Episode 822 transcript

FLOSS-822

Jonathan: Hey folks, this week, Rob joins me and we talk with Shimon Shachin about NAND to Tetris. That is the open course for college and high school students, all about starting with the NAND block as the basic building block of computer architecture, designing an ALU, a CPU, a full computer your own programming language, a compiler, all sorts of great stuff, and eventually ending up with the playable game of Tetris.

It's a lot of fun, and you don't want to miss it, so stay tuned. This is Floss Weekly, episode 822, recorded Tuesday, February the 25th. Nand to Tetris.

Hey folks, it's time for Floss Weekly. That's the show about free, libre, and open source software. I'm your host This is Jonathan Bennett, and today we've got a lot of fun planned. We've been fighting with technical difficulties, but that's okay. We're used to that. We're Linux users. And speaking of Linux users, I've got a co host today.

The one, the only Mr. Robert P. Campbell.

Rob: Excited to be here as always. Yeah, it'll be

Jonathan: fun. So you've got, you've got a little, a little bit of a teaching experience, right? Sort of on the, on the college level.

Rob: I have a little bit of college teaching experience, taught for a year mostly computer hardware software kind of like A plus stuff and then a little basically a security plus class.

So I've done a little bit nothing really in the programming side of of teaching.

Jonathan: Still, you've done more teaching at that level than I have. So I think probably than most of our other co hosts. So that's why it came to mind to invite you on because today we have Shimon Shacham of the Nand to Tetris.

Project and book and course and I have been recommending this for a long time since I first found out about it I just immediately went. Oh my goodness. That's one of the coolest things i've ever seen And so I reached out several months ago. I reached out to shimon and said Hey, would you be on the show and he was he was kind enough to to agree to it And this morning we went to get everything set up and ran into some technical difficulties So i'm gonna bring shimon on the show and he could say hi But we won't for those of you watching we won't actually be able to see him You have to be stuck with looking at Rob and I but welcome, welcome, sir, to the show.

I'm glad you're able to be here, even though it's just as a sort of disembodied voice today.

Shimon: Yeah, sure. Very glad to be here with you folks. And sorry that we have some camera problems, but I'm not, I'm not a great looker, so the voice will

Jonathan: do better. Alright, so I'm gonna start with the assumption that most people listening have no idea what we're talking about.

Give us kind of the intro, like what is Nand to Tetris, where would someone run into it, and why should geeks like us find it cool?

Shimon: Well, Nand to Techbase is a project, or a course, or a book, as you said, that guides you through the process of doing actually 12 standalone projects, 6 hardware projects and 6 software projects, in which You start with the humble, low level NAND gate, which is possibly the lowest logic gate that you can think of, and work your way through building a complete, modern, general purpose computer system.

Consisting of a typical von Neumann hardware platform and a typical software stack consisting of an assembler, a virtual machine, a compiler. And a little operating system so that at the end of this journey, you have a computer on which you can run or write and execute any program that comes to your mind, including, of course, Tetris.

So. That's the story.

Jonathan: Yeah. Where did, where did the idea for this come from? What was the spark of inspiration where you said, let's, let's put all of this together in a course?

Shimon: Well, I developed this course together with my colleague Noam Nisan, who is like me, a computer science professor. And both Norm and I served as, um, directors of various programs computer science programs, undergraduate, graduate, and so on.

And we were always frustrated from the fact that we are teaching Many different fragmented courses and students are getting all sorts of I, I, I should say rather deep views about various corners of computer science and, you know, mostly applied computer science, but they have no idea how things work together.

So, you know, they may understand the hardware and they understand the compilation. They may know to write programs in Java and Python, but there's no one course that puts all these pieces together and lets them sort of see the forest for the trees. So, the idea was to To sort of plug all these holes, all these interfaces between the different courses that one typically gets in a computer science program and glue things together.

So this was sort of the underlying pedagogical motivation. And the other motivation was just the, you know, the fun of building something big. And And, and taking the time to do it because both of us are, you know, busy people like, you know, all of us. And so this became kind of on the side work of love that that we did for several years, you know, before we actually.

Jonathan: I would, I would pull my copy of the book out and show it to everybody. It is on loan to a teenager here in town that I heard he was, you know, interested in computers and languages, and he'd done a little bit of programming. It's like, I want to take the next step. Well, if you're adventurous, I've got something for you to try.

So that's where my copy of the book is at. And I've, I've got to admit, I've not made it all the way through the process. I've, I've built the, the ALU. I may have finished the CPU design. I was somewhere, somewhere along those steps when. Real, real life busyness took over and I had to step away from it. But I was actually, and this is what I want to ask you about.

I was actually going through one of the courses that are offered. And so this, this is a course that just anyone that's out there that wants to take it. There are, there are some opportunities for just someone to jump on the internet and take it, right?

