[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: What are the characteristics of a good type system for XML

• To: John Cowan <cowan@m...>
• Subject: Re: What are the characteristics of a good type system for XML?
• From: bry@i...
• Date: Wed, 14 May 2003 21:38:12 CET
• Cc: xml-dev@l...

> I agree that regexes are not enough, but 
supplemented with ranges and precision
> for suitable types, they do almost the 
whole job.  Integrity checks are important,
> but I haven't seen anything yet that is 
really satisfying in this respect except
> full Schematron processing.
> 
> 

whenever I think about what kind of types 
xml requires I think first of tree 
descriptions, which is what xpath provides, 
and then regex for strings where xpath is 
weak. 

I don't think of it necessarily as document 
validation but like is possible with 
Schematron partial validation relative to 
the instances of a type. 

let us suppose that one had a Schematron-
like language with assertions etc. that have 
ids. 

one declares a type date with a list of 
regexes for allowable date values. 

one declares a type MeetingDate which is an 
element name "Date" that inherits from date 
for its text content and has two required 
xpaths, a parent element Meeting, and a 
following sibling | preceding sibling Time.

the idea behind having ids for assertions is 
that one could have a syntax for removing 
assertions from an inheriting type and 
replacing them with others. 

With a regex extension function for xslt I 
think this would be reasonably easy to 
implement in the same way that skeleton1-
5.xsl is used to do a sample Schematron 
implementation. There would undoubtedly turn 
out to be a lot of other things needed 
later, but I sort of think its nicer to find 
out what people want instead of second 
guessing them. Although it seems that a lot 
of people are supposedly clamoring for the 
xsdl integration in xpath 2.0, xslt 2.0 that 
everyone here is groaning over... I would 
just like to find some of these xsd 
integration proponents (not sitting on a W3C 
working group) but out in the xml wilds. 

Anyway I just figured I'd remark on this 
idea that's been kicking around in my head 
for a while, mainly cause you mentioned 
Schematron. 

• Prev by Date: RE: XML:lang, clarification demanded
• Next by Date: Re: What are the characteristics of a good type system for XML?
• Previous by thread: Re: What are the characteristics of a good type system for XML?
• Next by thread: RE: What are the characteristics of a good type system for XML?
• Index(es):
  • Date
  • Thread