Algorithms + Data Structures = Programs - Episode 58: Otto & Eric Niebler! (Part 1)

Episode Date: December 31, 2021

In this episode, Bryce and Conor interview father and son Otto and Eric Niebler!Date Recorded: 2021-12-02Date Released: 2021-12-31ArrayCast PodcastThe Tweet that led to this episodeC++20 RangesUNIVAC Solid StateUNIVACKen IversonAPL (A Programming Language)GE 200 SeriesGeneral ElectricNYSE (New York Stock Exchange)Fortran Programming LanguageCOBOL Programming LanguageALGOL Programming LanguageAssembly LanguageIBM (International Business Machines Corporation)Working with Asynchrony Generically: A Tour of C++ Executors (part 1/2) - Eric Niebler - CppCon 21IBM Model M keyboardZork IIBM Personal Computer XTIBM Personal Computer/ATIntro Song InfoMiss You by Sarah Jansen https://soundcloud.com/sarahjansenmusicCreative Commons — Attribution 3.0 Unported — CC BY 3.0Free Download / Stream: http://bit.ly/l-miss-youMusic promoted by Audio Library https://youtu.be/iYYxnasvfx8

But I was with Univac between 61 and 64. That's when I discovered the APL line. Wow. Wow. So you're actually probably one of the very, very earliest users of APL, like literally in the world. Welcome to ADSP, the podcast, episode 58, recorded on December 2nd, 2021. My name is Connor, and today with my co-host Bryce, we have a special episode with two special guests, father and son, Otto and Eric Niebler.

All right, so how should we do this? Before we hop into and we let Eric and Otto introduce themselves, maybe we should tell how do we get to this point? I think at one point, Eric, like so for those that aren't aware and how this would be possible, I'm not sure. I guess maybe this could be the first episode if this episode goes a little bit more viral for whatever viral means for ADSP. I'm a huge array language fan. APL is my favorite language. And most folks are aware of this. And so I think I tweeted something and then Eric, you either retweeted or commented and said, my dad used to code in APL. And then I said, oh, maybe we should have him

as a guest on ArrayCast. One thing led to another thing. Bryce got upset that I was trying to recruit people to my other podcast. And then we ended up sitting on me. He's cheating on me with his other co-host. And, and so then we ended up setting up this episode where we have both Eric and Otto. So our first ever, and probably one, maybe one of the only, I'm not sure if there's ever been a father-son combo guest on a tech podcast out there.

I'm like, it's possible it's happened before, but this could be the first time ever. This could be an exclusive. I actually know of another pair, which is Michael Spencer, who works on LLVM and is a C++ committee member, his dad was on the POSIX committee, one of the POSIX committees back when they were part of ISO. And so they actually like they can have conversations about like committee stuff. Have they been on a podcast, though, is the question. No, no, no, not not not yet. We could change that. So, yeah, that's how we got here.

And so I'm not sure how we want to go from here. Eric, you want to introduce yourself first and then transition to Otto, or do we do it in chronological order of who entered the tech industry first? What do you think? Let me go first. I'll go first. So, you know, I'm Eric Niebler, currently employed at NVIDIA and working mostly on C++ stuff, of course. And, you know, I have, you know, stood ranges in C++ 20 is a proposal that I brought forward, Currently working on making abstractions for asynchronous programming in C++ suck a little bit less. That seems to be going all right. The sender-receiver

stuff is going swimmingly. You know, so that's what I'm excited about right now. I've been a member of the C++ Standardization committee for about 20 years. You know, I've been around in the software industry. I've worked at Microsoft and Facebook. I've consulted independently. And, yeah, I mean, I owe an awful lot of it to my dad, who bought an IBM PC in 1986 when they were extremely underpowered and very expensive. And he thought it would be really great for me and my sister.

And he was right. Both me and my sister are in the in the software industry. And yeah, I mean, I mean, I hated computers because it's what he wanted me to do. And I and I refused. I refused to learn anything about computers. I remember in high school I wanted to write my my book reports out by hand, you know, because I didn't like using the computer. My dad wanted me to use this typing, this awful, awful typing tutor program to learn how to type faster. And I hated it. And it really wasn't until college when I kind of discovered computer programming on my own. And so, you know, in some sense, you know, I credit my dad. And in another sense, I got involved in computers, you know, in spite of him. I actually feel great.

