Lecture: Rediscovering and Reviving Archie

Introduction

• Purpose: Bring back Archie, the first search engine
• Problem: Archie has vanished, FTP (File Transfer Protocol) is obsolete
• Goal: Run an Archie server ourselves to preserve Internet history

Background

• Archie:
  • Created in the early days of the Internet
  • Indexed FTP sites, allowed users to search public files
• Current State:
  • No Archie servers online
  • FTP protocol is no longer supported by major web browsers

Initial Approach

• Finding the Archie Software:
  • Contacted Alan Mage (Archie creator)
  • Searched Computer History Museum (no relevant items found)
  • Explored online records and old server listings (e.g., 1999 Bunny up site)
• Challenges:
  • Old servers shut down, telnet interfaces inactive
  • Potentially lost source code and data

Leads and Exploration

• Anarchy (Macintosh Client):
  • Graphical interface for Archie
  • Contacted creator Peter Lewis (didn’t have Archie source)
• Archiplex (WWW Gateway):
  • Web gateway for Archie servers
  • Contacted creator mar coaster (didn’t have Archie source)
• Internet Old-Timers Group:
  • Searched for leads among early internet users
  • No copies of Archie found

Breakthrough: Universities

• Rutgers University:
  • Contacted former administrator Dan Ricean
  • Largest Archie server in 1990s, but backups lost
• University of Warsaw:
  • Had an old database, found backups of source code and binaries
  • Contains the last known copies of Archie

Resurrecting Archie

• Files Received:
  • Four tarballs, README, and Postscript file
  • Contained source code and server backups
• Initial Steps:
  • Created Kimu virtual machine, installed Sun OS 5.6
  • Followed installation documentation to set up environment
• Key Configurations:
  • Created required users and groups
  • Set up databases and configuration files (AR domains, AR retrieve, etc.)

Encountered Issues and Solutions

• Errors:
  • “re undefined variable” due to Sun OS version mismatch
  • Missing make utility, File not found errors (missing Perl)
• Solutions:
  • Hard-coded variables, installed the missing utilities
  • Re-ran indexing process

Success: Operational Archie Server

• Indexing FTP Sites:
  • Configured via hostmanage program, ran indexing commands
  • Successfully retrieved and parsed data
• Telnet Client:
  • Operated by starting the Prospero server
  • Tested and validated successful searches

Public Launch

• New Public Archie Server: arch. sereral port. org
• Interface: Based on Arlex, allows FTP file downloads via browser
• Contributions: Community effort, thanks to contributors like Mike Schwarz and Alan Mage

Conclusion

• Achievement: Successfully brought back Archie
• Utility: Useful for modern-day searches and file downloads
• Future Vision: Preservation and usability for historical and practical purposes