RE: xml literal and xslt (resend)

Graham:
>
> I'm bothered with all the solutions you propose --

Me too, the alternative is to specify less, and still leave room in which
implementations can differ.

>                                              XML schema uses qnames
> in attributes, so I think we'd ignore this at some risk.  The solution of
> using rdf:parseType="literal ns1 ns2 ..." doesn't seem to be XSLT-safe.

Why not?
I think XSLT preserves namespace prefixes (except when using the aliasing
mechanism).


>
> I find myself foundering on exactly what it means to be XSLT-safe:  it
> seems that any XSLT solution that naively extracts literal content is
> likely to fall foul if this goal.
>