But, you know, it was it was there and it was like part of the the water that I was swimming in as a kid. You know, and I I do remember playing text based adventure games on my dad's IBM PC in the 80s. And I and I do remember thinking like, wow, this would be cool. And I remember trying to write one myself in basic when I was very young. So like, you know, the seeds were planted from way, way back. So when I got exposed again to programming in college, I think, you know, I took to it like a fish in water. So what about you, Otto? So were you a programmer, I assume? I was an assortment of things. Right now, I'm currently employed by retirement. I've been retired since 2001. I started out actually, it was the late 1950s. I was working on, quote,

computers at that time, which were no less than punch card equipment. I started really when I was in my second year of Fordham at university. I was looking for a part-time job, and one of my classmates was working in the Wall Street area. He said, come on now. I'll show you what a computer looks like. Well, the computer was a Univac Solid State 80, I think it was. It was a punch card equipment. He was doing stock record, which was a basic brokerage back office account.

I was training to be a dentist at the time at Fordham. I saw the computer and I think it was love at first sight. I went home and I said, what do you want to do for the rest of your life?

Walk down the throats of people or you want to do some programming? And I said, I think programming would be my life. So at the time I was living at home with my parents and I said to my father, I'm going to quit Fordham. I want to go into this field called computers. This is 1956 now. My father is old school. And he says, computers, he says, they're toys.

He says, they're not going to last very long. Get a real job. Become a dentist, you know. I says, no, Dad. I think there's a future there somewhere for me, you know. So I quit Fordham, and I got myself a job at a Revital Rand service bureau. And basically, it lasted for about, oh, three years, I think. Then I realized I have to get a college education.

So I went back to my family. I said, I think I want to go back to school my father says this time it's your nickel you're going to pay for it I paid for the first half, you pay for the second half so I went back to Columbia as a math major because they didn't have computer science in those late 50s

and it was there I read up on ken iverson and uh the notation that he had and i really liked his work uh so when i graduated i was hired by univac and all of a sudden i found this language called apl and it was Ken Iverson all over again. So I fell in love with the language. It was an amazing language, as you probably can tell, talk about a lot better than I can. But that's what got me really started into computers, too. From there, I had an assortment job.

I started out as a programmer, supporting a marketing team. I was in Manhattanortment job. I started out as a programmer supporting a marketing team. I was in Manhattan at the time. So we marketed mostly lower Manhattan applications. And that's what Univac. And then I left Univac, went to GE. GE was just starting up their computer line. And I programmed the, let's see, it was the 200, the GE 200, I believe it was. I was at GE for about three years.

And then I went on to New York Stock Exchange and helped automate their ticker. This is back in the late 60s. The late 60s in Wall Street was a very traumatic period of time where Wall Street finally got the word about computers and started becoming a little more efficient. And they converted most of their operation to technology. And I was there for about three years. We automated the tickericker among other projects. The project I was working on, the Ticker renovation,

we teamed up with IBM. IBM did the control program and developed a totally fault-tolerant computer. It was a three 360 50s lashed together and they shared a mass storage device. It was a ferrite core mass storage, but it was a totally full tolerance system. No time lost, no transactions lost. And it was a classic at the time. It really was.

And then from there. What was that written in? The software part? The the I believe it was written in assembly language. IBM did the programming on it, the control program programming on it. I wrote it was COBOL, mostly COBOL. We had a little bit of Fortran, something called Algol at that time. And that's a variety of languages that we used.

And mostly a sublingual language. That's where most of the tricky programming was done. Yeah, working with the New York Stock Exchange, I left the exchange and I didn't since that was the primary interface with the New York Stock Exchange to the IBM. They gave me a job. They said, hey, we want you to work on brokerage for Wall Street applications. So we developed applications for about a year or two.

