Tuesday Afternoon

1:30 pm - 3:00 pm
Convention Center — Ballroom B-C
Session A
Invited Talk: Software: The New Realities
Tom DeMarco, Atlantic Systems Guild

We live in an age of acceleration. Whatever the formula was for business success a few years ago, it won’t work today. Today there needs to be more and more work crammed into less and less time. There are fewer people doing more and doing it faster in less space with less support and with tighter tolerances and higher quality requirements than ever before. So we have spent the last decade becoming ever more efficient. Now enter the need for change. In the super-accelerated corporation, meaningful change of direction is almost impossible. The very improvements that we made to go faster and cheaper have undermined our capacity to make any other kind of change. An organization that can accelerate but not change direction is like a car that can speed up but not steer. In the short run it makes lots of progress in whatever direction it happened to be going. In the long run it’s just another road wreck.

Tom DeMarco is a Principal of the Atlantic Systems Guild, a New York and London-based consulting practice. He is the author of seven books on software method and management, including the recent Random House book Slack: Getting Past Burnout, Busywork, and the Myth of Total Efficiency. Past titles include The Deadline: A Novel About Project Management, as well as PEOPLEWARE: Productive Projects and Teams (with co-author T. Lister). In 1986 he was named winner of the J.D. Warnier Prize for “lifetime contribution to the information sciences.” He won the 1999 Stevens Prize for contribution to software methods. He is the author of a mainstream novel, Dark Harbor House, (Down East Books, 2000). His consulting activity is mostly in the area of project management and litigation involving software-intensive endeavors.


Convention Center — Ballroom A
Session B
Papers: Program Analysis
Chair: Peter F. Sweeney, IBM T.J. Watson Research Center

This year’s program upholds the OOPSLA tradition of high-quality contributions to program analysis. The first paper in this session uses a constraint-based approach to implement points-to analysis for Java. For a restricted Java-like language, the second paper guarantees that well-typed programs are race-free, even as it allows objects from the same class to use different locking mechanisms. The last paper describes a combination of static and run-time techniques for detecting potential data races in concurrent Java programs.

Points-to Analysis for Java using Annotated Constraints
Atanas Rountev, Rutgers University
Ana Milanova, Rutgers University
Barbara G. Ryder, Rutgers University

A Parameterized Type System for Race-Free Java Programs
Chandrasekhar Boyapati, MIT
Martin Rinard, MIT

Object Race Detection
Christoph von Praun, ETH Zürich
Thomas Gross, ETH Zürich


Convention Center — Ballroom D
Session C
Papers: Performance
Chair: Brian Foote, The Refactory, Inc.

Programmers often shun programming language features they see as exhibiting poor or unpredictable performance. This session highlights three advanced implementation techniques that promise to relieve programmers of such concerns. The first paper shows how exceptions in Java can be implemented with unexceptional overhead. The second paper examines how to minimize the cost of asking a deceptively simple question: Is this type a subtype of another? Sophisticated implementation strategies described in the third paper nearly eliminate the customary performance penalty of Java interface invocations.

A Study of Exception Handling and Its Dynamic Optimization in Java
Takeshi Ogasawara, IBM Tokyo Research Laboratory
Hideaki Komatsu, IBM Tokyo Research Laboratory
Toshio Nakatani, IBM Tokyo Research Laboratory

Efficient Subtyping Tests with PQ-Encoding
Yoav Zibin, Technion
Joseph (Yossi) Gil, Technion

Efficient Implementation of Java Interfaces: invokeinterface Considered Harmless
Bowen Alpern, IBM T.J. Watson Research Center
Anthony Cocchi, IBM T.J. Watson Research Center
Stephen Fink, IBM T.J. Watson Research Center
David Grove, IBM T.J. Watson Research Center
Derek Lieber, IBM T.J. Watson Research Center

3:30 pm - 5:00 pm
Convention Center — Ballroom B-C
Session A
Panel: How Do Requirements Relate to Objects?
Moderator: Hermann Kaindl, Siemens AG Austria

While most object-oriented methods today include something like object-oriented analysis (OOA), they do not say much about requirements. Typically, they treat requirements just as a textual input in the form of a short problem statement. In the field of requirements engineering, in contrast, object-oriented modeling is often considered as a means to create and represent a requirements model. So, the relationship of requirements with objects still seems to be an open issue.

