W3C home > Mailing lists > Public > xmlschema-dev@w3.org > April 2002

Re: [Fwd: [Fwd: Schema Design: Composition vs Subclassing]]

From: Jeni Tennison <jeni@jenitennison.com>
Date: Thu, 4 Apr 2002 23:55:23 +0100
Message-ID: <100228118767.20020404235523@jenitennison.com>
To: paul.w.daisey@census.gov
CC: costello@mitre.org, Simon.Cox@csiro.au, dwc@mitre.org, plj@mitre.org, dhoward@mitre.org, <xmlschema-dev@w3.org>, <gml30.rwg@opengis.org>

> I thought Jeni Tennison's comments on potential disadvanges of
> definition by composition were quite thorough. The "killer"
> advantage of design by subclassing was the one I was going to raise,
> that it explicitly retains type inheritance identification; E isa D
> isa C isa B isa A, in the extension / restriction base="" attribute.
> This is an advantage not just in XML, but also in using XML/Schemas
> as the quasi-MDA base for object definitions in languages like Java,
> for example with a product like Castor, whereby the type inheritance
> identification is passed to the generated language classes. I'm not
> sure I entirely agree with the conclusion: " information about the
> model groups that have been composed to create a content model is
> lost(I believe)". I think that it is possible to write xslt to
> determine the implicit type inheritance hierarchies in XML/Schema
> types defined by composition by comparative inventory of their
> contents. But in any case the type inheritance information is
> certainly less accessible than in the design by subclassing case.

Possibly. I was talking about the fact that there's nothing in the
PSVI that keeps tabs on what model groups were used within a
particular complex type - the model group references are resolved to
their model groups to create the particles in the content model of a
complex type. So given a schema (in the abstract sense rather than an
XML representation), I don't think that you can tell where a
particular model group definition has been referenced.

You're right that given an XML representation of a schema, you could
use XSLT to work out which complex types and model group definitions
share particular model group definitions and construct a hierarchy
that way. But in general I don't think that information (about the
original XML representation of the schema, given there is such a thing
in the first place) will be available within a schema-aware processor.

(The same argument applies to attribute groups, which are resolved to
attribute uses when a complex type definition schema component is



Jeni Tennison
Received on Thursday, 4 April 2002 17:55:25 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:55:56 UTC