And then I went on to becoming a manager a while. And then I was into consulting for a while. We did disaster recovery planning for a lot of the major firms in Manhattan, which was probably the most interesting job I had with IBM. But I look back on my record, and the term Rolling Stone comes to mind. I didn't last any longer than three years on any job I worked on, including when I was with IBM. Every three years, I changed jobs. And luckily, IBM had a variety of jobs that kept my interest going. So now I retired. So that's my career in 25 words at once. You told me a great story about the work on automating the ticker.

That work didn't go flawlessly, though, did it? Not really. I think it also illustrates the mentality of Wall Street investors. We took a ticker system that was based on pneumatic tubes. Reporters on the floor of the exchange would report trades on pieces of paper that they'd stick in the amount of tubes, pump it up to the third floor of 11 Wall, which is where the exchange was. And young ladies would take the slips out, get onto a teletype machine, actually a punch paper tape machine, punch it out to Reuters and Reuters would blast it across the world. That's basically the system we had to replace. We replaced it with a Marksense reader on the floor, and the reporter, instead of using pieces of paper,

he would use a Marksense card. They're using a pencil and a card, and a card in the shape of a punch card, actually. Read it into the reader, and the reader would basically feed it into the 360 configuration that we had. You would format a ticker message and put it out on ticker and send it out. In order to train the reporters on how to use the Marksense reader, we had a training system on the third floor which consisted of a Marksense reader. And instead of the tape that we sent out, it would come out on a Teletype machine, which is about

the Marksense reader. But there was a slight glitch in the program, my program actually, where instead of going out to the teletype printer, it went out over the ticker itself. And we had a fictitious trade of 1 million shares of XYZ. And XYZ does not exist on the floor of the exchange. And all of a sudden, there was a panic on the floor of people wanting to buy X, Y, Z, but they couldn't find the damn post for Australian, you know? So it hit the papers the next day and it was a chuckle,

but most people had, but my manager was not very happy with that. Credibility was a big, big, an important feature of the new york stock exchange so anyway that's that's the story and it was it was probably pretty hard to get the you know the wall street uh industry to um to believe in and have confidence in this new you know that was the big concern that was the big concern by the New York Stock Exchange. They pride themselves in being the blue chip exchange of New York, the world actually. And credibility is very, very important.

That mistrade is very, very important. The way they recovered from that was a message on the ticket following that ticker, the following day, which we did it during trading hours. We put a correction message out of the ticket saying, disregard this message of XYZ. But it was too late. The horses were out of the barn at that point. They were running like hell.

And so it gives you an idea of the mentality of how much an investigation goes on in investing on a hot stock, which they thought it was a hot stock. A company that didn't even exist. Yeah. So to rewind to the back because i'm super curious so you said that you basically set eyes on a computer for the first time in 56 um it was a punch card it was the solid state i think the univac solid state 80 yeah so that was 56 then you went uh and worked

for three years then went back to school and it sounds like you actually ran into Iverson Notation first. And that's what you mentioned by Ken Iverson's work. And then later on. I was in Columbia, yeah. Yeah, so that was at Columbia. Yeah. And then later on, you ran into APL. So for folks that don't know the timeline, if that's the that's the late fifties or 1960 at that time, when you're at Columbia,

APL doesn't exist at that point in time. It's not until 1962 that Ken Iverson actually publishes his book and they don't actually have an APL implemented that you can code it until I think it's 66 or 67. So that's super, super easy. I was working for a Univac 61 through 64. It was in Univac. I saw, I used the APL line. Oh, really? So that must, that must've been in its first, you know, in its early development stages. It was, it was a relatively new language as I recall. It was myself and this other guy that was very interested in it as well.

He introduced me to the Iverson notation, and I got hooked into it. And together, we were just screwing around with it to see what we could do with it. But I know that I was able to do a hell of a lot more with that language than I can with BASIC, which is what I was used to writing or an assembly language. But I was with UNIVAC between 61 and 64. That's when I discovered the APL language. Wow. That's even, I think, that must have been like an implementation by UNIVAC or something.

