Re: Generating namespace declarations given a result tree?

Subject: Re: Generating namespace declarations given a result tree?
From: John-Paul Sicotte <jps@xxxxxxxxxxxxxxxxxxx>
Date: Wed, 24 May 2000 15:13:19 -0600
On Thu, 18 May 2000 12:18:58 +0700 James Clark <jjc@xxxxxxxxxx> wrote:

> John-Paul Sicotte wrote:

> XSLT says you're supposed to output a tree T so that if you parsed it
> back in again and constructed a tree using the XSLT data mdel you would
> get a tree T' that was the same as T except that
> 
> - the order of attributes in T' and T may be different
> - T' may have namespace nodes that are not in T

I read this several times before I sent the mail but one critical thing
never clicked in:  the tree must be the same including the namespace 
nodes.  I was always wondering what "the same" meant with regards to 
elements (ie. what about prefixes) but if the namespaces nodes must be 
the same all of this becomes clear.

> In this specific case, if you didn't output the namespace declarations
> for jpns[1-3] then T would have namespace nodes that T' did not.  That's
> not one of the differences that's allowed: XSLT only allows the
> converse.  So you must output the namespace declarations for jpns[1-3].

If the namespace nodes need to be the same then the prefixes must 
remain as well.

> If XSLT didn't do this, it would be difficult to use XSLT to transform
> XSLT stylesheets.

Or anything that used xpath expressions as attribute values which was 
my direct concern.

> James

Thanks very much

				John-Paul

----------------------
John-Paul Sicotte
Software Composer
MessagingDirect
mailto:John-Paul@xxxxxxxxxxxxxxxxxxx



 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list


Current Thread