re: an xslt question re modeling namespaces

[james anderson <james.anderson@attglobal.net>]

I have received one response to my original question, which drew
attention to issues which arise where "nodes" are displaced from their
original context elements.
It was not clear in the response whether the suggeted issue concerned
ns-nodes or qname-values, but that doesn't matter, as one must handle
displacement so as to ensure that the binding is of indefinite extent in
either case.

Support for which one can implement in various ways.
One of which is shallow prefix-namespace bindings.
I ask, why does the generic specification prescribe this as the
necessary model?

Are there things one needs to do with namespace bindings which leave no
other alternatives, or which suggest convincing efficiency advantages
for shallow bindings?

...