• To: "'Jeni Tennison'" <jeni@j...>,"'bob mcwhirter'" <bob@w...>
• Subject: RE: Quick Xpath
• From: "Michael Kay" <michael.h.kay@n...>
• Date: Thu, 18 Jul 2002 09:17:29 +0100
• Cc: <xml-dev@l...>
• Importance: Normal
• In-reply-to: <43991867019.20020718001529@j...>
• Reply-to: <michael.h.kay@n...>

> It would certainly be annoying if, for example, (//foo)[4] 
> didn't return the 4th foo element in the document, because it 
> wouldn't be compliant.
> 

On the other hand, an XPath processor given the expression //foo is free
to return the results in any order. XSLT defines that the results will
be processed in document order (e.g. by apply-templates, or for-each, or
copy-of), but that's an XSLT rule, not an XPath rule: a different
environment that uses XPath isn't obliged to have the same rule. 

If the DOM interface for XPath specified that nodes could be returned in
any order, this would be a conformant use of XPath, though many users
would find it irritating.

Michael Kay
Software AG
home: Michael.H.Kay@n...
work: Michael.Kay@s...  

• Follow-Ups:
  • Re: Quick Xpath
    • From: Jeni Tennison <jeni@j...>

• References:
  • Re: Quick Xpath
    • From: Jeni Tennison <jeni@j...>

• Prev by Date: Re: Article: Keeping pace with James Clark
• Next by Date: Re: Schema Namespace name, schemaLocation, and Schema V ersioning
• Previous by thread: Re: Quick Xpath
• Next by thread: Re: Quick Xpath
• Index(es):
  • Date
  • Thread