Re: Validation API, was: Regarding the vote on XML Schema.

[Wayne Steele <xmlmaster@hotmail.com>]

I think this discussion has happened before.
IIRC, the consensus was that more is needed than a 'pluggable validator' 
API; each schema technology brings its own infoset contributions.

'Validation' can be seen as a special case of an infoset contribution.

What's needed is a mechanism to plug-n-play infoset annotations into a 
document processing stream.

Example:
   I have this XMLDocument.
   I want to validate a fragment of it against this DTD I've got (this also 
fills in some default attributes I need).
   I want to use an XML Schema to 'paint' data types onto the doc.
   I run an XSLT2 transform on the document, keeping the datatype 
information intact in the output document.
   I want to overwrite _some_ of the datatype information in the new 
document, with some specific rules I hacked up.
   I want to validate the document using a set of Schematron rules.

I also want to do this without scribbling on the document instance, because 
it's on a read-only medium and it's huge.

Maybe that is too hard.

I think people are still trying to narrow down what these 'infoset 
contributions' are, and what they look like. Then we can talk about an 
architecture for providing them.

-Wayne Steele



>From: Tony Coates <Tony.Coates@reuters.com>
>
>On 24/04/2001 07:06:48 Rick Jelliffe wrote:
>
> >Merely saying "XML Schemas bad!  RELAX good!" keeps the cart before the
> >horse.  If there is no modularity or ability to plug-n-play with 
>different
> >kinds of schema, then every little engineering trade-off has to be 
>subjected
> >to exhaustive discussion (as in XML Schemas) with no guarantee that the
> >result will satisfy everyone.
>
>Agreed.  What would help us more is a validation API that allows pluggable
>validators, and that allows (multiple) validation of any part of an XML
>document, not just the whole document.  If your Schema wraps a legacy DTD, 
>why
>shouldn't the legacy tags be validated using the DTD?  If you have 
>validation
>code in your application (and there is always some, doing the things that
>DTDs/Schemas can't do), separate it out and build your own pluggable 
>validator.
>That will make your application architecture cleaner.
>
>My major concern in achieving this is that Schemas can be applied to a DOM 
>tree
>(which is very nice), but I'm not sure what tricks there might be in trying 
>to
>do the same with DTDs, which were not designed with that in mind.  Of 
>course,
>you would want not just these two, but also TREX/RELAX, Schematron, and any
>other likely suspects that might come along.
>
>All of this said, is anyone else interested in being able to do this kind 
>of
>thing, or is it all too hard?  My impression is that for a lot of the 
>world,
>DTDs are the now, XML Schemas will do what most people need for the future
>(remembering that most people aren't on the "xml-dev" list ...), and 
>anything
>else tends to be marginalised as a toy for XML weenies.  The existence of a 
>good
>API could change that (just as SAX and DOM made XML parsers accessible and
>acceptable by removing the lock-in), but I'm interested to know whether the 
>rest
>of you see things the same way.  All comments gratefully received.
>
>      Cheers,
>           Tony.
>========
>Anthony B. Coates
>Leader of XML Architecture & Design
>Chief Technology Office
>Reuters Plc, London.
>tony.coates@reuters.com
>========

_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com