00:00:00.001 --> 00:00:06.040

It's time to look deep within the machine and understand what really happens when your Python code executes.

00:00:06.040 --> 00:00:11.720

We're code walking through the CPython code base and visualizing it at pythontutor.com.

00:00:11.720 --> 00:00:17.660

This is episode number 22 with Philip Guau, recorded Monday, August 3rd, 2015.

00:00:17.660 --> 00:00:47.540

Welcome to Talk Python To Me, a weekly podcast on Python, the language, the libraries, the

00:00:47.540 --> 00:00:49.160

ecosystem, and the personalities.

00:00:49.160 --> 00:00:51.180

This is your host, Michael Kennedy.

00:00:51.180 --> 00:00:56.200

Follow me on Twitter where I'm @mkennedy, and keep up with the show and listen to past

00:00:56.200 --> 00:00:58.720

episodes at talkpython.fm.

00:00:58.720 --> 00:01:02.160

Be sure to follow the show on Twitter where it's at talkpython.

00:01:02.160 --> 00:01:06.480

This episode is brought to you by Hired and CodeChip.

00:01:06.480 --> 00:01:11.740

Thank them both for supporting the show on Twitter via at HiredHQ and at CodeChip.

00:01:11.740 --> 00:01:14.160

Now let me introduce Philip.

00:01:14.540 --> 00:01:19.360

Philip Guau is an assistant professor of computer science at the University of Rochester in New York.

00:01:19.360 --> 00:01:24.880

He researches human-computer interactions with a focus on user interfaces for online learning.

00:01:24.880 --> 00:01:29.780

He's especially interested in studying how to better train software engineers and data scientists.

00:01:30.000 --> 00:01:35.120

He created a free web-based visualization tool for learning programming called Online Python Tutor

00:01:35.120 --> 00:01:45.020

at pythontutor.com, which has been used by over 1.2 million people in over 165 countries to visualize over 11 million pieces of code.

00:01:45.020 --> 00:01:46.760

Philip, welcome to the show.

00:01:46.760 --> 00:01:47.620

My pleasure.

00:01:48.300 --> 00:01:50.160

Yeah, it's really exciting to have you here.

00:01:50.160 --> 00:01:53.020

We're going to talk a lot about many things.

00:01:53.020 --> 00:02:02.820

We're going to talk about CPython and a really cool project that you put on your website and on YouTube called CPython, a 10-hour code walk.

00:02:02.820 --> 00:02:04.860

And so we'll be digging into CPython.

00:02:04.860 --> 00:02:11.900

And we're also going to talk about this thing called Python Tutor at pythontutor.com that you are working to help people understand the internals of Python better.

00:02:12.400 --> 00:02:14.000

So that's going to be great stuff.

00:02:14.000 --> 00:02:14.680

Cool.

00:02:14.680 --> 00:02:16.140

I'm looking forward to it.

00:02:16.140 --> 00:02:16.520

Yeah.

00:02:16.520 --> 00:02:22.660

Before we get into the details, though, you know, everyone likes to know how people got into programming and how they got started in Python.

00:02:22.660 --> 00:02:23.460

What's your story?

00:02:23.460 --> 00:02:29.860

So my story was I was always interested in computers as a kid, like many people who got into computer science.

00:02:29.860 --> 00:02:34.720

But I never really had a strong programming background until I went to college.

00:02:34.720 --> 00:02:38.520

So I tried to learn QBasic by myself when I was 10.

00:02:39.080 --> 00:02:43.220

And that, you know, I had a book and then I failed after a few weeks because I had no one teaching me.

00:02:43.220 --> 00:02:46.080

I took an AP computer science course in high school.

00:02:46.080 --> 00:02:46.980

That was in C++.

00:02:46.980 --> 00:02:48.660

And that was really fun.

00:02:48.660 --> 00:02:52.440

And that was kind of my first introduction to really doing programming.

00:02:52.440 --> 00:02:57.580

And in college, I decided to major in electrical engineering and computer science.

00:02:57.760 --> 00:03:01.320

And that's when I started just learning programming formally.

00:03:01.320 --> 00:03:09.640

But really, the Python relevance is I didn't actually start hacking for fun until my about my senior year of college.

00:03:09.640 --> 00:03:15.980

And the first language that I learned for programming for fun and not just because I had to do it for class was actually Python.