Shimon: Yeah, absolutely. The, the course is available presently in Coursera.

And pretty soon it'll be available in, in other platforms as well and anyone can take it. And I think you have to pay something to Coursera in order to get a grade in the course.

Yeah. But you don't

have to get, but you don't have to get a grade. You can just take the course. As an auditor and do everything, do all the projects.

And, you know, when you do a project, you submit it to a server that we control. So even, even if you submit it through Coursera, it ends up in our server. So anyone, you don't have to be a registered to Coursera to submit any, to submit stuff. And so you submit it to our server. We grade it, you know, we, we, we test everything that you do.

We, we, we send you a grade. And and then, you know, the grade is just a nice feedback that you did what was expected for you. And so, you know, because we work with a version of HDL, how to description language that we Designed for our course. And we also wrote a language that, you know, we called it TDL, Testing Description Language.

And we wrote many scripts for, for every chip that you develop. And so when you submit your chip to our server, we run a test script on it. And if the chip produces the expected outputs, you know, we compare it to a compare file, then, you know, you get a pass grade or, you know, whatever you want to call it.

And then the student knows that. that he or she did what was expected. And so anyone can take this course. And surprisingly, you know, we didn't expect it, but it turns out that the hardware part of the course does not require any prerequisite knowledge. Now, you don't have to be a computer science student because everything is based on the basic logic.

And you know, logic is something that anyone can understand or pick up easily. Yep. So You can build, you know, you mentioned the ALU. So the ALU is built, you know, is designed on, you know, it requires to put together some, you know, endgates and muxes and so on. Nothing much more beyond that. And then you have a working ALU, which is really exciting.

You know, moving on to the CPU is, you know, a few more steps. It gets a little bit more complex because now you need what is known as sequential logic. You know, you need to print this clock and so on. But once again, we have very detailed lectures, video lectures. The book is very detailed. And so.

Anyone, practically speaking, including high school students, can, can build a computer. The hardware part of the computer.

Jonathan: So I wanna, I wanna dive more into that. I, I have a question first though, because you mentioned it. That you guys have your grading server. And I'm sure Rob has questions he wants to get to as well.

But you have your grading server where people send in their programs and you kind of simulate them. Has, has anyone ever tried to be particularly clever and do something like really surprising? Have you, have you had anyone, you know, escape your sandbox and, and cause havoc on the grading server?

Shimon: Not that I remember, but no, it's an interesting thought.

Yeah, I should have expected questions like that from you guys. But no, I think so far, you know, people behaved like happy campers. And I think all they want to do is submit a cheap. Get a grade, you know, stay alive and move on to the next gyp, so.

Rob: Yeah. Go ahead, Rob. So, you know, this being all online, is the book required to take this course or is that supplemental material?

Shimon: The book is not required. It's everything that written in the book is given in the course lectures. It's nice to have the book, you know, you know, in your hand. It's either as a souvenir or as some additional source of documentation, but but it's not required. And and so I think most people who take the course don't buy the book.

Rob: Yeah, but it is reasonably priced, especially for any kind of course textbook kind of kind of book. Cause I, I took, looked it up before the show and US dollars, it's on Amazon, it's less than 30. So that's definitely reasonable for most people.

Shimon: Yeah, we insisted when we When we made the deal with the publisher, we insisted, we insisted that the price would be relatively low because, you know, many textbooks in computer science and maybe in other subjects as well are very expensive.

So in our case, it's I think the price is okay.

Jonathan: Oh yeah, definitely. And also, I

Shimon: must say that we put half of the book online, so the first half that I mentioned, the hardware part, is available in PDF form online, and so that's another reason why, you know, you don't need to buy the book, although I'm happy to say that many people buy it, so.

That's always

Jonathan: nice.

Shimon: Yeah. You're welcome. Yeah, I think it's a nice souvenir.

Jonathan: Yeah, oh, I would encourage anybody that wants to go through the class, or even that just wants to learn this stuff, buy the book. It's a great book as well. And to your point of it being simple enough for anyone to go through, that was what I was struck by, is that like, the way, not only the way you approach it, but even the way The whole, the Von Neumann architecture works.

When you start with the NAND gates, it's all very incremental. Like, there's, this is a weird thing to say. There's nothing about modern computing that's actually black magic, right? It's all very incremental, one layer on top of another. And once you understand what this layer does, the next one really makes sense.

And it's, I would, I would make the argument that it really is accessible for a lot of people if they're just, if they have the time to do it and are willing to put the work in. And that's one of the things that I think is just the coolest about this is that it gives someone that, that top to bottom understanding of a, admittedly, a very simple computer in comparison to the, you know, the behemoths we have on our desks, but it's still the, the exact same principles at play.

And that's, that's, it's, it's a service. honestly to the community. And so for that, thank you.

