of xt or saxon and change it
to implement that, but then it isn't XSLT any more.


> No - to compare reelative URI references as literals is broken
> as has been shown many times on this list.

The only thing shown to be broken is that if you write stylesheets
assuming the "absolute" interpretation of namespace names and run it
on a system that implements the "literal" interpretation, then bad
things happen. This doesn't actually prove anything about either
interpretation,


David

Received on Monday, 19 June 2000 15:31:46 UTC