It may have been a beta version that we got our heads on. I got no idea. All I know is I didn't use it during that period of time. Wow. Wow. So you're actually probably one of the very, very earliest users of APL,

literally in the world. I wouldn't categorize myself as a user. I mean, we, we feel, we fooled around with it, but yeah, I did. We did not do any kind of productive work with it. I don't think I did any productive work with it either. And so what would you say of the languages that you've sort of programmed with? Cause it sounds uh the three apl or the three 360s that were tied together for that you mentioned a slew of languages fortran cobalt

algol um i'm not sure if you mentioned basic in that run and assembly was the main one uh did you spend like most of your career working with any one of those or did you were you basically a polyglot actually i was not the one working. I was a manager at that time. I had a staff of people who were working on it. My particular responsibility for that project was to develop the application side of it and to interface with IBM on a control program.

We laid down specs for the control program on what IBM was to do. But the team I had, we were basically application programs. And so how did the team or the folks that were working for you, did they, I guess at that time, like these days, a lot of times when you're building something, you have a team of engineers, not that there's a vote or something, but everyone has their, oh, I want to code this in C++ or Rust. It was not up to them. It was divided up by the function that they were addressing with the software go with the macro languages like Cobalt and, I'm sorry,

Fortran and Album. The more stringent type of, the more complicated type of programming, we usually wound up doing it in assembly language because we had the most control at that time. And when, so Eric, when did you enter this? So I assume, you know, back in 56, you clearly weren't alive yet. So at what point, at what point did you enter? Yeah. I don't know how old you are, but I'm guessing it was, you're not that old. No, no. I came into the picture in 1973. Yeah. That's when I was born.

And you said you got that first IBM PC in 86? I think so. Yeah. Because, you know, I know that there was a computer that was bought in 86. I don't know if it was the first IBM PC that my dad bought. It was. Okay. Okay. Okay. Because I know this because several years, maybe 10, 15 years ago,

I went down to my parents' basement and the machine was still there, actually, in the basement. And I was very excited because the keyboards that came with those PCs are legendary. The IBM model M keyboard. And, and so I, I, I took it. I mean, it was, it was filthy. It was like completely encrusted over. So I like meticulously took it apart and cleaned every piece of it and reassembled it and brought it back to Seattle with me, which is where I live. And, and I know,

I know it's from 1986 cause cause all IBM model M keyboards have, have their birth date on, on the back of them. And that keyboard was made in October, 1986. And do you still have it? so this is a really sad story oh no um yeah this is a sad story and i can tell you haven't seen my cpp con talk yet because i talk about it there so shame on you from this uh 2021 right yes okay i haven't seen it yet i haven't seen any of the cpp con talks because they're not they're like hidden on YouTube still, I think. No, no, this one is actually public on YouTube. You can find it now.

The keyboard traveled with me wherever I went and I loved it. And this is the keyboard that I use day to day. And I started working at Facebook and of course I brought it to work with me and it was there on my desk at Facebook when lockdown happened due to the pandemic. And I, you know, like the rest of the world, didn't really appreciate at the time exactly how long lockdown would happen. It would be so. So I left a lot of stuff at my desk i'm like i'm gonna come back to this in a few weeks it'll be fine um but um but then uh then you know like we were still off uh you know in lockdown and then like office moves happened and then uh i i changed jobs to NVIDIA. And like, I never got back to the office to reclaim my stuff. And nobody knows what happened to that keyboard. It has just vanished into thin air.

So I am like bereft. I am like heartbroken because this is the keyboard that I learned to program on in the 80s. It's the keyboard that I played, you know, Decathlon on in 1987. You know, it's the keyboard that I played Zork 1 on, you know, this text-based adventure game. And, you know, I remember sitting next to my, my best friend at the time, and we were, you know, hammering away on that keyboard, playing, playing games, like really awful eighties PC games, really bad games that we absolutely loved. And of course that,

that keyboard, you know, the reason why it's so legendary is because it's like absolutely indestructible. Like that keyboard, you know, the reason why it's so legendary is because it's like absolutely indestructible. Like that keyboard weighed more than like more than my laptop. You could clock someone over the head with it. It would do real damage. It had this big steel plate on the bottom of it, you know, to make it like super like heavy and stable. So it wouldn't bounce around when you were typing on it.

