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

Re: Relax NG, Thoughts, etc.

• To: Michael Kay <michael.h.kay@n...>
• Subject: Re: Relax NG, Thoughts, etc.
• From: Burak Emir <Burak.Emir@e...>
• Date: Mon, 29 Nov 2004 09:56:09 +0100
• Cc: xml-dev@l...
• In-reply-to: <20041128212719.JNVK2076.aamta04-winn.mailhost.ntl.com@Turtle>
• Organization: EPFL
• References: <20041128212719.JNVK2076.aamta04-winn.mailhost.ntl.com@Turtle>
• User-agent: Mozilla Thunderbird 0.8 (X11/20040913)

Michael Kay wrote:

>>There are 5 standard Validation Languages:
>>
>>1. DTD
>>2. XML Schemas
>>3. RelaxNG
>>4. Schematron
>>5. OASIS CAM
>>
>>But the following technologies have been created to only 
>>utilize one of them
>>(XML Schemas):
>>
>>a. XSLT 
>>b. XPath 
>>c. XQuery 
>>d. SOAP
>>e. WSDL
>>
>>    
>>
>
>Although all of 1-5 might do validation, only one of them does data typing,
>and XSLT/XPath/XQuery are using XML Schema as a data typing language, not
>primarily as a validation language.
>  
>
Daniela  already answered to that.

>However, integrating output validation into XSLT is, I think, the biggest
>"quick win" that schema integration brings, and there's no technical reason
>why this part of the story shouldn't work equally well with other schema
>  
>
There are actually many reasons why validation should not work equally 
well with other schema languages.

>languages. You won't see anything standardised in this round, but it
>wouldn't be surprising to see vendor extensions in this area. In principle
>it should be straightforward for an XSLT processor (even a 1.0 processor) to
>support result tree validation using any class that implements the JAXP 1.3
>ValidatorHandler interface. 
>
>  
>
Granted, dynamic validation (=at runtime) is always possible, but it is 
certainly not a big help. Errors are discovered after having gone to 
production, doesn't sound good to me.

But one should not forget to mention that static typing (using the 
validation language to determine at compile time) is certainly more 
useful, and simply not possible with some of the mentioned approaches.

I am sure you will not try to add exact static typing based on Relax NG 
for XSLT in the next version of Saxon? There are papers that analyze 
this for subsets of XSLT, but doing it exactly, for the full language, 
is - as far as I remember - undecidable.

cheers,
Burak

-- 
Burak Emir

http://lamp.epfl.ch/~buraq

• Follow-Ups:
  • RE: Relax NG, Thoughts, etc.
    • From: "Michael Kay" <michael.h.kay@n...>

• References:
  • RE: Relax NG, Thoughts, etc.
    • From: "Michael Kay" <michael.h.kay@n...>

• Prev by Date: Re: Relax NG, Thoughts, etc.
• Next by Date: Re: Seeking SAXException pattern
• Previous by thread: RE: Relax NG, Thoughts, etc.
• Next by thread: RE: Relax NG, Thoughts, etc.
• Index(es):
  • Date
  • Thread