The Archie search engine: I was there...

On 11th (some sources say 10th) September 1990, a small team of people from McGill University in Montreal, Canada released what is generally recognised as the world's first internet search engine. It was called archie. The team members were Alan Emtage, (J) Peter Deutsch, Bill Heelan and Mike (der Mouse) Parker. As surprising as it may seem, I used to work with these clever people and regarded them as my friends. But I played no part in the development of archie. Or did I?

My story starts in Montreal in 1986 at a small hi-tech company called MUX LAB, where Peter and I were working as part of a small team of programmers developing a product called ACCESS 36. (I still have a copy of the user manual.) It was both a rewarding and a frustrating experience at times, and after a few incomprehensible management decisions, Peter left to work on his Master's degree at McGill full-time. My time at MUX LAB came to an end in early September 1987 -- I should have seen it coming...

About the same time, the post of System Manager at McGill's School of Computer Science had become vacant, and it was offered to Peter -- which he accepted. Not long afterwards, knowing of my lack of employment, Peter asked me whether I would like to work for him as a (Unix) systems administrator. As the saying goes, the rest is history! I was delighted to accept, although I knew it would be a steep learning curve. Peter was already assisted part-time by Luc, whom I also knew from my stint at MUX LAB, and who was also working on his Master's degree. At that time, the School of Computer Science was located on the eighth and ninth floors of Burnside Hall on McGill's downtown campus.

The first few months were challenging but by early 1988 Peter was looking to expand the team. If I remember correctly, Bill came recommended by a friend of mine (Stephen) at Concordia University. One of my earliest memories of working with Bill was helping to set up and configure a new Sun 3/280 server by booting it from a remote 1/2" tape drive. Based on Bill's recommendation to Peter, Wanda joined the team in late spring or early summer of 1988. And by the summer, Alan had swelled our ranks -- although I think it might have been earlier as he shared an office with Luc, separate from the full-time staff.

One probable reason for the relatively sudden increase in staff numbers was that The School of Computer Science (SOCS) was to be relocated to the McConnell Engineering Building, into newly refurbished spaces on the second and third floors. The SOCS system staff were responsible for all the computers, serial cabling and most of the network cabling. So it turned out to be a very busy summer running and terminating cables, getting everything ready, and then moving everything into new rooms and offices.

By 1989, we were settled in to Engineering and the pace of life seemed to slow down to more of a routine. Frankly, and in retrospect, we were over-staffed for the size and complexity of the School at the time. But they were ideal conditions for what was to come.

As has been related elsewhere, Alan had accumulated a number of large files listing the contents of Internet download sites. (I have a feeling that I grumbled when I discovered that some of them had appeared on the server milo that I used to maintain!) In time and quite logically, this transformed into the archie project, and then took off. Peter announced that we were welcome to volunteer to work on the project, but I remember thinking that most of the (interesting) programming was already underway and I didn't think I could contribute anything useful. (Where was the crystal ball when you needed it, eh?) Bill's desk was adjacent to mine so I was aware that he was working on the user interface, and was inspired by the VMS shell that he had used at Concordia. I remember hearing Alan talk about memory mapping as a way to speed up database access. Mike (Mouse) was based in McCRIM a couple of floors above, and was our friendly local Unix guru who would drop by regularly and offer advice.

So I made no direct contribution to archie that I can remember. However, I did once solve a problem that was causing the archie server (a Sun 4/280 running SunOS 4.1 I think?) to crash fairly frequently while under load. I believe it was Luc who came to see me with error messages from the server log files. It looked as though it was related to the serial ports. This rang a bell: the fix was to physically disconnect the DB25 plugs (still cabled-up to the patch panel) from the serial ports on the server. I then heard that this seemed to help with extending the uptime...

This all happened long before the arrival of consumer digital photography. I had taken a few photos of the work environment on colour print film, and a couple of the most relevant appear here. (Apologies for the general low quality: an inexpensive all-manual SLR and matching lens.) The top picture shows an out-of-focus (oops) Bill (L) and Peter (R), and was taken around the spring of 1988 in the old Burnside Hall office. The photo of Wanda (L) and Bill (R), taken from my desk, dates from the summer of 1991 and shows the office where a good proportion of the archie discussion and development took place. Alan would sometimes use the desk in the centre of the photo as his office was next door. (I thought I had a photo of Alan, but nothing has come to light yet. I'm fairly sure that I didn't take any pictures of Mouse.)

My former colleagues, wherever they are now, may remember things a little differently -- some of the events mentioned happened more than 30 years ago, and memory can be unreliable. I felt it would be useful to record a little more detail about the background and context of the creation of archie as I still recall it. (Some of the references I have found about it are misleading or inaccurate!) I am of course proud of the achievement of my former sys. admin. team members: at the time, none of us suspected how it would change the world!

Programming references?

I must have been fortunate in that my higher education took place about the same time as the personal computer was starting to make an impact on everyday life and the world was starting to turn digital. It was a good time to be in computing -- despite what my parents thought, especially my father (bless him). Over the years I have been able to make a modest but comfortable living from IT, and so it continues.

When I started out I wanted to be a programmer, and my wish was fulfilled immediately after I had finished university. Even though I then moved into (computer) Systems Administration, I have always enjoyed programming and try to keep my hand in, even if it is only simple shell scripts which are then subsequently discarded.

Pre-university, probably like thousands of other people, I taught myself BASIC, no doubt with some common bad habits. In university, I was taught -- or was expected to learn at various times -- FORTRAN, Pascal, PL/1, SNOBOL, PDP Assembly language, and C. I subsequently picked up Intel 8086/8088 assembly language , Bourne shell scripting, and Perl. (Some of the programming reference books I used can be seen in the photo.) I briefly dabbled with C++ and PHP, but Java and JavaScript completely passed me by. Either shell scripting or Perl are my weapons of choice for repeatable systems administration tasks and projects.

In recent years, I have heard much about the scripting language Python. It is the favoured language for a number of my users, and the Debian Linux packaging system is written in it. I have received a number of requests to install new Python packages, or update existing ones. (This does not require much Python knowledge, fortunately.) Perhaps it was time that I learnt more about Python?

With this in mind, I sought a suitable reference book -- as I have found a portable, printed, read-nearly-anywhere guide had been essential in the past. I have long been a fan of the O'Reilly series of books, so this seemed like the obvious place to look. My copy of Programming Perl had served me well over the years; I wanted an equivalent written for Python. So, rather than check the wisdom of the internet and look more closely at what I was asking for, I put O'Reilly's Programming Python on my Christmas gift list...

My sister was kind enough to get it for me. What arrived was this monster volume of 1600+ pages that can be used to prop doors open. It is not such a pleasant experience to carry it any distance, and its size makes it difficult to easily keep the book open on certain pages. Had I done just a little more research, I would have noticed that it is big by the page count: 1632. And also that it should have been named more appropriately as Using Python. (Perhaps I should have been looking at Learning Python instead, all 1648 pages of it?...)

To use a bit of (Monty) Python: Stop, this is silly! The general consensus is that it is easier to learn Python than it is C, so why does it take 1600 pages to cover what the seminal The C Programming Language seems to do in 200? (I know this is being provocative, as the C book assumes a single version of the language, a Unix OS, a command line and a limited number of libraries. Things have moved on since the 1990s.)