Eric van der Vlist, CEO, Dyomedea
Date: Thursday, July 29
Time: 10:45am - 11:30am
Data driven classes are a well-known subject and a unique advantage of dynamically typed programming languages as shown last year at this conference (see for instance Data-Driven Classes in Ruby). XML documents are by nature non-regular and constantly evolving data structures and a field where data driven classes greatly improve our ability to manipulate XML documents.
This talk describes a simple Python implementation of such a class that dynamically maps XML documents exposed through the DOM into a hierarchy of objects following the principles of data driven classes and late binding (the nodes are converted into objects when they are accessed for the first time). Attendees will be able to understand the internals of data-driven XML
libraries and eventually to write one.
What was van der Vlist's inspiration for this session? "The starting point has been Data-Driven Classes in Ruby, a presentation given at OSCON 2003 by Michael Granger and David McCorkhill.
"I have found the concept of classes dynamically created after data
stored in a database or anywhere so elegant that I have wanted to apply
it to my domain of choice, XML. Without knowing it, I had already used XML data driven libraries, but to
be sure I understood the concept, I have wanted to write one that would
be 'more data driven' and 'more late binding' and started to do so in
the plane taking me from Portland back home in Paris..."
van der Vlist describes the coolest part of his session: "When you are using statically typed languages (such as C++, Java or C#),
you need some kind of schema if you want to automate the binding of XML
documents to object trees. This is, for instance, the approach of Java's
JAX-B that generates Java classes from W3C XML Schema schemas.
"The approach of 'data-driven classes' shows that dynamically typed
languages (such as Python, Ruby, Perl...) can dynamically bind XML
instance documents to object trees which classes are dynamically created
at run time from what is found in the documents themselves. This provides a very lightweight and 'pythonic' method to access to the
information available in XML documents without using any XML specific
stuff such as SAX, DOM or schemas exactly as this information was an
"This is also bringing the principle of 'late binding' which is one of
the most important benefits of XML to its extreme and the library that I
have developed actually does that binding when you first attempt to use
one of these objects."