00:03:15.980 --> 00:03:28.380

So the first kinds of programs I wrote were scripts to manage my photos and, you know, kind of manipulate and manage my own personal photo gallery and, you know, put it up on a simple website.

00:03:28.380 --> 00:03:30.660

So that was where I got started getting hooked on Python.

00:03:30.660 --> 00:03:32.460

That was, you know, it was about 10 years ago.

00:03:32.460 --> 00:03:33.680

That was around 2005.

00:03:33.680 --> 00:03:36.040

That was like Python 2.4 or something like that.

00:03:36.440 --> 00:03:38.040

Yeah, that's a great way to get started.

00:03:38.040 --> 00:03:44.680

I think a lot of people have interesting stories like that, you know, just they have some small problem they're trying to solve.

00:03:44.680 --> 00:03:47.700

And, you know, it leads you down this path.

00:03:47.700 --> 00:03:52.240

And all of a sudden, you discover this world where, hey, there's this great thing, you know, programming or Python or whatever.

00:03:52.240 --> 00:03:54.160

Yep, that's exactly right.

00:03:54.160 --> 00:03:57.520

So I see you're calling in from Seattle, right?

00:03:57.520 --> 00:03:58.340

What are you doing up there?

00:03:58.340 --> 00:04:05.340

So I am currently an assistant professor of computer science at the University of Rochester in upstate New York.

00:04:05.340 --> 00:04:06.560

So that's nowhere near Seattle.

00:04:06.560 --> 00:04:07.200

That's what I was going to say.

00:04:07.200 --> 00:04:08.880

You're not – it's not at all in Seattle.

00:04:08.880 --> 00:04:19.860

So I get to – one of the real benefits of being a professor is that your summers are free to do research or to travel or to do other sorts of scholarly work.

00:04:19.860 --> 00:04:27.080

So most professors in most terms, they stay on campus in the summers and they do research full time for three months.

00:04:27.300 --> 00:04:41.460

What I decided to do this summer since I had some colleagues at Microsoft was to spend most of my summer at Microsoft Research doing research and both in software engineering and in online education at the lab in Seattle.

00:04:41.460 --> 00:04:47.020

And I came here because I actually was an intern here a long time ago when I was back in grad school.

00:04:47.020 --> 00:04:49.320

So I'm actually back interning in the same group.

00:04:49.320 --> 00:04:51.200

So it's sort of a homecoming of sorts.

00:04:51.680 --> 00:04:52.620

Back to the future.

00:04:52.620 --> 00:04:53.200

That's excellent.

00:04:53.200 --> 00:04:55.560

Yeah, I've done some work with some of the guys up at Microsoft.

00:04:55.560 --> 00:04:57.020

It's a cool place up there.

00:04:57.020 --> 00:04:57.720

So excellent.

00:04:57.720 --> 00:05:00.080

Is this related to PythonTutor.com?

00:05:00.080 --> 00:05:01.820

No, not really.

00:05:01.820 --> 00:05:04.620

I mean, this is just a completely separate sort of research project.

00:05:04.620 --> 00:05:08.520

So there's nothing Python related in the work here, unfortunately.

00:05:08.520 --> 00:05:09.880

All right.

00:05:09.880 --> 00:05:10.160

Cool.

00:05:10.160 --> 00:05:11.620

All right.

00:05:11.640 --> 00:05:16.020

So let's talk about your CPython internals class.

00:05:16.020 --> 00:05:19.580

This was a class you did at University of Rochester, right?

00:05:19.580 --> 00:05:21.160

2014, I think.

00:05:21.160 --> 00:05:23.220

At least the recorded version was 2014.

00:05:23.220 --> 00:05:24.600

Yep.

00:05:24.600 --> 00:05:27.480

So this was a class I taught in fall 2014.

00:05:27.480 --> 00:05:32.120

And the name of the course was Dynamic Languages and Software Development.

00:05:32.120 --> 00:05:40.500

So I actually inherited this course from another professor who was taking a leave and teaching another class that term.

00:05:41.000 --> 00:05:43.080

And that class was originally in Ruby.

00:05:43.080 --> 00:05:48.980

So it was sort of a graduate level programming languages class about these sorts of dynamically type languages.

