Hi, Martin v. Löwis wrote: >> BTW, from the POV of objectify, generating Python classes from a schema would >> basically mean inferring a document instance from an XML Schema (sort of a >> meta-model to model transformation). I find that an interesting relation, but >> maybe that's just me... > > It's just you. It would *not* be a meta-model to model transformation, > but a meta-model-to-meta-model one.
It obviously is a meta-to-meta model transformation to generate Python classes from a schema, but "inferring a document instance from an XML Schema" is not. > The schema defines a type system, > just as a set of Python classes does. Instances of the schema (i.e. > a document) then correspond to a set of instances of these classes. Objectify doesn't generate code. Instead, it comes with an extensible meta-model that resembles the basic Python type system, and which gets mapped on a tree at runtime. So the act of inferring the classes from the schema is actually linked to the instance, not the meta model. And the link is done through validation, which assures that the document really is an instance. So we end up with classes that represent an instance of a meta-model. There is no intermediate step of a meta-to-meta model transformation. Stefan _______________________________________________ XML-SIG maillist - XML-SIG@python.org http://mail.python.org/mailman/listinfo/xml-sig
