tree addressing language (was Re: New/old pattern syntax, why can't we have both ?)

Subject: tree addressing language (was Re: New/old pattern syntax, why can't we have both ?)
From: "James Tauber" <jtauber@xxxxxxxxxxx>
Date: Tue, 25 Aug 1998 22:52:29 +0800
I've changed the Subject: (but not the subject :-)

-----Original Message-----
From: Paul Prescod <papresco@xxxxxxxxxxxxxxxx>

>James K. Tauber wrote:
>> 1. XSL patterns are generally interested in a class of nodes (eg "all the
>> elements of type 'emph'") whereas XPointers are generally interested in a
>> specific node.
>So we're talking about a function that returns one node or one or more.
>Clearly the first is just a superset of the second.

The only reason I raised it was that it effects choice of defaults and the

XPointer presently doesn't assume you mean ALL candidate nodes unless you
use the term 'all'. In XSL, it makes more sense to assume ALL in the absence
of an instance number. Which should a general tree addressing language
(my leanings are towards XSL in this instance (no pun intended)).

>XPointers could be merely constrained from resulting in a result set of
more >than one node.

I see this as an unnecessary constraint. I might want to link all instances
of a node to the one resources, for example.

>> 2. The processing models are likely quite different. XPointers are used
>> find nodes in a document whereas with XSL, you start with a node and try
>> find the pattern that matches with the greatest specificity.
>That's just one way to look at it. You could think of XSL dividing up all
>nodes into a sets (the result of queries). One node can reside in multiple
>query result sets. The set that most specifically describes the node is
>the one that wins.

Good point. I haven't checked how currently implementations do it.

>> I don't think XPointers are sufficient but I don't think it would take
>> to make them so.
>I agree. This is an important harmonization project. Addressing, filtering
>and querying are different views of the same thing, and should share a
>syntax and data model.

Yes. I might play around with recasting XPointers into XSL patterns and see
how it might work. Or is the XPointer or XSL WG already doing this?

James Tauber / jtauber@xxxxxxxxxxx
Lecturer and Associate Researcher
Electronic Commerce Network             (
Curtin Business School                  (
Perth, Western Australia                (

 XSL-List info and archive:

Current Thread