00:05:48.980 --> 00:05:50.840

And originally he did it in Ruby.

00:05:50.840 --> 00:05:59.600

But since I knew Python a lot better, I revamped the class to be in Python and basically turned it into what the videos are online.

00:05:59.600 --> 00:06:01.500

So I'd be happy to talk about that in detail.

00:06:02.280 --> 00:06:05.340

Just for everyone listening, the videos are online.

00:06:05.340 --> 00:06:08.260

And I actually spent like the last week going through your class.

00:06:08.260 --> 00:06:13.660

So I feel like I've had like some super intense summer course or something, you know, doing like 10 lectures.

00:06:14.440 --> 00:06:21.600

And people can find those on your website at pgbovine.net slash cpython dash internals dot htm.

00:06:21.600 --> 00:06:30.280

And I actually went through, unrelated to this conversation or maybe preceding this whole having you on the show, I just saw your videos and thought they were awesome.

00:06:30.280 --> 00:06:34.120

And I put them into a YouTube playlist at bit.ly slash cpython walk.

00:06:34.260 --> 00:06:35.580

So both of those work well.

00:06:35.580 --> 00:06:38.080

What was the main goal of the class?

00:06:38.080 --> 00:06:43.960

Sort of get people to understand what happens when you actually run dynamic code like Python?

00:06:43.960 --> 00:06:48.800

Yeah, I think that was basically that was basically a philosophy.

00:06:48.800 --> 00:06:54.200

So a lot of programming languages classes are taught from more of a theoretical perspective.

00:06:54.200 --> 00:06:54.640

Right.

00:06:54.640 --> 00:07:01.100

So it's usually kind of some formal syntax and semantics and maybe doing some proofs.

00:07:01.100 --> 00:07:04.300

And it's very, you know, kind of a formalism heavy.

00:07:04.300 --> 00:07:15.980

And I thought it would be interesting to do a very different sort of class for graduate students from the opposite side, which is something extremely applied to saying, you know, here is a here is a piece of Python code.

00:07:15.980 --> 00:07:20.460

Let's start with hello world or a simple for loop or a simple function call.

00:07:20.460 --> 00:07:29.200

And what actually happens throughout all the steps between that code being parsed and then the output appearing on your screen, let's say.

00:07:29.480 --> 00:07:39.160

So I wanted to dive into the interpreter and show students how everything worked under the hood and how there's really, you know, by deconstructing, you can show that there's really no magic here.

00:07:39.160 --> 00:07:43.160

There's just a lot of C code behind the scenes that keeps track of a lot of stuff.

00:07:43.160 --> 00:07:45.720

And eventually your program runs.

00:07:45.820 --> 00:07:50.960

So we don't do the parsing stage because I think parsing is fairly standard.

00:07:50.960 --> 00:07:54.860

And that's covered by most kind of introductory compilers classes.

00:07:54.860 --> 00:08:00.720

You write a grammar and parser generator and some code gives you a like an AST.

00:08:00.720 --> 00:08:03.940

And then that gets walked to turn into some kind of bytecode.

00:08:04.020 --> 00:08:07.800

So the class actually starts with assuming you have a bunch of Python bytecode.

00:08:07.800 --> 00:08:16.240

How does the bytecode actually get interpreted step by step by the by the interpreter runtime system to do your your programs operations?

00:08:16.240 --> 00:08:17.640

Yeah, that's really cool.

00:08:17.640 --> 00:08:32.920

And I think, you know, if I think about how like C code runs and then my intuition about how that C code actually executes, if you understand a little bit about registers and memory addresses and pointers, your intuition more or less will carry the day.

00:08:32.920 --> 00:08:36.940

I think with interpreted languages, all bets are off.

00:08:36.940 --> 00:08:37.180

Right.

00:08:37.240 --> 00:08:44.820

I mean, you have some concept of the programming language doing things, but then the way that happens, you really have to look inside.

00:08:44.820 --> 00:08:45.100

Right.

00:08:45.100 --> 00:08:47.060

Yeah, exactly.

00:08:47.060 --> 00:08:53.380

Because the these interpret languages are often not implemented like you would conceptually think of it.

00:08:53.380 --> 00:08:53.520

Right.

00:08:53.520 --> 00:08:57.760

You think of something as you have frames and variables and pointers to each other.

