RE: [xml-dev] Bug in XPath 2.0 spec?

["Michael Kay" <mike@saxonica.com>]

The XPath 2.0 grammar as presented in the specification does require look-ahead to parse unambiguously, and even with lookahead, it requires some extra-grammatical rules to disambiguate certain constructs such as

 

/ union /*

 

or

 

4 treat as item() + - 5

 

Occurrence indicators are one such case, and the rules for resolving the ambiguity are described in A1.2 (Constraint: occurrence-indicators).

 

Michael Kay

http://www.saxonica.com/

---

From: Timothy Washington [mailto:timothyjwashington@yahoo.ca]
Sent: 19 October 2008 15:35
To: xml-dev@lists.xml.org
Subject: [xml-dev] Bug in XPath 2.0 spec?

I think there might be a bug in the XPath 2.0 spec. I was trying to implement it in Java (using Sablecc). I get a shift/reduce error on a Multiplicative expression (when it is a treat expression followed by a +). This can conflict with the possibility of an occurence indicator (just a + sign). I fixed a few other similar errors and found I just had to be more precise in my grammar file. But this seems to be an honest ambiguity in language itself.

Is this a conflict in the spec? I've attached my error log and grammar file.

Tim

__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com