Re: C++ version of regexprep.cc

[Andy Adler]

On Thu, 4 May 2006, Bill Denney wrote:

> On Thu, 4 May 2006, Andy Adler wrote:
>
> > On Thu, 4 May 2006, Bill Denney wrote:
> >
> >> I personally think that elements should have a namespace associated
> >> with it only if it's different than the parent.  Then it is simple to
> >> climb the tree to see your particular namespace.
> >
> > That is a good idea. However, I'm not sure if this matches
> > the XML spec. XML namespaces provide a default namespace
> > for the document for all unspecified tags, not for those
> > that differ from the parent (correct me if I'm wrong).
>
> I don't think that's the case.  The general format of a document is that
> you can change the default namespace or use a specific namespace for a
> specific element.
>
> For example:
>
> <doc xmlns="http://my.namespace/uri"; xmlns:his="http://his.namespace/uri";>
>    <mine xmlns="http://other.namespace/uri";>
>      <data/>
>    </mine>
>    <yours>
>      <data/>
>    </yours>
>    <his:data/>
> </doc>
>
> The <data/> that is within mine falls into the namespace
> "http://other.namespace/uri"; because the default namespace was changed for
> all sub-elements under <mine>.  The <data/> under yours falls into the
> namespace "http://my.namespace/uri";, and the his: in <his:data> indicates
> that that data element falls into the namespace
> "http://his.namespace/uri";.  So the default namespace can be changed
> for the tree under any leaf.

Thanks for the clarification.

The basic issue is that if we do XML, we can't just do some ad-hoc
thing. We need to design a good API that does XML correctly.

One consequence, in my opinion, is that the Octave representation
of a loaded XML file must be the same for all xml-info-set
equivalent documents.

--
Andy Adler <address@hidden> 1(613)562-5800x6218