00:08:57.760 --> 00:09:04.680

But really, these bytecodes are this sort of the Python one is sort of this stack based kind of virtual machine.

00:09:04.820 --> 00:09:08.480

I think Java, the Java virtual machine is that way, too, but I forgot the exact semantics.

00:09:08.480 --> 00:09:11.200

But it's not something you would think about normally, but they do it that way.

00:09:11.200 --> 00:09:18.940

One, because it's really compact and it's kind of leads to really compact code and sort of easy to understand code for the implementer.

00:09:18.940 --> 00:09:24.480

But yeah, but that's very different than the conceptual model in your head at the very high level, how a program ought to work.

00:09:24.480 --> 00:09:30.060

And we can talk about that later when we talk about the Python tutor as well, because that kind of leads into that other tool.

00:09:30.060 --> 00:09:32.420

So we can keep talking about the CPython stuff first.

00:09:32.420 --> 00:09:43.380

Sure. So one of the things I thought was interesting was in your very first session, you did you have kind of a cool whiteboarding thing you're doing with a Microsoft Surface and like a pen where you can kind of draw on the code.

00:09:43.380 --> 00:09:44.080

And that's cool.

00:09:44.420 --> 00:09:50.900

You do a cool little sketch about what actually happens when you type Python space some file that py.

00:09:50.900 --> 00:09:54.700

And I mean, on one level, I knew it.

00:09:54.700 --> 00:09:56.480

On the other, it was a little surprising to me to say.

00:09:56.480 --> 00:09:57.740

And the first step is compilation.

00:09:58.500 --> 00:10:05.740

Can you maybe like talk just briefly about like what happens when I run my Python code before we get into the interpreter itself?

00:10:05.740 --> 00:10:06.960

Yeah.

00:10:06.960 --> 00:10:20.600

So many people are surprised when there's a compilation step in Python or in these sorts of dynamic or what people call scripting languages, because usually you think of running Python space, whatever, or Perl space, whatever, Ruby space, whatever.

00:10:20.760 --> 00:10:21.840

And it just runs, right?

00:10:21.840 --> 00:10:22.100

Right.

00:10:22.100 --> 00:10:24.660

I just thought, here we go with the interpreter.

00:10:24.660 --> 00:10:26.100

And now it's interpreting, right?

00:10:26.100 --> 00:10:27.420

Right.

00:10:27.420 --> 00:10:33.140

So with Java or C or C#, you have a compilation step and then you run a compiled binary.

00:10:33.140 --> 00:10:34.580

And there's two separate steps.

00:10:34.580 --> 00:10:40.680

But with Python, as with many other languages, the compilation happens before the execution.

00:10:40.680 --> 00:10:46.960

So what happens is as a standard kind of a front end to a compiler, it takes the source code.

00:10:46.960 --> 00:10:48.560

It does the lexical analysis.

00:10:48.560 --> 00:10:49.640

It does the parsing.

00:10:49.640 --> 00:10:52.700

It creates a AST or abstract syntax tree from that.

00:10:52.700 --> 00:10:56.400

And then it walks that tree and creates a bunch of bytecode.

00:10:56.400 --> 00:11:16.160

So the Python bytecode language, you can read in the documentation, it has, I don't know, a few dozen operations like add, load, store, and also some operations that are a little bit more Python specific, like build a list, build a dictionary, build a tuple, function call, those sorts of things.

00:11:16.160 --> 00:11:26.260

So the compilation step really takes your source code, which is in human readable, somewhat human readable form, and turns it into a linear stream of instructions.

00:11:26.260 --> 00:11:34.660

Very much like assembly language, except you can think of bytecode as an assembly language for a Python virtual computer.

00:11:35.440 --> 00:11:46.480

Right. That was kind of the impression I got as well, like a much, much richer assembly language where you have operations like build class and call method, push and pop stuff off stacks and so on.

00:11:46.480 --> 00:11:47.780

Yep, exactly.

00:11:47.780 --> 00:11:56.020

If we want to go work with this, right, we can go to python.org and download the code and decompress it or untar it or whatever.

00:11:56.020 --> 00:11:59.540

And it's just, it literally is a bunch of C code, right?

00:11:59.540 --> 00:12:03.640

The C in CPython is, here's your C implementation of this interpreter, right?