Shimon: Yeah. Well, thanks. Thanks for making this observation, which is very, very true. You know, both of us are computer scientists Norm and myself, and we are trained and and, and, and we enjoy. You know, this, this trend on and enjoy logic and and first principles.

And, you know, whenever we write the paper or do some work, we have to start with the first principles and prove, you know, every step in the way until we can make a solid point about anything. And we, we decided to build the course exactly in the, in that, in that spirit. So as you mentioned, you know, the computers that you use today are, are monsters and or monstrous and, and they are so complicated that that it's very difficult to.

To explain, you know, how a real computer works because there's so many layers of hardware and software and proprietary stuff and so many interfaces that it's impossible to explain how a real modern computer works in a typical academic course or to some, you know, layman or high school student. So one of the first decisions that we made.

Was that we're going to design an architecture, which first of all, will be relatively easy to build. And second of all, we'll have will be, you know, I guess I should use this word that it will be beautiful, something which is very elegant and very simple and the same time, you know, very powerful. So and I'm glad that You know, we, we managed to do it.

Jonathan: Yeah. Yeah. It's good stuff. With that idea of, of beauty said, I'm curious about the, the license behind all this, right? So we're open source enthusiasts here, and obviously you guys have made this sort of an open course. It's on open course where it's, it's open, open source friendly. What I guess the book itself, you said about half of it is completely available online.

And then I know a lot of the example code there's an IDE, I think. that is also open source. What what's the, what's the license and what's the story with all of that?

Shimon: Well, everything is wide open. I don't even remember the exact license that we are using, but it's a common what's the term?

I forgot.

Jonathan: I think, I think some of it's creative commons at least.

Shimon: Yeah, exactly. It's creative commons. Anyone can, uh, you know, fork our ID and, you know, do whatever he or she wants with it. And Same goes with the lectures. You know, it's, it's I think one reason why this course became very popular is that because not only anyone can take it, it's also anyone can teach it.

So, if you want to teach this course, all you have to do is send me an email, and I'll give you access to all the lecture materials and all the projects you know, past exams and so on. So, very easy to Adopt the course and teach it anywhere you want and as a result the course is now being taught in something like 400 universities and that's awesome numerous numerous high schools and hacker clubs and and whatnot so so yeah, it's widely available

Jonathan: I think it really appeals to that sort of original hacker mentality of, I want to understand this thing on my desk as much as I can from top to bottom.

And as far as I know, Nanda Tetris is about the only course that really offers that. I'm not aware of anything else.

Shimon: I think it's the only course that offers a complete ascent, if that's the right word, from basic logic gates to a high level programming language. There are quite a few courses that teach you how to build a computer, but typically they get stuck with well, getting stuck is not the right word.

They, they go all the way up to a symbolic machine language and an assembler. And that's it. We, we go the extra, you know, distance and we, we also cover a very nice gap between machine language and high level language. It looks pretty much like Java. You know, we, we decided to design our own language.

We call it Jack. Now the machine is called Hack. The language is called Jack. And it's a Java lookalike with object orientation and no inheritance. That's one major concession that we made. And, you know, we make all sorts of concessions so that students will be able to build significant tools like compilers and little operating systems in maybe two or three weeks, you know, each.

So, so you have to, you know, you have to. make some compromises. And also, you know, when you, when you stick, when you, when you, let me give you another example. Take a compilation. If you look at a typical compilation course, college level, a great deal of the course is devoted to handling errors.

Compilation errors. Now, this is, you know, terribly important, you know, error diagnostics, error handling, and so on. But, when you deal with errors, you often sort of lose sight of the most fundamental ideas, which are program translation. You know, taking one language, parsing it, analyzing it, and generate, generating code from it to another language, which I think is, is the essence of, of translation.

And so we decided that when we write a compiler, we will make an assumption that the source code is error free. There are no errors, and once you don't have to deal with errors, you can focus on, on the most important ideas in compilation, in my view. And so we made several such concessions in, in, in many different steps along the way.

And the nice thing about all these so called missing things, in Antutu is that every one of them is a great extension project. You know, you want to handle errors? Great. You know, you want to add some features to the high level language? It's a very nice extension of the compiler, you know. So, for example, we don't have a for loop.

You know, we have while, we have if. We don't have for. You want for? Great, edit. You know, we give you everything that you need in order to edit yourself. And so, indeed, over the years people made all sorts of incredible extensions and additions to the BASIC platform. And which, you know, gives us great joy too.

Every once in a while, I get an email from someone who says, look, look what I did. And and I try to publish some of in in the non detectors website. So, You know, we have a section called cool stuff in which you can see some of the extension that people built. That's

Rob: great. So, so in this project, you build up a virtual computer, virtualized, you know, real chips and all up to the computer to, I think you said, like, even like an assembly language layer to your own Java like programming.