In the light of such controversial views, this panel intends to discuss the following questions (among others):

  • Are requirements just preliminary textual statements?
  • Can requirements be modeled as objects?
  • Can an OOA model serve as a requirements model?

Panelists:
Ivar Jacobson, Rational Software Corp.
Craig Larman, Valtech
Joaquin Miller, Financial Systems Architects
Bruce Anderson, IBM Component Technology Services
Desmond D'Souza, Kinetium
Steve Mellor, Project Technology, Inc.


Convention Center — Ballroom A
Session B
Papers: Resource Management
Chair: Doug Lea, SUNY Oswego

When applications share resources, infrastructure services and virtual machines must help prevent interference. They must also manage resource consumption and optimize resource access among the competing activities. The authors of the first paper illustrate how JVMs can isolate applications to appear as if they were separate operating system processes. The second paper explains how to limit resource consumption in Java applications. The session’s closing paper describes how to maximize the efficiency of database resources by optimizing queries.

Multitasking without Compromise: A Virtual Machine Evolution
Grzegorz Czajkowski, Sun Microsystems Laboratories
Laurent Daynès, Sun Microsystems Laboratories

Portable Resource Control in Java: The J-SEAL2 Approach
Walter Binder, CoCo Software Engineering
Jarle Hulaas, University of Geneva
Alex Villazón, University of Geneva

Incremental Computation of Complex Objects Queries
Hiroaki Nakamura, IBM Tokyo Research Laboratory


Convention Center — Ballroom D
Session C
Practitioner Reports: Software Reuse
Chair: Jim Doble, Tavve Software Company

Many organizations have looked to object technology as a means to achieve significant gains in software reuse. Results, however, have run the gamut from stunning success to unmitigated disaster. The first paper in this session presents a survey of results from a variety of reuse-oriented projects within Mentor Graphics Corporation over the past 12 years. The last two papers describe experiences and lessons learned on individual projects involving the development of reusable frameworks.

Mentor Graphics Corporation’s Experiences with Object-Oriented Software Technology Reuse
Phil Brooks, Mentor Graphics Corporation
Steve Aho, Mentor Graphics Corporation

Since 1989, Mentor Graphics Corporation has undertaken a number of large-scale projects where software reuse, in various forms, was a major goal. A team of seven senior software development engineers recently examined software reuse practices within Mentor Graphics in order to figure out what works, what doesn’t, and why. Several actual projects, some that were successful and others that were not, are reviewed. The findings from these project experiences, along with the combined experiences of the members of the team, lead to conclusions on how organizational structure combined with the various reuse strategies presented aid or hinder reuse.

Construction of an Object-Oriented Framework for Text Mining
James W. Cooper, IBM T. J. Watson Research Center
Edward C. So, IBM T. J. Watson Research Center
Christian L. Cesar, IBM T. J. Watson Research Center
Robert L. Mack, IBM T. J. Watson Research Center

This paper describes the construction of a Java class library for text mining and information retrieval. The system consists of Facades around a database, a search engine, and a text mining tool. The authors discuss the design of the object models used for each of these elements, how these object models evolved as different databases and search engines became available, how they needed to evolve the system further in work with their customer, and the eventual fate of the system after the customer adopted the final version of the code, showing what was learned from the experience.

Applying Design Patterns and Other Technologies to On-Line Application Servers for Bill Payment, Bill Delivery, and Banking
Don Barton, CheckFree Corporation

This paper discusses how an organization gained high levels of reuse and developer concurrency, as well as scalability and availability by leveraging techniques from OOPSLA and other sources. The author discusses how a team and a process were built around the use of design patterns, frameworks, metaobject protocols, refactoring, code generation, and a custom build environment with integrated testing.


Jump to: Tuesday morning || afternoon Wednesday morning || afternoon Thursday morning || afternoon
Full descriptions of entire Technical Program

Back To Final Program