00:12:03.640 --> 00:12:04.540

That's right.

00:12:04.640 --> 00:12:07.280

So if you go to, this is what I do on the first day of class.

00:12:07.280 --> 00:12:19.640

We have everybody download the C interpreter source code from, sorry, the CPython source code from python.org and unzip it and do configure and make.

00:12:19.640 --> 00:12:26.440

Now, part of the class, I didn't require students to actually run the interpreter if they didn't want,

00:12:26.440 --> 00:12:29.800

because most of the class was actually reading through the code and walking through it.

00:12:29.800 --> 00:12:34.580

Now, the students who were a bit more adventurous, they could try to compile the interpreter themselves.

00:12:34.580 --> 00:12:40.500

And then try to, you know, put in debug statements or print statements to see how it works behind the scenes.

00:12:40.500 --> 00:12:49.400

But actually, compile interpreter itself might not be easy if you're on, say, especially, say, on a Windows machine, which doesn't have a lot of development tools, compilers.

00:12:49.400 --> 00:12:51.520

So I'm usually on Linux and Mac machines.

00:12:51.520 --> 00:12:58.860

If you install the standard developer tool chain with GCC and make and configure and all that stuff.

00:12:58.860 --> 00:13:01.760

In theory, right, building is always hard.

00:13:01.760 --> 00:13:10.120

But in theory, if you do dot slash configure and then and you type make all your your you'll actually call the C compiler on your machine.

00:13:10.120 --> 00:13:18.040

And it will compile all the C files and the C and the H files in the CPython slash directory.

00:13:18.040 --> 00:13:22.500

And in the end, it will produce a binary executable file called Python.

00:13:22.500 --> 00:13:24.720

And that Python you can just run.

00:13:24.720 --> 00:13:29.220

And that is the Python interpreter that you just compiled from C source code.

00:13:29.340 --> 00:13:34.980

So most of the class, what we do is we go over what a lot of those C files actually do and see.

00:13:34.980 --> 00:13:44.140

Maybe you could give us like a 10,000 foot view of what are the interesting parts of that source code and what is just noise and details.

00:13:44.140 --> 00:13:49.360

So there's like objects and then there's include there's see eval dot C.

00:13:49.360 --> 00:13:52.980

There's there's like a few really common parts that you come back to over and over and over.

00:13:53.080 --> 00:13:54.320

And then there's a bunch of details.

00:13:54.320 --> 00:13:55.580

Yeah.

00:13:55.580 --> 00:14:00.500

So on the Web site with all the videos, I actually show the files that they reference.

00:14:00.500 --> 00:14:08.520

But really, the core file that I keep on going back to what you're saying is in Python slash C eval dot C.

00:14:08.520 --> 00:14:12.840

And what that is that that file at its core is the main interpreter loop.

00:14:12.840 --> 00:14:19.720

So conceptually, how you execute how Python executes code is a byte code is just a bunch of them.

00:14:20.700 --> 00:14:22.580

It's just a list of instructions.

00:14:22.580 --> 00:14:28.120

Each one is add or subtract or build list or function call or so forth.

00:14:28.120 --> 00:14:40.960

And all the interpreter does is just go through one instruction at a time, take it off the list of instructions, do something and then move to the next instruction, do something, move the instruction and then do something else.

00:14:41.580 --> 00:14:46.440

And it might jump around the stream of instructions if you have, say, a function call or a loop.

00:14:46.440 --> 00:14:54.560

But really, the main interpreter loop in C eval dot C, all it does is it's just a big, wild, true, infinite loop that just.

00:14:54.560 --> 00:14:56.700

Yeah, there's like a huge switch statement.

00:14:56.700 --> 00:14:57.700

And it is huge, right?

00:14:57.700 --> 00:14:58.200

That's right.

00:14:58.200 --> 00:15:02.200

Yeah, there's like a 3000 or whatever line switch statement.

00:15:02.320 --> 00:15:03.860

There's a fun fact in there.

00:15:03.860 --> 00:15:15.840

If you actually I don't know if it's in all the versions, but at least in some of the versions I saw, there's some kind of comment in there saying that they needed to like break up the switch statement in some weird way.

00:15:15.840 --> 00:15:21.040

Because some C compilers just can't take switch statements that are that big.

