W3C home > Mailing lists > Public > www-xpath-comments@w3.org > January to March 2002

RE: Something missing with XSLT and XPath

From: Kay, Michael <Michael.Kay@softwareag.com>
Date: Wed, 20 Feb 2002 16:17:46 +0100
Message-ID: <DFF2AC9E3583D511A21F0008C7E6210602679C63@daemsg02.software-ag.de>
To: "'Lee Humphries'" <Lee_Humphries@softworks.com.au>, www-xpath-comments@w3.org
Cc: "'Anders Berglund'" <alrb@us.ibm.com>, "'Norman Walsh'" <ndw@nwalsh.com>
Issue 143 in the functions and operators document [1] refers to the
requirement for a tokenize() function. Thank you for providing this use case
which (in my view) helps to justify the requirement.

[1] http://www.w3.org/TR/xquery-operators/#need-tokenize-function

(Anders/Norm: could you cross-refer the issue to this mail item, please?)

Mike Kay

> -----Original Message-----
> From: Lee Humphries [mailto:Lee_Humphries@softworks.com.au]
> Sent: 19 February 2002 22:31
> To: www-xpath-comments@w3.org
> Subject: Something missing with XSLT and XPath
> 
> 
> Hi,
> 
> I've been making fairly heavy use of XPath and XSLT for around a year
> now.  But I've found what appears to be a hole in functionality that I
> don't see fixed in version 2 of XPath or XSLT.
> 
> We have an interest in using the 'list' simpleType in XML 
> Schemas, yet I
> can't find any function to readily perform the tokenisation 
> of any such
> element or attribute. For example, we have an element that contains a
> list of identifiers of varying formats, I want to be able to loop
> through the list extracting each identifier in turn and processing it.
> At present I've resorted to recursion, I call a template 
> passing in the
> list, the template extracts the first token, performs the required
> action, and then calls itself with the remainder of the list.
> Unfortunately this limits what I can do, as some XSLT processors won't
> allow you to build up XML fragments within a variable if those XML
> fragments are built within a new template scope (e.g. MSXML). 
>  However,
> I've also had need to do things like building up dictionaries from the
> contents of elements - so it's not just lists of IDs or keys 
> that I need
> to be able to tokenise.
> 
> Ideally what I would like to see is some XPath string function that
> allows you to identify a string to tokenise, a set of tokenising
> characters and an index of the token desired (if there was some way of
> easily working backwards through a set of tokens that would be really
> handy also).  Alternatively, an XPath function that could be 
> used within
> the select in a for-each that results in iterating across the tokens.
> 
> Regards,
> Lee Humphries
> SOFTWORKS Australia
> email: Lee_Humphries@softworks.com.au
> phone: +61-7 3511 7000
> Level 1, 33 Park Road, Milton, Queensland 4066, Australia
> 
Received on Wednesday, 20 February 2002 10:17:59 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 3 October 2007 16:05:54 GMT