And you said, you know, the language is the Java language is your own, but is the assembly like at the step where you're at assembly, is that like. Real assembly or is that assembly like that works with the chips that are made in this? Yeah,

Shimon: it's the assembly language is also our own quote unquote, proprietary language.

I mean, we made it up and, you know, when you build a computer, you have to design also the instruction set. And so, we designed the instruction set, and it comes in both binary and symbolic versions, and it's a very elegant language. And it's, it's really a pleasure to write programs in our assembly language.

And it's, it's much more user friendly than a typical assembly language. And you know, a little bit difficult to talk about it. I mean, you have to put your hands and play with it and write some programs and see for yourself. But it's I think students who write were introduced to assembly language using Our language, you know, gets very smooth and user friendly intro.

And indeed, this was, you know, we made many different sort of major decisions along the way, and one of them was should we develop sort of a stripped down version of an x86 or a RISC machine? And, or, or maybe design our own. And, and we decided that if you want to do this elegantly we should, we should do, we have to build our own our own design.

So that's what we did. So the assembly, once again, is our own language.

Jonathan: Yeah. It's real assembly, but it's custom.

Rob: Yeah.

Shimon: Exactly.

Rob: So, so having that assembly layer theoretically one could create their own. language above that instead of using your, the jack, I think you called it. Yeah, or change their jack, however they want.

Shimon: Well, yeah, you can. Well, you know, first of all, you can easily, and this is something you can do in high school also, you can build a macro assembler and design all sorts of macro commands and bring it to a level where you have a C like language or a basic like language. And And this, you know, is a significant step up in writing high level code.

But if you want to implement a powerful object based language, you need a compiler. I mean, you have to, you have to go through parsing, code generation, and we decided also to introduce a virtual machine. And the reason we did that was because You know, most computer science students today work with either Java or Python, and both Java and Python have a two tier translation model.

So the high level code is translated into some virtual machine, some intermediate language, and then the intermediate language is translated further into assembly. So that's what we do also in our course. You know, we thought it's important to, to train students to understand this elusive VM that sits in the middle of everything they do.

Jonathan: I, I, I'm looking at the, the cool stuff and one of the ones here that basically for as long as I've known about Nand to Tetris, it's what I've wanted to do with it. And that is running, running the, the virtual CPU as a real CPU in an FPGA. And I assume I'm not the only one that just immediately thought of that and thought it would be the coolest thing to do.

Shimon: Yes, absolutely. It's a very nice project in which you take the HDL programs that we wrote that you wrote in Nandu Tetris, and you Quite easily migrate them into either Verilog or VHDL. It's a very simple process. And then using the incredible FPGA world that we have today, you can actually sort of commit this whole thing to silicon.

And have a little board that, you know, with which you can play Tetris and any other program that you can think of and, you know, show off this board in job interviews or to your parents or whatever and, you know, tell them, you know, that that's a computer that, that I built from scratch and built it all the way from NAND.

It's a very empowering feeling.

Jonathan: Has anybody ever laid it out on, on actual NAND chips without the FPGA?

Shimon: Yes. There was, there was one project in which, uh, someone built the whole thing from using what is known as RTL or sort of real elementary logic gates. And, I don't remember if you did, if you went all the way up or just built the the CPU, but yes, people did that as well.

Jonathan: Yeah. I, so I'm sure you're familiar with Tiny Tape Out.

Yes. Yeah. Has any, has anybody managed to convince the, the powers that be that, The NAND to Tetris CPU, that it should be on one of the Tiny Tape Out builds? I don't know,

Shimon: I have to check, I don't know.

Jonathan: I see here where it looks like someone tried to, I just, I can't, I can't find out for sure if they managed to You know, because lots of people send in their designs for Tiny Tape Out.

For those that don't know, Tiny Tape Out is, I think Google is one of the companies behind it, but several companies. What they do is they say, okay, we're going to do a run of this. It's an older stepping, but a run of a lithographic process to build, essentially like you would build CPUs. But instead of doing one big CPU design, they'll say, people from the community, give us your designs, and they'll pick.

Oh, I don't know, 40 or so designs because they're all much smaller than a modern CPU. And so they'll do chips with all 40 of those designs on them and then just pin them out differently. And so that you can get, just like students and people researching, they can get real hardware designs done in lithography on these chips.

And being able to get a hacked CPU on a tiny tape out chip would just be It would be the, the, the ultimate. Well, I mean, it would be a real feather in y'all's cap too, but it would, it would be neat.

Shimon: Yeah. Actually, I have a friend by the name of URI, who who is very active in this in this area.

And, uh, we've been talking about it. So I'm sure that we'll do it. Pretty soon. And so if we get it done, I'll let you know. And I'm sure you know, maybe you can put something about it in a heck of a definitely. Yes, definitely.

Jonathan: That is our that sort of thing is our bread and butter. So I okay.

