Action Item: Editors to add a note to the spec about the subset of XPath in XSLT

Hi Felix,

Here's my suggestion for our action item

	Editors to add a note to the spec about the subset of XPath in XSLT

recorded in http://www.w3.org/2006/12/13-i18nits-minutes.html

Let me know what you think.

I would suggest to add the note to 2.1.2 (Global Approach). On the one hand
that's where
we first extensively talk about "selectors". On the other hand, it is in a
non-normative part.

Cheers,
Christian
===

Caveat Related to XSLT-based Processing of ITS Selector Attributes

The values of ITS "selector" attributes are XPath absolute location paths.
Accordingly, the
following is a legitimate value:

	verbatim/descendant-or-self::*/@*

Unfortunately, values like this cause trouble when they are used in
XSLT-based processing
of ITS where the values of the ITS "selector" attributes are used as values
of "match" attributes
of XSLT templates. The reason for this is the following: "match" attributes
may only contain
a restriction/subset of XPath expressions, so-called "patterns" (see
http://www.w3.org/TR/xslt#patterns).
Basically the following restrictions hold for patterns:

- only axes "child" or "attribute" allowed
- "//" or "/" possible
- id() or key() function possible
- predicates possible

Using only XSL patterns in ITS "selector" attributes helps to avoid this
issue. In many
cases, this is possible by using predicates. The value above may for example
be
rewritten as follows
	
*[self::verbatim]/@* | verbatim//*/@*