During the time between GeoVision laying me off and the end of the DMR gig, I must have went on over a hundred job interviews, and I was getting pretty good at it.
Actually, what I think was really happening is I was getting more and more convinced that interviewing was a waste of time, and I stopped caring. Anyway, that made me more relaxed in interviews than I had been, and that was paying dividends. I had a job offer from Bell Northern Research (aka Big Nerd Ranch) but it was for way less than I’d asked for and the job didn’t seem all that exciting to me. And I had an interview coming up with Gandalf, for a job that sounded VERY exciting.
Gandalf was a company that made networking equipment from the early days of networking. When I was a student at the University of Waterloo, every room with computer terminals had these ubiquitous blue boxes that had had a little dial on the front – we all knew that you dialed “3” (I think) to get onto the Engineering CMS network. Some of the more adventurous of us found that if you switched it to some other numbers you got the library catalog system, and even the Math Department’s Honeywell computer (known as Honeybun, or “the bun”). The really adventurous got an account on the bun so we could play Colossal Caves Adventure, the biggest time sink you’ve ever encountered in your life. The other thing that Gandalf was famous for during my university days was cool posters showing a wizard, and a company logo that looked sort-of rune-ish.
Somewhere in the intervening years Christopher Tolkien had continued sucking the dry husk of his father’s corpse and corpus and had sued Gandalf so they had a new logo and didn’t refer to wizards at all.
By this time, however, if you asked anybody about Gandalf, they would say “Oh, don’t they make the 2400 bps modems that cost 8 times as much as anybody else’s, even Hayes?”, and they’d be right. Gandalf’s stock wasn’t doing well. But their main business wasn’t these crappy over-priced modems, it was great big crappy overpriced routers.
The job I was being interviewed was to continue the work somebody else had started, with a new team. My title was going to be “Senior Automation Designer”, and I was going to have two people working for me. The project was
taking a lab full of Linux computers, each with one regular “Network Interface Card” (NIC) and one special one, and making an automatic test program that would use these special cards for testing Gandalf’s router equipment. The special cards could emit and detect all manner of custom and invalid packets, so they could emit them on one side of the router and make sure they didn’t cause the router to halt and catch fire, and make sure the bad packets didn’t get through to the other side but the good ones did, etc. Before this automated tester came along, the QA people had sat at Windows boxes manually constructing these packets and emitting them and then examining the router through its serial interface, and so on ad nauseum.
The job sounded challenging as all get-out, and while I was a little scared about the prospect of being an official manager instead of an ad-hoc team leader like I’d been much of the time at GeoVision. What team leadership at GeoVision had taught me was that if my team members were good and self-motivated, I can do a good job of giving them advice and feeding them work as needed, but when they are incompetent or liars, I couldn’t spot the
problems to save my life.
However, the aura of “I don’t give a shit” seemed to be shining through during the interview. After the interview, the head hunter who found me the position called me to say they were incredibly impressed by my “aura of quiet
confidence”. I didn’t do anything to dissuade that impression. He also suggested that I write them a letter thanking them for the interview and stressing what I liked about the job. I’d never done that before (or since), but I did it and it seemed to work – not only did I get the offer, I got just a hair under what I’d been asking.
I was especially excited about this job because I’d been wanting to try Linux for a while – I’d been on a friendly (non-tech) mailing list with Matt Welsh, one of the guys at the forefront of evangelizing Linux, and he’d made it sound like everything I wanted for a home operating environment (mostly because for a home operating environment I wanted to duplicate what I’d had at work, which was usually olvwm on SunOS). With this job in hand and the money from DMR, I bought a used but fairly up-to-date computer, a 486-33 with a whopping 16Mb of
RAM. The seller was a student who realized after buying it that he didn’t need to play the latest games so much as he needed to make his next tuition payment.
