RE: Alternatives to XML Schemas

[Michael Brennan <Michael_Brennan@Allegis.com>]

It will be interesting to see how much interest there really is in a PSVI.
There is a lot of activity going on around data-binding, and most approaches
do the data-binding in response to SAX events (or even at a lower level).
They don't build some in-memory representation of an Infoset, validate it
and add additional info, then process it. They use some mapping layer to map
atoms in the XML stream to application-specific structure more geared toward
the intended use of the data, then process it. The schema is used in a
code-generation phase during development. At runtime, any necessary
validation is done by application code generated in part with metadata from
the schema.

We'll have to see which approaches really predominate. I don't know the
answer to that. Here where I work, we've started heading down the latter
path. I can see value in the other approach, though; especially with
applications intended to deal with XML structures not anticipated at the
time of implementation. Such applications will need to be able to deal with
generalized structures that can express a PSVI in some manner.

> -----Original Message-----
> From: Rick Jelliffe [mailto:ricko@allette.com.au]
> Sent: Wednesday, March 07, 2001 1:16 PM
> To: xml-dev@lists.xml.org
> Subject: Re: Alternatives to XML Schemas
> 
> 
> 
> > Unfortunately, there are many implementors who will heed 
> your advice. I
> have
> > so far identified only a few implementations of XML Schema. Each is
> focused
> > on one very narrow use case (typically validation) and each 
> supports only
> a
> > subset of the spec.
> 
> The reason why XML Schemas will initially be used for 
> validation most is
> because, until we get DOM and SAX APIs for the PSVI or XSLT 2 
> and XPath2
> come along (for better or worse), all PSVI solutions are 
> proprietary or
> hidden inside an application and so invisible.
> 
> I would guess there will be three waves of XML Schemas 
> adoption: first for
> valiation (wave already broken), second for the generation of 
> interfaces
> (this wave is on fast apon us) and third for PSVI processing 
> (only just on
> the horizon, practically speaking, because the requirements 
> are still being
> gathered).  If your need is only for validation or the generation of
> interfaces, then you will clearly have a different evaluation 
> of the feature
> set of XML Schemas than someone (assuming they actually exist 
> :-) who needs
> a PSVI (or if you could just as easily use the schema for 
> providing type
> information to queries rather than to the PSVI data itself.)
> 
> Cheers
> Rick Jelliffe