The current trend in web development is to have pages produced using a four-layer sequencial model like such: The client receives the output of the Layout layer which is often intertwined with the Logic and Style layers. Under this model, languages like PHP and mod_perl have thrived. However, in the next year or so, it is likely that most sites will be using a new, more efficient model, where the Content, Style and Layout are sent directly to the client, like such: This reduces memory and bandwidth requirements. Proxying and caching are also more effective under this model, especially for dynamic content. In the former model, our data are generally stored in relational databases and queried from a scripting language using SQL. With such radical evolution of the web, we have to ask ourselves whether this is still the best method for extracting Content. This paper discusses some of the drawbacks of current Web Application frameworks such as JSP/J2EE and PHP, and introduces a new framework called XMLDB.

Pete Ryland attended the University of New South Wales under the Co-Operative Education Scholarship Scheme in Electrical Engineering. He majored in Computing and Networking and for his thesis he wrote a Java Decompiler. As part of the Co-op Program, Pete worked for Alcatel Australia and was subsequently offered a full-time Java/C++ development position with them upon graduation. He has also worked for a web solutions comapny, @www (the "Dubz") and a London finance company called Reech Capital whose main products are a set of risk analysis web applications. He is now working for Tangozebra, a digital advertising and marketing company.

Last modified: Mon, 9 Jun 2003 21:09:27 +0200