So anyway, like I have no idea what happened to this keyboard. It's out there somewhere. It unfortunately doesn't have my name on it. I should have written my name on it. But if somebody at Facebook hears this and they, you know, they have somehow inherited an IBM Model M keyboard, they don't know where it came from. And if they check the bottom and it says October 1986 on it, please reach out to me. Listeners, this is our mission to you. Help us

locate my lost IBM Model M keyboard. Yeah. This podcast now has two goals. One, to be number one in Slovenia. We love you, Slovenia. And two, to track down this keyboard. I mean, I'm not saying that's why Facebook rebranded to Meta. I've heard lots of reasons, but potentially Zuckerberg is trying to cover up the tracks of this keyboard. It sounds like this keyboard is probably worth quite a bit if you sold it online. I was, I was actually going to ask Otto, do you remember how much that IBM PC would have cost in like when you bought it in 86? I remember. You remember the amount that we paid? I paid for it. I believe it was $2,500. I'm sorry. It was $2,500. Yeah.

I remember, I remember mom making a stink about it because it was five grand. Five grand. You know, you're probably right. You're probably right. In today's dollars, it's well over $10,000. Yeah, yeah. Yeah, five grand.

And yeah, it was like a toy, right? It was like a pocket calculator. It was almost grounds for divorce divorce too, by the way. Yeah. Well, yeah, it's interesting. The reason I ask is because, you know, it's not something, you know, it was not a small purchase at the time. It was not common for people to have a computer in the house.

That's true. It was. But it was a must for me. It was a must have. You know, computers were very, very novel. I think people who were addicted to it, like I was, had to have it, okay, regardless of how much it cost. Like with an IBM machine and you were working for IBM at the time, did you get some sort of company discount?

I probably did. And I'm trying to remember how much. It was not a lot. In that respect, IBM was not that generous in terms of company discounts like that. The IBM, that particular model of IBM was hard to come by because they felt customers had to come first before the employees. So we were at the bottom of the food chain on this one. Was that an AT or an XT or what was that? It was an XT. I believe it was an XT. XT, I think, preceded an AT, didn't it? I'm not sure. I think, I don't remember.

I'll get it wrong. I thought the AT came first, but I'm not sure. I could be right. It's been over 50 years for me. So one thing I do know is that keyboard, like it has this long, like, telephone cable, this black telephone cable, right? And it's coil, so you could, like, you know, walk around with it, you know. It was portable. Yeah, sure.

And the plug, like where you stuck it into the computer, it was this big fat, like nine pin, I think, nine pin thing. And those plugs are called AT plugs. So that's why I think the AT came first because that was probably the first computer that used that plug and so it got called the AT plug.

It looks like the AT came second. The AT was 84. The XT was 83. I had the earlier one. Is this just a computer model for those that are out of the loop like me? I think that was an 8088 processor. 8088. You're right.

I think, but this is like the IBM's first home PC. Yeah. Was that the first one, the XT? I believe it was the first one. But the keyboard was not the first one. That came out of its office products division. That was being used by the office products for their devices. So the keyboard preceded that desktop

long before the desktop was created. I didn't know. So that was like the second chip that Intel made, basically. Oh. Yeah. That's what I learned to program on. Wow. Yeah. I was writing was writing basic for that.

Really bad text-based adventure games. One event that sort of made me consider Eric for computers, that he wrote a basic program, I think while he was in his early teens. And it was a very simple program, but it tested my hearing. And it discovered that I had a hearing loss in the higher range frequencies, which happens to be true today. But he discovered it when he was, I think, 14 or 15 years old with his basic program. I knew that he was meant for computers, even though he fought it tooth and nail. I knew he was meant to be for computers. It was the simplest program. It was probably like