00:15:21.040 --> 00:15:28.760

So they had to actually break up the code into pieces because, you know, it wouldn't compile on some kind of computers because that code was just too giant.

00:15:28.760 --> 00:15:30.100

Yeah, that's pretty funny.

00:15:30.100 --> 00:15:31.680

It's like a 3000 line switch statement.

00:15:31.680 --> 00:15:32.380

It's pretty cool.

00:15:32.380 --> 00:15:37.280

But those are more or less the steps that have all the opcodes.

00:15:37.800 --> 00:15:50.340

And so if I look at Python, it's not necessarily mapping one to one the Python code I write to these opcodes, which is a good thing for Python programmers, right?

00:15:50.340 --> 00:15:52.000

That means you're working in a high level language.

00:15:52.000 --> 00:15:54.220

You're not working like down in the detail, right?

00:15:54.220 --> 00:16:01.700

But it also means it's hard for me to understand if I write, you know, create a class and I say, you know, T equals new test class.

00:16:01.700 --> 00:16:03.420

What does that actually mean?

00:16:03.420 --> 00:16:04.620

Like, how do I line that up?

00:16:04.620 --> 00:16:07.480

And so you had a cool way to disassemble that, right?

00:16:07.720 --> 00:16:08.280

And look at it.

00:16:37.640 --> 00:16:42.960

Currently, candidates receive five or more offers in just the first week and there are no obligations ever.

00:16:42.960 --> 00:16:45.060

Sounds pretty awesome, doesn't it?

00:16:45.060 --> 00:16:47.100

Well, did I mention there's a signing bonus?

00:16:47.100 --> 00:16:51.200

Everyone who accepts a job from Hired gets a $2,000 signing bonus.

00:16:51.200 --> 00:16:55.540

And as Talk Python listeners, it gets way sweeter.

00:16:55.540 --> 00:17:03.100

Use the link Hired.com slash Talk Python To Me and Hired will double the signing bonus to $4,000.

00:17:04.100 --> 00:17:04.820

Opportunity's knocking.

00:17:04.820 --> 00:17:08.440

Visit Hired.com slash Talk Python To Me and answer the call.

00:17:18.560 --> 00:17:19.620

Right, right.

00:17:19.620 --> 00:17:25.740

So the disassembler actually comes in the standard Python library.

00:17:25.740 --> 00:17:40.640

So if you do, right, so if you do Python space dash M space DIS, which runs the disk module space, the Python file name, name of Python file, I'll actually run the main function in the DIS module.

00:17:41.360 --> 00:17:46.900

And what that will do is I'll actually print out a somewhat human readable representation of the bytecode.

00:17:46.900 --> 00:17:55.640

And the cool thing about that is that it shows the line number of which line of your Python source code compiles into which bytecode.

00:17:55.740 --> 00:17:57.880

And as you mentioned, it's not a one-to-one mapping.

00:17:57.880 --> 00:18:03.220

So one line usually compiles to several bytecodes because the bytecode is at a lower level.

00:18:03.220 --> 00:18:05.420

So you can run that DIS command.

00:18:05.420 --> 00:18:16.060

And the DIS module, you can just search for, if you search on your favorite search engine for Python space DIS, you should see the documentation for this disassembler module.

00:18:16.060 --> 00:18:20.720

And that is in the standard library, and that gives you all of the stuff.

00:18:20.720 --> 00:18:24.860

So now that said, though, that only prints out the instructions.

00:18:24.860 --> 00:18:31.480

There was somebody who made a library called byteplay, which is B-Y-T-E-P-L-A-Y.

00:18:31.480 --> 00:18:39.700

And that library actually is an enhanced version of the disassembler that lets you get the disassembled bytecode into objects.

00:18:39.700 --> 00:18:41.440

You can actually play with it yourself.

00:18:41.440 --> 00:18:42.980

You can manipulate it.

00:18:42.980 --> 00:18:44.840

You can, you know, take it apart.

00:18:44.840 --> 00:18:45.800

You can analyze it.

00:18:45.800 --> 00:18:51.520

So this byteplay library, I haven't used it myself personally, but I know people who really like playing with it.

00:18:51.520 --> 00:18:52.840

Yeah, that's cool.

00:18:52.840 --> 00:18:53.660

A little more powerful.

