Re: [xsl] Calling subdocuments with processing directives

[Steven Ericsson-Zenith <steven@xxxxxxxxxxxxx>]

It is because there is no defined order that I find it counter  
intuitive to have to make what is essentially and order  
specification. From a functional point of view a key really should  
aggregate IMHO.

With respect,
Steven

--
Dr. Steven Ericsson-Zenith
Institute for Advanced Science & Engineering
http://iase.info
http://senses.info



On Oct 26, 2007, at 10:41 AM, Michael Kay wrote:

> > The "default" you refer to is on retrieval. It is not clear
> > to me why that context is relevant in the case of keys. Why
> > do you have to specify the source of the value? That seems
> > very odd to me. The name of the key includes the source path?
>
> The key definition includes a pattern - a condition that nodes must  
> satisfy
> in order to be eligible. You seem to be suggesting that you would  
> like the
> search in some sense to be global - to search all known documents.  
> There
> seems to be something of a closed world assumption there. All  
> documents
> previously loaded using the document() function? But what does  
> "previously"
> mean in XSLT - there's no defined order of execution.
>
> Consider
>
> <xsl:variable name="doc1" select="document('foo.xml')"/>
> <xsl:variable name="bbb" select="key('k', 93.7)"/>
>
> Should $doc1 be searched? Why? It typically isn't loaded until it's  
> needed,
> and if no-one ever refers to $doc1 then it never will be loaded.
>
> Michael Kay
> http://www.saxonica.com/