Re: namespace node implementation

On Friday, Oct 24, 2003, at 00:56 Europe/Berlin, Kay, Michael wrote:

> > > The tricky thing is that when you come to serialize, the XML 1.0
> > > serialization is
> > >
> > > <c xmlns:ns2="NS2"><b xmlns:ns1="NS1" ns1:x="X"/></c>
> > >
> > > but the more accurate XML 1.1 serialization is:
> > >
> > > <c xmlns:ns2="NS2"><b xmlns:ns2="" xmlns:ns1="NS1" ns1:x="X"/></c>
> >
> > ? which of the 1.1 documents implies this? is there something other
> > than the december namespaces cr or the recent xml cr, or is this
> > somehow expressed in c14n?
>
> It's not implied by the namespaces specifications, it's implied by the 
> XPath serialization spec, which places the requirement on 
> serialization that it should "round trip": that is, it should generate 
> the textual XML which, when reparsed, will give you back the data 
> model you started with. The data model allows a namespace to be in 
> scope for a parent element without being in scope for its child 
> element; this cannot round-trip through a 1.0 serialization, but it 
> can round-trip through a 1.1 serialization.

as there are no references to a prefix binding or "ns2" in the actual 
encoded identifiers in the "b" element, i presume that this devolves to 
the "QNames in content" argument, which is the topic of the related 
post.

>
> Michael Kay
>

Received on Friday, 24 October 2003 02:51:10 UTC