Dennis Ritchie will "desconstruct" the original CACM paper -what have we learned, what we got right, what did we overlook. Computing has changed in the past 25 years and the ideas of distribution and networking were less central then than they are today. Clearly, modern systems have been impacted by UNIX and the ideas presented in the original paper. But how has UNIX been impacted by our changes in thinking?

 
20th Anniversary of the First Port of UNIX
Steve Johnson, Transmeta; Richard Miller, Miller Research; and Juris Reinfelds, New Mexico State University

Nowadays, the portability of UNIX is taken for granted. The first ports of UNIX were audacious projects. Two teams independently succeeded with ports of UNIX at about the same time, only to find out about each other when the ports were finished. Both teams used different techniques for porting, and these talks will present the strategies used and how they hold up to current porting practice.

The UNIX code was so well designed that it could be picked up and ported without any consultations with the authors of the code. Reinfelds and his research team ported UNIX to Interdata 7/32 at the University of Wollongong in Australia, where Richard Miller proposed an innovative implementation of the port and proved its effectiveness by single handedly porting the kernel code and most applications. The Wollongong port later became the first computer vendor-supported UNIX.

Johnson and Ritchie were doing the port at Bell Labs on Interdata 8/32. Porting Unix required that C "get serious" about portability. Portability concerns led directly to such innovations as separate name spaces for structure members, sizeof, and indirectly to tools such as lint and social conventions such as the use of header files. This talk will also dicuss places where we were less successful, notably alignment, byte order, and bit fields.

Thursday, June 18   11:00am - 12:30pm    see [FREENIX Track]

SimICS/Sun4m: A Virtual Workstation
Peter S. Magnusson, Fredrik Larsson, Andreas Moestedt, Bengt Werner, Swedish Institute of Computer Science; Jim Nilsson, Per Stenström, Fredrik Lundholm, Magnus Karlsson, Fredrik Dahlgren, Dept. of Computer Engineering, Chalmers Univ. of Technology; Håkan Grahn, Dept. of Computer Science, Univ. of Karlskrona/Ronnebyl

High-Performance Caching With The Lava Hit-Server
Jochen Liedtke, Vsevolod Panteleenko, Trent Jaeger, and Nayeem Islam, IBM

Cheating the I/O Bottleneck: Network Storage with Trapeze/Myrinet
Darrell C. Anderson, Jeffrey S. Chase, Syam Gadde, Andrew J.Gallatin, and Kenneth G. Yocum, Duke University ; Michael J. Feeley, University of British Columbia

Software projects have developed using different models to control the outcome. As software began to be developed within a world-wide culture, particularly when the source code was freely available, two distinct models of development have become popular:

* The Cathedral model, where one or two "master builders" define the basic structure of what will be built. BSD UNIX and the GNU project are typical.

* The Bazaar model is free running; there is no central control or basic building plan. Different programmers, like vendors at a bazaar, offer up different approaches to different problems. This approach has been the core of Linux development.

The speakers will each present their views and the history of what has worked and why. Neither Linux nor BSD was purely either approach, but the ideas presented by these two styles should make for a fascinating and lively discussion.

Thursday, June 18   12:30pm - 2:00pm   Lunch (on your own)

Thursday, June 18   2:00pm - 3:30pm    see [FREENIX Track]

Mhz: Anatomy of a Micro-benchmark
Carl Staelin, Hewlett-Packard Laboratories and Larry McVoy, McVoy, Inc.

Automatic Program Transformation with JOIE
Geoff A. Cohen and Jeffrey S. Chase, Duke University; David L. Kaminsky, IBM

Deducing Similarities in Java Sources from Bytecodes
Brenda S. Baker, Bell Laboratories and Udi Manber, University of Arizona

Many programmers are far too ready to roll up their sleeves and start writing C (C++, Java, Fortran, etc.) when they should be considering alternatives first. Interpretive languages often provide a better way to do things, even for fairly ambitious projects. Sometimes a certain amount of C (or whatever) is indeed indicated, but even then, better results can often be had with a partnership between primitives written in C and overall control written in something else. This talk will discuss why the instant resort to C is a bad idea; what some of the alternatives, including mixed solutions, are; and how to make the choice.

Thursday, June 18   3:30pm - 4:00pm   Break

Thursday, June 18   4:00pm - 5:30pm    see [FREENIX Track]

Short, pithy, and fun. Works-in-Progress Reports (WIPs) introduce interesting new or ongoing work. If you have work you would like to share or a cool idea that is not quite ready to be published, a WIP is for you! We are particularly interested in presenting student work.

The Works-in-Progress session will consist of five minute presentations. If you have work-in-progress, we invite you to submit a one or two page abstract via email in plain text (please, no MIME or Post Script, etc.) to: wips98@usenix.org by May 1. Please include your name, affiliation, and the title of your talk.

A schedule of presentations will be posted at the conference. Speakers will be notified in advance. The five-minute time limit will be strictly enforced.

Philips n.v. is one of the world's largest companies. Its internal network is one of the largest in the world, containing approximately 150,000 hosts spread over 70 countries. Corporate culture and policy makes the management and administration of DNS a particularly difficult problem. Each site has the budget to choose their own way of handling DNS, but the skills and resources available vary between the extensive (a few) and the non-existent (the majority). This talk will discuss the many problems that result from this diversity and the difficulties of deploying proprietary tools to alleviate or fix them. It will also describe the solution that Origin (the IT services company of Philips) developed to overcome these problems - ADAPT.

ADAPT also entailed the roll-out of a new DNS backbone architecture. Operational problems caused by the introduction of this new backbone - essentially renumbering the network's internal root DNS servers - will also be described.