RE: Namespace-Alias + XSLT2 feedback

> The following comments are forwarded from my colleague Eric Bratton at
> Software AG:
> XSLT 2.0 draft feedback:
> Get rid of the example stylesheet and output in section 
> "8.1.3 Namespace Aliasing".  The input is missing, the output is wrong
> (missing namespace declarations in a section dealing with 
> namespaces!),
> it is too complex, XSLT-FO hasn't been introduced at this level, and
> it serves no identified purpose.  A simple stylesheet which produces
> another minimal xsl stylesheet would be a far better example.

Good point; I'll try to do a new example.
> Mention in "8.1.3 Namespace Aliasing" which prefix should be used in
> the output document.  In my example stylesheet above, the prefix used
> in the output document was "out".  What is wrong with either using
> "xsl" (stylesheet-prefix) or creating your own prefix?
> One answer is, of course, QNames hidden in textnodes that end up in
> the output.  I think the prefix choice and rationale should be
> explicitly stated, at least in a note.

I don't think we can prescribe which prefix should be used in the result
tree (because prefixes aren't held in the data model), but we could
certainly add a note suggesting that the "result-prefix" is preferred.
Certainly, users get confused by products such as Saxon that use the
> Also in "8.1.3 Namespace Aliasing", mention that aliasing supercedes
> [xsl:]exclude-result-prefixes and [xsl:]extension-element-prefixes
> processing.  Excluding a namespace before aliasing it yields nothing.

I think it's actually the use of a namespace in an element or attribute in
the result tree that causes xsl:exclude-result-prefixes to have no effect
(rather than this being explicitly due to the use of namespace aliasing),
but I agree this could be clarified.

Mike Kay

Received on Monday, 18 February 2002 16:04:53 UTC