💾 Archived View for stack.tilde.cafe › gemlog › 2022-09-05.comdex90.1.gmi captured on 2024-07-08 at 23:36:56. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2023-09-08)
-=-=-=-=-=-=-
James is noticeably upset. His hands are shaking, his voice is cracking, and his grey toupee is slightly askew, but I can't really tell him that. He takes another sip of his Donald Duck juicebox and looks around furiously, sensing that I am not a good audience. I can't follow what he's saying: he looks even more ancient than usual, and all I can think about is what would happen if he dropped dead right here and now...
Behind us there are 11 computers with dark screens, the newest of the IBM PS/2 line with Micro Channel architecture, sitting on tables on an otherwise empty stage. A couple of actors are running lines off the side, and my partner is sleeping in the first row of chairs.
James takes a deep breath and his rant finally enters my ears: Las Vegas electricians' union, blah blah, not allowed to plug the computers into the extension cords two feet away. He gulps down the juice, and with a dramatic arm wave, shuffles back to the bar at the end of the convention floor, where Donald Duck juiceboxes are refilled with vodka.
I look at the dark CRT monitors. Most of the machines are '286 boxes, primitive even for 1990 -- the '386 has been out for five years! Apple, Atari and Amiga are using 32-bit 68030 processors. Sigh.
I close my eyes for a second, then force them open, as I really can't afford to fall asleep right now. We got in on a late flight, checked into the damp, dungeon-themed Excalibur hotel, and got a few hours of sleep on short-sheeted beds. I just need a bit of quiet time; in a few minutes (or hours) the computers will be plugged in, and it will be up to me to make sure they all boot and sync up.
The rest of the production team gathers in the back row. James, now lit, is gesticulating wildly, as he tells stories of old-time theater projects he had directed, and the fabulous people he knows. I take another sip of my Duck juice and sit down dangling my feet off the stage.
"... And the IBM Token Ring Network connects our computers to make this possible!", the actor extends her hand to the row of machines flashing pictures on their screens. Final images bounce across the 11 screens on stage, turning into striped IBM logos (conforming in color and proportions to the IBM Logo book). We made it through another show without a crash, three in a row. Whew. I take my between-the-shows position on stage, guarding the computers.
A big German guy steps up on stage. "I look at computers?", he says. I try to explain that the area is off limits. He is not giving up: "I push button?", reaching his extra-long arm towards a nearby keyboard. No, no, no... "I push button, crash computer?" How does he know? Thankfully, a security guard steps up and escorts him off-stage.
The computers are networked using the IBM Token Ring network. "Networked" is stretching it, since it's used exclusively to start the show simultaneously on all machines. It's an overkill -- I voted for twiddling a parallel port pin as all PS/2s come with a glorious bi-directional parallel port with many IO pins.
But IBM wanted to say that the show was made with a Token Ring network, and so it was. The network drivers were faulty on a couple of new machines, and after a few weeks of crashes IBM sent Brad and Jackie, always-impeccably-dressed network specialists, to help us out. In the end we set up a low-level mailbox interface that was relatively crash-free, to kick off the show.
Doing anything more than that, such as synchronizing actual cues, was too much, and one of the computers invariably locked up. So the network would start the show, and each computer was free-running to the end. Most of the time it worked fine, although sometimes, hitting a hard drive would take an extra second as it did some kind of recalibration, and we would wind up slapping our heads backstage.
We finally coded a keyboard-controlled way to get back on track. In case of trouble, one of us would pop up on-stage, behind the sweating actor doing his/her schpiel, and hit the spacebar at the right time to sync the computer back up.
But between shows, touching the keyboard would cause a scary few minutes of stepping through the show or cancelling the loop and hoping that it restarted correctly. The show is strung together with a DOS batch file. It's an imperfect system, but we had very little time to prepare. And it's 1990.
The year is 1990. The Wall came down in Germany. Al Gore is busy inventing the Internet. Larry Page is a high school junior in Michigan. Finding coding gigs in NYC is hard. New York is still an analog town of bankers and ad execs (doing print, of course) recovering from the Big Crash of '87.
I stumbled into this job, a strange last-minute affair. A floundering theatrical production company somehow got the IBM Comdex 90 booth. No one knows shit about computers, except one kid who shows up when he isn't too drunk. The show starts in two months.
I am in no position to pick and choose jobs, so I take it, fully expecting all-nighters. I manage convince them to pay for my partner from my recently-failed company, since I figure we'd have to code our way out of this mess.
The Intel rep in the back row catches my eye, makes a gun with his fingers and points it at me with a wink. He toasts me with the Duck juice in his other hand; I raise my Duck in return. The best part of the show is coming up.
"... And with IBM's new eXtended Graphics Array, XGA, you will never run out of colors!", the actor's voice booms in the darkened stage area. The displays explode into a rainbow of colors, in a slow, 90's way. The audience is impressed at the coolness of XGA graphics.
I take a sip my juice, endlessly amused. The truth is that the XGA hardware -- or maybe the drivers -- crashed in XGA graphics modes. We faked XGA.
XGA -- theoretically -- can display 32K colors at 640x480, amazing for the time (and unsupported by any software). There is no easy way of faking it. Intel stepped in and sent us a bunch of DVI boards last minute. DVI ia a pretty obscure $5000 board (if you can get it, which requires knowing someone at Intel), with an on-board VLIW microcodable chip optimized for graphics. It too has a 16-bit color mode. DVI boards can even play full-screen (but really blocky) video, but doing that would ruin the illusion of XGA.
The Intel guy slaps me on the shoulder and congratulates me. He's never seen a DVI board behave with such stability before, and this is several DVI boards! I manage to mumble a thank-you. I wrote a few wipes and effects in clumsy C in a marathon 3-night coding session. I did not know it yet, but I would spend the next 5 years writing DVI code (and microcode too).
We are running MS-DOS 4.0. Version 5.0 is out, but IBM machines here boot 4.0, and as I found out the hard way, the IBM version of DOS 4.0 is not the same as the Microsoft version of DOS 4.0... It is full of bugs and crashes.
I did a year-long stint coding on Atari 520 ST, a 32-bit 68000 machine with 512K of linear RAM, three years ago. Working on a DOS box with its segmented 16-bit CPU seems like a slap in the face, and I hate every second of it. Apple and Amiga had 32-bit machines for years, I am stuck with dressed up PIC microcontrollers.
At home I wrote a credible Forth-like development system using the PharLap DOS extender, in flat, 32-bit protected mode on my 16MHz 386SX box. Needless to say, it is not possible on 286 boxes here, in IBM-land.
Windows 95 is still 5 years away...
Bob Denver is hawking some product in a booth nearby. The set is a an island, with a palm tree, a pile of coconuts and a computer. He is babbling something about a word-processing program even Gilligan can use.
I am almost done with my Duck juice, and it's still one more show before we shut down for good.
I briefly wonder what the future may bring. I am still a kid, albeit with 10 years of coding experience. Can I survive in NYC without joining a NJ COBOL consultancy my distant cousin keeps trying to talk me into? "What are you doing, playing with these toys," he says. "We make real money on mainframes!" I really don't want to do that, but work is hard to find and I still owe a chunk of money from my last fizzled business venture...
I am tired. Someday all this will be a distant memory; in a few decades I will write down the few things I can still remember. But that is far in the future, and who can foretell the future anyway? Right now, Comdex '90 is coming to a close.
Not my photos, but will give you the taste:
https://www.flickr.com/photos/linuxjournal/2177873471/in/photostream/
https://www.flickr.com/photos/lupichan/albums/72157623563277740/
(c) 2022 Stacksmith. All rights reserved.