Re: [xsl] XPath context evaluation

[Victor Toni <xsl-list@xxxxxxxxx>]

Michael Kay wrote:
> > In the meantime I found a nearly trivial solution which lets 
> > my keep my
> > environment:
> >
> > "subsequence(//info,index-of(//info/id, .), 1)"
> >
> > These functions were easy to implement and solved the issue I 
> > had. Since performance doesn't count for this use case I am 
> > glad it works :-)
> >     
>
> This isn't 100% equivalent to //info[id = current()], unless you make
> assumptions about the data, for example that there is no more than one
> match.
>   

This is indeed the assumption made and is even required in my use case.
> But if you're using XPath 2.0, then what's wrong with
>
> for $x in . return //info[id = $x]
>
> (which is equivalent: in fact it's essentially what Saxon compiles the
> original code into)
>   

I am not really using XPath 2.0. Actually it's dom4j with some extension 
functions for Jaxen.since Jaxen 1.1 beta X is not XPath 2.0 "ready" it 
does not support all the features (e.g. current() ).
After writing the functions it's a one-liner to register it to the 
function context making it very easy for me to add functionality (with 
the limitations given  by the current Jaxen model)
I have written some XPath 2.0 functions for Jaxen on my own which seemed 
trival to write and still supported by the current Jaxen model 
(string-join(), subsequence(), index-of() ) and some more not belonging 
to the standard e.g. an XSL like sort function which uses additional 
XPath expression to determine the sort order.
Especially the sort function is crucial for my use case and as long I 
can work around Jaxen's limitations I prefer not to bother about porting 
it to another XPath implementation :-)

Victor