RE: SOM

Steffen,

I don't think that the fact that you use SAX for _instance_ data should
steer you away from a schema model which happens to be in the DOM spec.

If you look at the Content Models spec, you'll see that it isn't very
DOM-centric; the meta-model really corresponds to the _information set_ for
schema, not the XML representation.

You won't find any discussion of using a DOM content model for validating
with SAX in the spec; that would be "out of scope".  But this seems like a
very natural evolution; perhaps a future version of SAX might specify it.
Since a schema definition must be held in memory for practical usage, a
"SAX-native" schema model doesn't really make sense, unless the DOM Content
Model is inadequate in some way.

[I'm not an expert in this area... if anyone is better informed on this
stuff, please comment!]

-Ed Staub


> -----Original Message-----
> From: David Valera [mailto:dvalera@pcl-hage.nl]
> Sent: Thursday, December 14, 2000 9:22 AM
> To: xmlschema-dev@w3.org
> Subject: RE: SOM
> 
> 
> > Sorry to jump in at the end of this thread, I've been otherwise
> > occupied for the last few days, but this is too misleading to go
> > unchallenged: an XML Schema is _not_ an XML file.
> 
> I got scared when I saw this...
> 
> > One schema may
> > correspond to many schema documents, they are _not_ the same thing,
> > and cannot be made so.  Schemas involving more than one 
> namespace are
> > of necessity represented by at least a pair of schema documents.
> 
> But now I can relax. :-)
> 
> You are indeed correct, since an XML schema can be divided in 
> different
> documents. An element declared in one part (file) of the 
> schema could have a
> type which is defined in an different part (read file) of the 
> schema. DOM
> would need to be aware of the semantics of XML schema's to access the
> corresponding components.
> 
> I would rather see this developed as an extension of DOM than as an
> 'independant object model' like Steffen is proposing.
> 
> > The appropriate objects to access from a Schema API are schema
> > components.  The spec. mandates that these be available as 
> part of the
> > PSVI.  The DOM WG is working on making the PSVI available, there is
> > still discussion about whether this will be a schema-specific
> > facility, or done via a generic infoset-access paradigm.
> 
> PSVI? I am not sure if I missed something, but what does PSVI 
> stand for?
> 
> David Valera
> 

Received on Thursday, 14 December 2000 10:37:48 UTC