Every student that goes through this, do they sort of end up with the same design? Or like, is there, is there room for creative differences? Right? So like, am I gonna put all of my instructions are the same, you know, machine codes gonna be the same instructions for everybody that goes through it?

Just for one example.

Shimon: Yeah, well

you know, this course is It's very demanding because every week you have to come up with a significant module, like like, for instance, in project one, you have to design 1515 basic chips, and then in project two, you have to design an ALU, in project three, you have to design memory unit, RAM, and so on.

So, there isn't much time to come up with your own designs. And yet, toward the end of the course, if time permits, We have a special session, sort of one or two weeks, in which we encourage students to come with all sorts of extensions. So, it's not that they come up with different computer models, but they come up with different extensions.

So, one student may develop a literal disk. Another student can develop a nice extension to the high level language. There are many different possible extensions. You know, you can develop an, you know, HTTP server using our system, you know, make two head computers begin talking to each other and so on.

By the way, I mentioned before I forget, I mentioned Uri Shaked his website is called walkway. com. It's W O K W I. com. And it's a wonderful website for TinyTap and TinyTapOut and related projects. So you may want to take a look at this and maybe interview him at some point.

Jonathan: Yeah, yeah, that'd be great.

So to that, to that question about the differences between different students, you got in, in the course, the, the sort of the framework is provided then, like you, you guys have already assigned, you know, this machine code is going to be this instruction. You know, this one is gonna be this one. This is the way that the CPU is gonna be pinned out.

Shimon: Yeah. Well, is that, you know, we, we provide only the OB obstructions, you know, we mm-hmm. We describe, you know, if you take, for example, the alu, we describe the language or we describe the the, the arithmetic and logical operations that the air should perform. Mm-hmm . And, but, but we say nothing about implementation.

You know, we expect students to come up with their own implementations. Mm-hmm . So you have to decide how you put together the lower level chips in order to deliver the the required functionality. So the functionality is the obstruction that's on us. The implementation is of new. That, that's the deal in, in every one of our projects.

You know, we and you know, one reason why. Students can build, for example, a compiler in two weeks is because we, we are very, you know, very careful in giving, giving them very specific obstructions, very specific specifications, and We also give them the design in the form of APIs. So we might give them an API for a passer, an API for, for a code generator.

And the APIs are proposed. They're not you know, they're not mandatory. Mm-hmm . But if you follow our APIs, you will realize that. You know, every method that we specify in the API comes with a, with a test method that we wrote. So you can immediately put your work to test and you can, you know, you can unit test everything.

So, once again, you know, you get the overall design. Which is what I call an obstruction. And then you have to implement it. And you can implement it, you know, in any language that you want. And that's another nice feature of the course. You know, if you build a compiler, you can build it in Java or Python or Scala or Ah, right, right, right.

Yeah, or Perl is probably the best language for these things. It's up to you.

Jonathan: How about in Jack? Is it possible to build the, is it possible to eat, for Jack to eat its own dog food?

Shimon: It's, it's difficult. It's difficult. It's, it's possible. But, but it's, it's challenging. It's, it's a nice project. Yep,

Jonathan: understood.

So it's, it's one of those good add ons where someone can do it if they want to. But you're not going to require it from everybody. So that answers the question that I was thinking of, so like, if one person writes A Jack program on their CPU, they can share it with their classmate or someone around the world for that matter, and pretty much know that it's going to run on the CPU and the, the, the system that someone else designed because you guys kind of lay out the, the, the, the bones.

Maybe we could say that that assures that all of this, for the most part will work together the way it's supposed to. Yeah.

Shimon: Executed on any machine that was built in any of the Tetris cores.

Rob: Are there, are there any Jack repositories? Well, we don't have a

Shimon: repository. We should, yeah, we should build one. There are numerous computer games that people built with our computer. If you go to YouTube and type non tetris games, you'll find, you know, thousands of games.

And these are typical, typically they are kind of classical retro games like Space Invaders, Snake, Tetris, Sokoban, Maze Games, and so on. And You know, Jack is surprisingly good for interactive graphics applications, you know, as long as you don't do 3D modeling and things like that. Although, I must say that, you know, some people even did ray tracing and very fancy image processing without computer, which was quite astonishing to see.

And yet, once again, most programs are kind of classical computer games and there are thousands of them.

Jonathan: I, I see a YouTube video here. It's linked to from your cool stuff page. It's Hackenstein 3D, and it's a it's black and white, but it's sort of a re implementation of the old Wolfenstein 3D game. Yeah, and

Shimon: it's impressive.

It's impressive. It's like, you know, real time graphics. It's a very nice game.

Jonathan: Yes. Yes, very much so. It's very cool.

Rob: Do you know if anyone's re implemented your jack on just a regular computer at all?

