Re: A theory problem

Kay Michael wrote:
> 
> >
> > [James Clark]
> > >> A path expression has the property if
> > >>
> > >> (a) it doesn't use / and the axis is a forward axis, or
> > >>
> > >> (b) it is a / expression, and the left hand operand has the
> > >> "single-level" property and the right hand operand has the
> > >> "stays-in-subtree" property.
> >
> > [Michael Kay]
> > >This is nice and clean, but it doesn't seem to catch
> > >child::A/child::B/child::C.
> >
> > Do you mean as a select or as a match pattern?  As a select,
> > child::A is
> > single-level and child::B has stays-in-subtree; taken together,
> > child::A/child::B is single-level and child::C has stays-in-subtree.
> >
> We're talking about expressions not patterns, since document order is
> irrelevant in patterns.
> 
> James' definition of "single-level" was that a set of nodes was single level
> if they all had the same parent; therefore child::A/child::B is not single
> level. This is why I was trying to use a more general concept of "peer"
> nodes.

You're right: having the same parent isn't quite the right concept and
it's not quite what I implement in XT. The definition should be such
that if E1 and E2 are single-level then E1/E2 is.  I think the right
concept is that there's a constant difference in the levels of the
context node and the result nodes.  More precisely, define the level of
a node to be the number of ancestors it has, and then define an
expression to be single-level if and only if there exists an integer n
such that for any node x, for any node y in the result of evaluating the
expression with respect to x, the difference between the level of x and
the level of y is equal to n.

In XT I implement / as right associative; this makes the left hand
operands of / in child::A/child::B/child::C all be single-level even in
the narrower sense of the original definition I proposed.

James


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