The atmosphere at Gandalf was pretty great – I had a large cube with a floor to ceiling window making up one very long wall. I found out later that the cube was empty because most people didn’t like how much cold air came off the
window in the winter, but there was a hedge in front of the window that was frequented by a number of birds, so I didn’t mind the cold. My team turned out to be one guy who was devoted to the team full time, and another who was supposed to be on the team half time and a QA tester the other half. I thought this would be a good arrangment, because the QA testers were our “customers”, so having somebody who knew that end of it would be a major advantage. Unfortunately, it didn’t work out that way – I never got more than a day or two a month of his time. And like the other QA testers, he turned out to be good at telling me what they were doing at the moment, but really bad at thinking of what would be a better way of doing their jobs.
The Linux computers in the lab consisted of two or three that were reasonably up to date (486s with S3 VLB video cards) which were used as the controlling computers, and the rest were salvaged and scrounged from around the company. Most were 386s, and two of them got a hell of a lot faster when I discovered that the “Turbo” button hadn’t been pushed, so they were running at 8MHz instead of 16MHz. They were running one of the first “commercial” Linux distributions, SLS 1.03, and had been upgraded to kernel version 0.99.14. Installation or upgrading involved marching approximately nearly 100 floppy disks around the lab. Some time later, I switched to Slackware, which came on a CD-ROM (although we didn’t have CD-ROM drives in most of the lab computers). The CD-ROM was actually 100 or so directories, each one of which had the contents of one floppy from the floppy distribution. The other nifty feature of Slackware was that it had an “Install from NFS” option, so I copied the
Slackware CD onto the workstation on my desk and exported it, and then I didn’t have to sit by each computer swapping floppies in and out of them. After doing that once, I discovered that each “floppy” directory had a file
that specified what files where mandatory and which were optional, so I modified the file so that everything I wanted on the machines was mandatory and everything else was optional, and installations got a lot more hands-off. Even later, I switched to an early version of RedHat Linux, which had the same networked install, customized installations using “Kickstart”, and an upgrade-in-place option. Suddenly I didn’t have to reinstall and reconfigure every computer in the lab every time I needed the latest of something. And Linux was moving so fast back then that I was always discovering bugs that the latest version was supposed to fix – and sometimes it did. During this same period, I went through pretty much the same evolution with my home computer, except between Slackware and RedHat I had a brief flirtation with a weird distribution called Yggdrasil, named after the Norse Tree of Life. I can’t remember why I decided it would be unsuitable for the computers at work, but I do remember that it had a kernel patch that would emulate a sound card through the built-in speaker, which was cool but useless.
My boss, Pierre, was a great guy. In the summer time, he had a couple of barbeques at his cottage in the Gatineau hills, and invited everybody who worked for him and their families. Nice place, and he had a boat for taking
people waterskiing. His boss was also a really nice guy, and used to take the group out to lunch every now and then, but after I’d been there a few months he left to take a job with Eriksson. Unfortunately his replacement was an
incredible jerk. It’s probably just as well I don’t remember his name, because I understand that truthfulness is no defence against libel laws. I’ll just call him “Jerkface”.
Jerkface decided on day one that my project was a big waste of money. As a matter of fact, based on the fact that he’d ship product when we’d identified hundreds of major bugs in it, I don’t think he believed in QA at all. Soon
after he arrived, Pierre told me I had to write up a cost justification for my project for Jerkface’s benefit. I did, citing the last product test cycle where we had done thousands of tests on every hardware/firmware revision in
the time when they used to do a few dozen, and how this was contributing to releasing a higher quality product. I included detailed explanations of what the product did, what I planned to add to the product in the future. After
hearing a nothing from Jerkface, Pierre suggested that I start putting out a monthly news letter for Jerkface, the hardware designers and the QA people detailing what we had done, what we planned to do, the methods we used, and also what the QA people had done with the system. But still Jerkface kept asking Pierre to explain what we were doing, why it was costing so much, and what benefit this was having for the company. And each time he did, I’d have to find yet another way of wording the same stuff that I’d put in the first document, and try NOT to sound sarcastic or annoyed that I’d already done this every other week since he’d arrived at our company.
