Lee Provoost

Planning

2004-2005

• Software Technology Colloquium

Period 1

• Functional Programming
• Programming Correctness
• Grammars and Parsing

Period 2

• Software Architecture
• Database Architecture

Period 3

• Distributed object systems
• Research on Semantic Web
• Working for a software company

Period 4

• Complex Data
• Research on Service-Oriented Architecture
• Working for a software company

2005-2006

• Software Technology Colloquium

Period 1

• Software Engineering
• Seminar: Networked Objects
• STC paper + talk: The evolution of the Java Memory Model

Period 2

• Implementation of programming languages
• Seminar: Semantic Web (from CKE)
• Thesis Proposal

Period 3

• Master Thesis

Period 4

• Master Thesis
• Philosophy of Computer Science

Software Technology Colloquium

Title

The evolution of the Java Memory Model

Abstract

The Java Memory Model specifies the legal behaviors for a multithreaded Java program. Although it influences every Java programmer, unfortunately for most people it is an unknown concept. Even the targeted audience (virtual machine, compiler and processor implementers) does not have a uniform understanding of the Java Memory Model, because the original model is hard to interpret. This leads for instance to virtual machines that violate the constraints of the model, resulting in unexpected behavior of Java code. The big problem when you start reading papers and articles related to the Java Memory Model is that you should really pay attention to the timeframe when they were written. A good example is the confusion about the semantics of the volatile keyword. This paper tries to give an evolution of the Java Memory Model, starting from the original Chapter 17 from the Java Language Specification of Java 1.0 to the implementation of the JSR-133 specification in the latest Java 5. To clearly illustrate the problems and give a better understanding of the Java Memory Model, a big part of this paper will deal with the famous lazy initialization/double-checked locking programming idiom.

Paper

JMM.pdf: The Evolution of the Java Memory Model

Thesis Project

Topic/Area

Service-Oriented Architecture, Enterprise Content Management

Project

Project title: A SOA-enabled Enterprise Content Management system
Advisor: dr. A. Bijlsma, prof. J. van den Berg
Start date: 8 November 2005 (first meeting)
End date: 28 June 2006 (thesis defense)

Description

Service-Oriented Architecture is a new way for designing distributed systems and will probably become the de-facto standard. It tries to bring a solution for the problem that for instance CORBA tried to tackle, i.e. interoperability between heterogeneous systems.
Enterprise Content Management is regarded as a key asset for large enterprises to efficiently manage the flow of huge amounts of data. Several vendors have identified the importance and the advantages of using Service-Oriented Architecture and they are gently "servicing" the functionalities of their Enterprise Content Management system.
This thesis proposes a novel approach for the design of an Enterprise Content Management system. It shows that the concepts of Enterprise Content Management fit nicely with those of Service-Oriented Architecture: document flows can be represented by BPEL processes, software components by Web Services and complex features like human involvement and process time-out are already built-in features of SOA frameworks that we get for free.
However, Enterprise Content Management systems are a critical part of large enterprises and have often high-demanding non-functional requirements like performance, security and maintainability. This results in several issues like service granularity, single sign-on and reliance on remote interfaces. This thesis project tries to identify all those problems and propose solutions. The concepts are tested in an application that wants to serve as a proof of concept.

Document

thesis_public.pdf: Thesis Document