10 lines of code. Somehow I learned how to control the speaker and play tones. And it just started real low and went up real high. And it was like, did you hear this? Yes or no. Right. And we were absolutely stunned when when my dad was like, I think there's something wrong with the program. Like it didn't. And we were all dumbfounded because you're like, you didn't hear that. Wow. So at first you at first you thought that the program didn't work uh and that maybe he wasn't uh destined to be computed and then you realize don't wait it does work it's just a couple lines it's my hearing that's uh not working wow that's uh that's pretty amazing um that you were but you were like a little medical

programmer back then. I guess so. That and the text-based adventure games, the only programs I actually remember. Oh, that and copying programs out of Byte Magazine with my friend, you know, and trying to get those programs to work. But they never did, you know, because, you know, the print in those Byte Magazines was like, OK, I can't tell if this is an O or a zero or, you know, what, you know, the print in those bite magazines was like, okay, I can't tell if this is an O or a zero or, you know, what, what, what, what is this thing anyway? And then, and then you'd make like one typo and the whole program wouldn't work. And then you'd have to go back meticulously and find, you know, it was, it was pain, painful. But I guess that was, that was the fun of programming back then. I don't know. Yeah. Because that used to be how,

for all the young listeners, that used to be how programs would get distributed before you... Yeah, that was open source. ...sent them around on floppy disks. They'd be in some magazine. You'd have to... They'd be in a magazine.

You had to type it in. Yeah, exactly. That was programming in the early days of my early days, you know, not my dad's era. Oh, my days. The actual testing of a program was a hell of a lot more difficult than it is today. We would get maybe one or two turnarounds a day. And by a turnaround, it's submitting a test to a computer and getting the results back within one day. So the term interactive programming did not exist with us.

It was you had to literally assemble, test and then print out the core dump. And from that, you were able to debug a program. So we were lucky if we got one, if not two turnarounds a day. Yeah, that's that's actually it's very similar to when I worked in supercomputing, high performance computing. If you had an application where you were running it at scale, you know, you were running it on half the supercomputer, the whole supercomputer, and you had some bug at that scale, you know, it would take you 24 hours or 48 hours to submit the job to the system that you know would go and run there and then you'd get your result back and and you wouldn't you wouldn't it wouldn't be interactive at all it would run you know at 3 a.m sometime and you'd have to do your best to debug based on what you got back well we had a few other complicating factors we had something called punch cards that you guys don't use. Most of the programs are on punch cards. So in addition to debugging the program, we had to get

a key puncher lined up to repunch the cards and then have someone feed the cards in along with the test material and generate a printout after that. So it was a little more involved than that. It really was. We had also programs. The programs would, at least on a Univac, early Univac machine, I programmed the Univac 3 originally. We had a drum arrangement where the drum actually laid the program. I'm sorry, the program would lay the actual instructions onto a drum. And to execute it, it would execute off the drum. So after debugging the program, you had to go through another exercise where you had to do a, it was a lattice refinement where you had to reduce

or replace the words so it would be an efficient execution of the program. The drum would rotate, and hopefully by the time the drum rotated to the next instruction, it would then execute. Now, if it missed that word and go to another rotation, you had a latency problem which slowed the program down. So in addition to debugging the program, we had to make a latest lattice refinement, make it an efficient distribution of words on a drum itself. So it ran efficiently.

It's like pipelining in the 1950s. Yeah. Yeah. Yeah. I, wow. That's amazing. I've never, you know, that that's something that, you know, today is done by your processor for you or by your compiler at the worst. It's amazing to think that you used to have to do that by hand. We had, we had to do that ourselves. So that shows you how far you guys progressed. When you were writing a program, like when you were doing the development,

would you just do it on paper and then you'd transfer it to the punch card? We basically used the same techniques you use today. You float shot the logic and you basically translate that into machine language or compiler language, whichever one is appropriate. This would be on coding sheets. You submit the coding sheets to a key punch operator who would punch up the instructions, punch it into the cards and basically submit that to the operator who would do the test. You have to provide your own test material as well, which includes tapes. We did not have things like random access storage at that time.

So we had to provide data in the form of tape as well. And they basically then ran it. And if it didn't work, they would dump it and put it onto a printer and say, fix it. That's basically the process at that time. We thought it was fabulous. The whole technique, we thought it was very, very efficient. But it shows you how far we've gone. Thanks for listening.

We hope you enjoyed and stay tuned for part two next week.