Shimon: Yeah, that's a great question. And I don't know, but I think it's a very nice project, you know, to implement Jack on, you know, maybe on, I don't know, on your cell phone or whatever.

There's no reason why it cannot cannot be done because all you have to do is, you know, you have to generate VM code. For a virtual machine, and then you have to write a VM translator that translates the VM code into the assembly of any computer that that is out there, you know, so what is missing is is that trust is sort of what is Sometimes referred to as the back end of the compiler That's the the translator that translates VM commands You know, push, pop, and so on to assembly.

And the target assembly can be on any computer's assembly language.

Jonathan: Yeah, that's cool. So obviously there's, there's a lot of these add on projects where people have done extremely impressive things. Have you guys thought about a follow on course, like a next year? Because I think Nanda Tetris is a full year long, right?

Shimon: Well, it depends. I think the recommended pace is, yes, it's a, well, I'm kind of, I'm, I'm, I'm selecting my words carefully because I'm, you know, I guess some people are hearing this and they're not sure, you know, what's the right format to teach this course. So I should say that in many universities, Nantu Tetris is taught in one semester only.

Jonathan: Okay,

Shimon: but it's very intensive and you know, it's, it's, it's also tends to cannibalize, you know, many other courses because the students are very engaged and they end up spending, I don't know, 15, 20 hours a week just on this course. And so that's, that's what many people do. That's what I do. But in other universities, it's It's being taught in two semesters.

One semester to build the hardware, another semester to build the software, which I think is a more sort of normal pace. So it depends. It's either a two semester or one semester course.

Jonathan: And then the follow on question to that is, have you guys thought about a second cut to it, you know, once you have the simple software

Shimon: done?

Yeah, so, you know, we've been approached by many people who wanted to do something like that. And we always Encourage other people to do it, but I'm not sure that we have enough energy to develop a sequel. But you know, we've been thinking about some, some wild things like, believe it or not, even something I called Named to Life.

And Named to Life is a course in which you, you know, in the same style of doing everything from the ground up, you begin by building You know, very basic sort of DNA sequences.

And then

from the DNA sequences, you build proteins. And from the proteins, you build ribosomes. And before you know what, you have a little functioning cell.

Yeah.

You know, so, so that's, you know, one thing that we've been thinking about. And in the same spirit. We thought about building logic itself, you know or a Turing machine or something like this from basic logical constructs in the process explaining, you know, what is an axiom, what is the Theorem, what is the proposition and building everything from the ground up in the form of little objects, you know, that you can build in a language like Python.

And so I think, you know, the basic spirit of things, you know, which I think is very characteristic in what, what Norm and I are doing is that I should say, you know, what innovation of non contemporaries. is this notion of bottom up construction. And I'm saying this because computer science is very much a top down field.

And when you start a computer science program, the first thing that you do is you learn a language like Java or Python, you know, which is, if you think about it, it's it's, it's something which is so farther away. From the real metal, you know, from the real computer. And so, you begin with the most high level things in computer science, and then, through the program, throughout the program, you work your way, sort of, downward.

And at some point, you begin to talk to learn about hardware, operating systems, typically TEPPEN. Your third year of studies or something like that. And we, I think Nantou Tetris has shown that there's also a great virtue in doing things the other way around and starting from scratch, you know, and building everything from very simple building blocks.

And, you know, one reason why it works is because it's very exciting. You know, it's very exciting to do something simple and see it becoming more and more powerful, and more and more potent, and to understand every step along the way, which is terribly important. You know, there's nothing hidden. You know, everything is out there, and you can play with it and extend it and do whatever you want with it.

So So unlike a typical computer science student who who thinks that the underlying computer, you know, maybe, I don't know, a coffee machine or something, you know, you don't know what's really working there under the hood. In Nandu Tetris, you build the whole damn thing, you know, from, from A to Z. And I remember when we, when we, when we wrote The abstract of the book, and we sent it to the publisher.

The publisher sent it for a review, and two of the reviewers were two well known professors who wrote a wonderful book called Structure and Interpretation of Computer Programs. It's a very famous book that came out from, from MIT. And it's the ultimate top down computer book, because it starts with a language called LISP.

And, you know.

Which is a very high level,

yeah, very, very high level abstract language. And so when I heard that they're going to review our book, you know, I was kind of very worried. But the review that they wrote was very warm. It was, you know, exactly the opposite of what I thought. They said that you know, what a refreshing thing it is to see a bottom up approach to teaching computer science.

Because we, we used to do everything the other way around. So, so I think that this, this notion of bottom up Construction has you know, many people find it engaging, many educators, and so I'm sure that we'll see more non tutorialist courses non tutorialist like courses in, in, in the future.

Different areas. I'll give you one more example communications and networks, you know would be very nice to build, you know, all the, all the protocols, you know, from, you know, starting with having two machines that send bits to each other. And then build the whole, you know, OSI layers and everything and end up with a web browser and an HTTP server somewhere.