00:18:53.660 --> 00:19:02.120

One thing about the DIS module is it's super easy to look at just sort of flat code in Python files.

00:19:02.120 --> 00:19:07.640

But if I want to look at the functions or I've got nested functions and classes, it's a little more work to do that, right?

00:19:07.640 --> 00:19:09.000

Yeah.

00:19:09.000 --> 00:19:15.540

So the default with the DIS module is it just disassembles the top level of your program.

00:19:15.540 --> 00:19:20.580

So all the top level says is that if you define a function, it'll just say function definition.

00:19:20.800 --> 00:19:27.960

And then what you have to do is you actually have to go inside that function and disassemble that function itself.

00:19:27.960 --> 00:19:29.680

So it is a little bit more hairy.

00:19:29.680 --> 00:19:34.060

And I don't know if byteplay handles all that out of the box, but it might.

00:19:34.060 --> 00:19:39.760

But the idea is that the DIS module, if you just run it by default, it will just disassemble the top level program.

00:19:39.760 --> 00:19:42.580

And any functions will not be disassembled automatically.

00:19:42.580 --> 00:19:47.460

You have to actually grab the code of those functions and go in there and call dis on that.

00:19:47.460 --> 00:19:50.400

So it is a little bit more tricky to do that.

00:19:50.400 --> 00:19:51.260

Sure.

00:19:51.260 --> 00:19:59.640

The other thing I thought was interesting is if I've got a function, let's say foo, in Python, I could say, what is it?

00:19:59.720 --> 00:20:03.400

Foo.func underscore bytecode.

00:20:03.400 --> 00:20:06.480

How do I – the bytecode is actually there on the function.

00:20:06.480 --> 00:20:12.220

And you can look at it in its encoded form, which is kind of some binary string type thing.

00:20:12.980 --> 00:20:15.000

And then you can also disassemble that as well, right?

00:20:15.000 --> 00:20:16.320

That's right.

00:20:16.320 --> 00:20:18.640

And that's what I think we're just leading into that.

00:20:18.640 --> 00:20:26.280

So the idea is that DIS itself, if you just run it, it disassembles the bytecode of the, I guess, of the top level file.

00:20:26.280 --> 00:20:29.180

But each function itself has its own code.

00:20:29.180 --> 00:20:34.200

And like you said, I think it's – it's actually different in Python 2 and 3, the name of it.

00:20:34.200 --> 00:20:39.500

But I think in one version it's like the function object dot func underscore code.

00:20:39.500 --> 00:20:43.100

The other one is just like just dot code or something like that.

00:20:43.100 --> 00:20:50.340

But the idea is that the code of the function just appears inside of it as a binary string of data.

00:20:50.340 --> 00:20:53.960

So if you actually print it out, it just looks like some garbled string.

00:20:53.960 --> 00:20:58.740

But if you run it through some – you can run it through some pretty printing function or through DIS.

00:20:58.740 --> 00:21:01.740

And it actually shows you the bytecode of the function.

00:21:01.740 --> 00:21:13.620

Because all a function object is that it's some context plus an actual string of bytecode that represents what the instructions are that the function is supposed to execute when you run it.

00:21:14.280 --> 00:21:22.740

Yeah, the other thing I thought was pretty cool is – or interesting to understand is that sort of compile step that you talk about, right?

00:21:22.740 --> 00:21:30.200

When I run Python My Python file, I get first like a compile step to bytecode and then the dynamic interpreted execution.

00:21:30.200 --> 00:21:34.220

But all those functions and stuff, that bytecode is there and ready to roll.

00:21:34.220 --> 00:21:38.700

It's just not kind of wired together until it gets to the interpreter, right?

00:21:39.700 --> 00:21:40.240

That's right.

00:21:40.240 --> 00:21:48.200

So you can actually compile – I think it's just the Python interpreter does the compiling and running all at the same time.

00:21:48.200 --> 00:21:57.320

But I think there's actually a mode in Python that you can just compile to – you can just compile the bytecode and not actually run it yet.

00:21:57.320 --> 00:21:59.980

I'm not sure exactly which flags are that one.

00:21:59.980 --> 00:22:06.400

But sometimes people actually ship pre-compiled Python bytecode instead of the source code.

00:22:06.820 --> 00:22:12.500

So there's – I don't know what reason people do this because you can just run the source code.