W3C home > Mailing lists > Public > www-ql@w3.org > April to June 2005

Re: Variable references in path expressions

From: TAN Kuan Hui <kuanhui@xemantics.com>
Date: Thu, 28 Apr 2005 17:00:02 +0800
Message-ID: <001201c54bd0$b4e58a70$6402a8c0@pinecone>
To: "Michael Kay" <mhk@mhk.me.uk>, "'Charles Brooking'" <charles.brooking@research.canon.com.au>, <www-ql@w3.org>

> I believe the statement is true, and I don't understand your reasoning for
> saying it isn't. I don't understand the terms "evaluated against" and "blind
> looping".
> $a establishes the focus for evaluating the expression on the rhs of "/",
> but a variable reference has no dependency on the focus. It depends only on
> the set of variable bindings in its static context, and the static context
> for $b is not affected by $a.
> I can't see what's special if $b contains a sequence of xs:QName values. The
> interpretation of an xs:QName value doesn't depend either on the static or
> the dynamic context.
I have a problem with this interpretation. The lhs of "/" iterates each item
to establish the focus for evaluation of each item in the rhs expression. Yes ?
Now if each item is a QName, then this should be interpreted as a
Name test. No ?

If I interprete you correctly, then,

 $book := <book><title>blah</title><author>foo</author></book>,
 $fields := (xs:QName("title"),xs:QName("author"))
 return $book/$fields

would return

and not

While the variables are bound prior to the path expression,
the evaluation of "/" can interprete $b to make
the path expression in a meaningful and consistent

I need to understand how

returning xs:QName("title"),xs:QName("author")

can be seen to be useful as a path expression.

It this is so then the semantics of the path expression
as defined must then otherwise been changed. The
definition should appropriately define the above

Quote : XQuery 3.2 Path Expressions
[Definition: A path expression can be used to locate nodes within trees.
A path expression consists of a series of one or more steps, separated
by "/" or "//", and optionally beginning with "/" or "//".]

> In a conventional path expression such as a/following-sibling::*, how would
> I convert following-sibling::* into a QName?
You can't, syntactically, this is identified by the parser as a Wildcard
and a Wildcard token cannot be represented as an atomic type nor
a node in a sequence. Maybe it can be cast into a string but this will
will have significant impact on the overall grammar.

However, a variable is definitely bound to a sequence; you can
run through its items and produce a sequence of QName before
processing "/". I don't see why this is not implementable
to align with the objective of a path expression.
Received on Thursday, 28 April 2005 09:00:24 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:17:17 UTC