So that's another very nice project, you know, sort of, you know, building the internet.

I'm also, I'm

Jonathan: also struck by the idea that we should show the electrical engineers some love and that there should be an electrons to NAND course that actually talks about how you get to the point of a NAND gate.

Shimon: Yeah, that's a huge

Jonathan: missing,

Shimon: you know, missing gap in our journey. I've seen

Jonathan: somewhere where someone has written a chapter zero, there may be more than one, but at least one person has written a chapter zero talking about, okay, this black box that we're calling a NAND gate in this course, what's actually inside of it?

Rob: That'd be a great prequel.

Jonathan: Yeah. Yeah,

Shimon: absolutely.

Jonathan: Yeah, it, it really, it really does seem like there's a, there's a, there's a great niche there for writing a couple more courses, maybe more than just two that sort of slot into this same idea. And I think, I think that one would be great. The Electron to NAND as sort of an intro to electronics, getting up to the NAND and then.

You know, maybe on the other side of it, there needs to be a build your own GPU, you know, a very, a very simple GPU, like the old retro computers used to have that can do sprites and such. But then that idea of building network protocols and, you know, maybe you re implement token ring and then finally Ethernet that.

That would be very cool. And then, of course, on top of all of that, you have to, you have to talk about computer security. And so then, you know, hacking the hack chip just naturally falls in there.

Shimon: Yeah, absolutely. You know, another, another very important building block in, in, in all such courses is to build.

A suite of simulators that kick in, kick in, in every major step along the way.

So

for example, in Antutu, we have, you mentioned, you know, we have our IDE and the IDE includes a hardware simulator, a VM emulator. a CPU emulator, an assembler, and so whenever you build a significant piece of software, you can actually test it and play with it before you actually set out to build it.

And we have executable versions of everything, of the compiler, the VM translator, the operating system. So And you have to do something like that in every non tetris like course, you know. So, you know, you play with DNA. You need some, you know, DNA simulator. You, you, you get it up to the level of proteins.

You need some, you know, some way to, to simulate how protein Things are, you know, being constructed, and so, and when you do that, you also get some very nice decoupling of the various topics, you know, you can sort of teach, teach the course in different orders, if you like so yeah, you know, I was thinking about you know, preparing a little lecture that sort of focuses on On the various tricks that, that Norm and I used in order to make this course, uh, feasible.

And, uh, you know, all the concessions that we made, all the tools that we developed, and, uh, coming up with a whole litany of, of, of advice, of tips to, to people who want to develop similar courses. You know to find time to do that.

Jonathan: Yeah, that would be, that would be really interesting. I, I, I like I like this style of teaching and learning a lot.

So in, in my high school education, one of the educators it's actually a quote from scripture from, yeah, and it's to him that understandeth knowledge is easy. And I am, that was, that was very much their viewpoint on things. And I'm really seeing echoes of that in the Nanda Tetris course, because it's all about, if you understand the layer below, then when you go to the layer above that and go to work on it, you understand it.

So actually doing the work is fairly easy. And I think, I think that's, I think that's great. I just. I'm just in love with the course. It, I, it makes me want to go back and start it all over again, get my book back or go buy another copy and actually go through it again. I don't think I have actually. I have to

Shimon: tell you that, you know, the best, the best way to learn it and to build a computer is to actually teach it.

So, you know, find a few people who are interested in this stuff and just teaching them to test this. Tetris course, it will be a great pleasure. Oh yeah. Hmm.

Rob: Community class.

Jonathan: Do a community class.

Shimon: Yeah, exactly. Exactly. That's the way to do it.

Jonathan: That would be a lot of fun. I don't know if I have time to do that.

I've got, I've got too many businesses I'm already running and juggling, but it would be a lot of fun. Maybe one of these days. Speaking of which, if someone wants to either take the class or run a group through the class, what are some where would you tell them to start?

Shimon: Well you have to go to nanddetectives.

org, our website, and Everything's there. It's, it's NAND, N A N D, the digit two, tetris. org, and you will find there are links to the Coursera courses. You will find an email that if you send us email and just say that you want to teach a course, I will give you links to to all the materials.

You will find links to our IDE, to all our programs. Projects, everything is, is, is in the website.

Jonathan: Very cool.

Shimon: So that's the go to, that's the go to place.

Jonathan: Yeah, that's, that's neat, that's neat. And then people that are taking the course, is there sort of a a hangout place? Is there like a Discord server or anything for, for students going through it?

Shimon: Not really. I mean, we we didn't want to Sort of, you know, we didn't want to put together something that we cannot support. Right. So so you want to teach the course? I'd say you can build your own space and do whatever you want with it. Up to you.

Jonathan: Yeah, I would, I would love to see maybe a little bit more official, easy to use repository one of these days for, for people doing these add on projects.