I was also involved in another little battle with Jerkface at the same time. My desktop computer was a SPARCstation. But since I was developing code that had to run on Linux computers, I wanted a Linux computer for my desk. We didn’t have the budget to buy me a new computer, though. So I hit on an idea – I asked on the company wide mailing list, and there was a guy in Holland who needed a SPARCstation, had the budget to buy one, and was willing to buy me a top of the line Linux box for mine. So I spec’ed out exactly what I needed, and sent it off to him. He was very enthusiastic, especially since we’d be buying the equipment at Canadian prices rather than Dutch prices, which gave him a major discount. But Jerkface scuttled the deal. Even though it wouldn’t have cost him a penny, and we’d get a faster and more useful machine in return, he didn’t want to lose that SPARCstation from his empire, because every penny of inventory reduced was a reduction in his value to himself.
Another problem with Jerkface was that he had no respect for the “chain of command”, at least going downward. He’d come in to the QA lab or the automation lab and yell at people for any slight, real or imagined. We
weren’t working hard enough, we were leaving too early (never mind the fact that most of the QA guys got in before 8 and he usually strolled in at 10:30), we didn’t care, we were reporting too many bugs and slowing down the release too much (yes, really). He was so out of touch with what was going on, so even when there was something going wrong (a product taking too long to test, whatever), he’d yell at the wrong person. But it didn’t flow both ways – you couldn’t come to him with a problem, you had to go through Pierre.
At this same time, my friend from GeoVision, Dan Simard, also came to work at Gandalf. We were both nuts about mountain biking, but he was a lot more nuts about it than I was. I remember one spring where he spent $1500 on titanium screws and bolts and a seat post to save a pound and a half off of his already superlight titanium framed bike. That was about the total value of my bike. Anywhere, there was a storage area where I worked, and hardly anybody ever went in there. I took to keeping my bike there when I cycled to work, and invited Dan to keep his in there too. Gandalf actually had shower facilities, so riding to work would seem to be something to be encouraged, but not according to the head of Human Resources. She caught Dan carrying his bike into the building once, and told him he couldn’t do that because it would muddy up the carpets. When he pointed out that he was carrying it over his shoulder and it wouldn’t touch the carpets, she told him to take it outside and turned around and left. Then she discovered the storage area where I was keeping my bike and told my boss that the fire marshall had declared my bike was a hazard, because it would get in the way of people exiting the building. This in spite of the fact that the storage area wouldn’t normally have anybody in it, it only had one exit and the bike was stored along the opposite wall from the exit so the only way it could be in your way was if you decided to tunnel out. So I started keeping my bike outside. Dan refused to do that, since anybody could have casually walked off with a tiny bit worth a week’s salary. Can’t say I blame him, although my bike was never damaged.
Dan later left Gandalf to go work at a much, much more successful networking company down the road called Newbridge. We referred to them as “Gandalf West” because so many of our people left to go there, and they referred to us as “Newbridge Junior”.
One other little challenge I had to deal with, and I don’t think I dealt with very well, was when my team member, Todd, came up to two of the (male) QA testers in the lab and showed them a picture album. The pictures were of Todd in drag, doing a strip tease, and then jerking off. “See anything there you like?” Oh man, how do you deal with something like that? Well, probably not how I dealt with it. I tried to be supportive of Todd, but none of the QA testers wanted to even be in the same room with him. I gave him time off, I suggested he see a therapist, I tried to have him working on stuff where he didn’t have to interact with the testers. It was a very difficult time.
Anyway, Jerkface was such a horrible manager that nobody who worked for him liked him. I used to find appropriate Dilbert cartoons and stick them to his door in the mornings before he came in. Everybody else got a good laugh out of them before he strolled in late and removed them. When I left, I wrote an utterly scathing describing everything Jerkface had done to destroy morale in the QA department and praising Pierre for all his good work in attempting to insulate us, and gave it to HR in my exit interview. I found out some months later that they’d fired Jerkface, and I felt good, but then I found out that they’d fired Pierre as well. I felt pretty damn bad about that. I hope to hell it was just cost cutting, but maybe they just decided that when a manager and the person he manages can’t get along, they have to fire both of them, rather than just firing the jerk.