Like, look, I added inheritance to Jack. Here's my code. You know, I, I wrote this video game engine. Here's, here's my code, you know, put a GitHub or GitLab server up somewhere. I think that would be fine.

Shimon: I agree. I guess. I guess we have to do it. It's too, too lazy, but we'll do it at some point. I understand

Jonathan: that part.

Yeah, so do I. I mean, you guys surely don't have anything else keeping you busy, right? Exactly. Oh, okay. That's cool. Is there anything coming down the pike and beyond our source code server that we just established that we're going to create? Is there anything, is there anything you guys are excited that's coming up that you want to plug?

Shimon: Yeah. Well you know, I'm so glad that we. We managed to talk for about almost an hour without mentioning the two letters A I. But one thing that I'm working on now is pouring all this stuff, you know, all the course materials, the book and everything, into a larger sort of language model. And and providing a chatbot in which you can actually, in which you can actually interact.

with so called me, you know, with an avatar that represents me, and you can ask any questions that you want about Nand to Tetris, projects, teaching, lectures, whatever, and, and hopefully get some sensible answers, so Hopefully, yeah, so I'm working, you know, I'm not going to release it until until it will be good looking and sensible, but that's something that I'm working on.

And other than that plug in, I don't know, you know, I don't know. I think I'm quite happy with what we did so far, and I don't want, I don't want to sort of describe all sorts of half baked, half baked ideas. So, I'll let you know when I'm ready for another interview. Okay,

Jonathan: we can, we can do that. Yeah, for sure.

Alright, good. Okay, so I've got to ask, we are, we are required by our audience to ask you two final questions before we let you go, and that is, what is your favorite text editor and scripting language?

Shimon: Favorite text editor is Tico. If you, you know, sort of an Emacs like editor, and scripting language?

Yes. Perl. Ah.

Jonathan: Yeah, pretty cool. Okay. We don't get very many Perl answers, but somewhere Randall Schwartz is, is applauding and giving you, you know, a virtual high five. I started

Rob: on Perl, so.

Jonathan: Pearl, yeah, Pearl's one of my first languages too. I moved away from it pretty quick, but I enjoyed it back in the day.

Every once in a while I'll pull it out and do just a little bit in it. Okay. Alright, Shimon, thank you, thank you so much for being here. It was an absolute delight to get to chat with you and talk about the project. And like I said, I'm, I'm gonna have to carve out some time to get back to it and go through it again.

I just enjoy it so much.

Shimon: All right. Thank you for having me. Bye

Jonathan: bye. Bye. All right, Rob, what do you think? You gonna, you gonna teach the course?

Rob: Well,

Jonathan: I don't know if I'm going to set

Rob: up, I don't know if I'll set up a community class or anything like that, but definitely heard about it. It gets me excited to try it out.

I don't know if with all my commitments, if I'll be able to, but I may at least start with buying the book and, you know, baby steps.

Jonathan: Yeah, yep.

Rob: And then get into it, you know, take a, take a little bit into it. But that seems really awesome.

Jonathan: I, I really want to get the the recommended FPGA and screen and set up and do the do the actual hardware version of it.

To be able to have the thing running now in an FPGA, but still be able to run the thing for real. That's just, that, that tickles every geek bone in my body. I love it so much.

Rob: Maybe we know someone who can give us some chips.

Jonathan: Yeah, there you go. I don't think that company makes FPGAs, but.

Rob: Well, I don't know.

Guessing they're all NAN chips or something. Yeah,

Jonathan: yeah, I'm sure he can give us lots of NAN. All right, anything you want to plug, Rob?

Rob: You know, if you want to check out more of me, come find me. My website is robertpcampbell. com. There's links to various social medias there. Otherwise you can come join me and Jonathan over at the Untitled Lennox Show on the Twit Network every Saturday at 5 30 central time.

Jonathan: Because central time is all that matters, right?

Rob: Centralized for a reason.

Jonathan: Alright, thank you for being here man, I appreciate it. Alright, if you want to find more of me hiding behind that command line window where I've been moving virtual machines around, back there is Hackaday. You can find my stuff on Hackaday, that's where Floss Weekly is.

We appreciate Hackaday being the home of Floss Weekly these days. It's also where my security column goes live every Friday morning. Friday morning, central time. The only time that matters. And then if you want more of Rob and I, there is the Untitled Linux Show over on the Twit Network. And other than that, we just appreciate everybody being here.

We always have a lot of fun on Floss Weekly. And we will be back next week for more. And we don't have a guest scheduled yet. So if you know of somebody Let us know, either come into the discord and tell us, tell us you want to be on the show or I believe it's floss at hackaday. com. You can send an email there.

We will get it. It'll come to me and we'll make it happen. Appreciate it. Everyone that watches that listens both live and on a download, and we will see you next time.