WEBVTT

00:00:00.000 --> 00:00:06.120

Hey, YouTube, everyone out there. Thank you so much for showing up for the live show or

00:00:06.120 --> 00:00:09.440

watching the recording afterwards. If you're here live, be sure to throw some comments

00:00:09.440 --> 00:00:15.600

and thoughts into the live stream and we'll try to make it part of the show. All right,

00:00:15.600 --> 00:00:22.280

guys ready to kick this off? Yep. Let's do it. Andre and Aaron, welcome to talk Python

00:00:22.280 --> 00:00:28.880

to me. Thanks for having us. Yeah. It's great. Yeah, it's fantastic to have you both here.

00:00:28.880 --> 00:00:32.920

What a cool topic, symbolic math with Python.

00:00:32.920 --> 00:00:36.000

I think I'd heard of SymPy before,

00:00:36.000 --> 00:00:39.160

but I didn't fully appreciate how neat

00:00:39.160 --> 00:00:41.960

and how advanced it really was.

00:00:41.960 --> 00:00:44.740

So people are gonna enjoy diving into this

00:00:44.740 --> 00:00:47.640

and we can talk some internals and whatnot.

00:00:47.640 --> 00:00:51.520

And if you're doing anything scientific or computational,

00:00:51.520 --> 00:00:54.040

this is certainly a project worth checking out.

00:00:54.040 --> 00:00:56.960

But before we get to that, let's just hear your story.

00:00:56.960 --> 00:00:58.880

How do you get into programming and Python?

00:00:58.880 --> 00:01:01.120

Aaron, you want to go first?

00:01:01.120 --> 00:01:01.620

Sure.

00:01:01.620 --> 00:01:05.640

So I mean, as far as just programming,

00:01:05.640 --> 00:01:13.400

I've always been into computers ever since I was a little kid.

00:01:13.400 --> 00:01:18.120

Actually started messing around with programming

00:01:18.120 --> 00:01:23.360

and AppleScript and my family's Macintosh

00:01:23.360 --> 00:01:26.600

back when I was a really little kid.

00:01:26.600 --> 00:01:28.960

but didn't really do any serious programming

00:01:28.960 --> 00:01:32.120

till college when I started doing,

00:01:32.120 --> 00:01:34.680

taking like computer science courses.

00:01:34.680 --> 00:01:37.560

And actually for Python,

00:01:37.560 --> 00:01:40.720

my college had a free course that they were offering

00:01:40.720 --> 00:01:45.120

to teach this little language called Python.

00:01:45.120 --> 00:01:48.400

So, you know, the CS course that I was taking

00:01:48.400 --> 00:01:52.200

was in like Java and C and those languages

00:01:52.200 --> 00:01:54.880

or I wasn't a huge fan of those languages,

00:01:54.880 --> 00:01:57.180

but I learned about this language called Python

00:01:57.180 --> 00:01:58.880

and yeah, I thought it was great.

00:01:58.880 --> 00:02:05.100

And so, yeah, I was immediately hooked with Python

00:02:05.100 --> 00:02:08.620

just because it's so easy to use.

00:02:08.620 --> 00:02:11.500

It's so easy to just write a program

00:02:11.500 --> 00:02:14.620

and you don't have to worry about compilers

00:02:14.620 --> 00:02:16.700

or all this boilerplate.

00:02:16.700 --> 00:02:19.740

And then--

00:02:19.740 --> 00:02:21.880

- I'd always been a little suspicious of languages

00:02:21.880 --> 00:02:24.100

that described themselves that way,

00:02:24.100 --> 00:02:27.220

which is ironic given how much I do with Python.

00:02:27.220 --> 00:02:29.140

But so often when you hear that,

00:02:29.140 --> 00:02:32.080

it's like, "Oh, it doesn't need all the symbols and it's super easy to get started."

00:02:32.080 --> 00:02:33.500

That means like, "Oh,

00:02:33.500 --> 00:02:37.540

it's really easy to do easy stuff but then you can't do interesting things."

00:02:37.540 --> 00:02:40.980

I think one of the things that's special about Python is like you can keep going.

00:02:40.980 --> 00:02:44.980

You don't have to stop once you need advanced ideas. That's pretty awesome.

00:02:44.980 --> 00:02:47.340

>> Yeah, it really is. You're right.

00:02:47.340 --> 00:02:49.700

A lot of languages stop,

00:02:49.700 --> 00:02:54.340

But Python, I mean, it's got a lot of advanced stuff as well.

00:02:54.340 --> 00:02:58.340

And it's got a huge ecosystem of libraries you can do.

00:02:58.340 --> 00:03:00.940

Just pretty much anything you want in it.

00:03:00.940 --> 00:03:02.460

- Yeah, for sure.

00:03:02.460 --> 00:03:03.940

The big paradox for me is,

00:03:03.940 --> 00:03:08.020

there's a joke T-shirt that I've seen that says,

00:03:08.020 --> 00:03:09.980

I learned Python, it was a great weekend.

00:03:09.980 --> 00:03:12.020

Which I think is kind of true, you can do that.

00:03:12.020 --> 00:03:14.180

But at the same time, for many, many years now,

00:03:14.180 --> 00:03:15.100

I've been studying Python,

00:03:15.100 --> 00:03:18.020

and there's still just so much more to learn.

00:03:18.020 --> 00:03:20.600

One about the language and also just like, as you said,

00:03:20.600 --> 00:03:22.240

all the different libraries like,

00:03:22.240 --> 00:03:24.000

oh, I wanna learn Pydantic now

00:03:24.000 --> 00:03:25.600

or this other thing or so on.

00:03:25.600 --> 00:03:27.560

Very cool.

00:03:27.560 --> 00:03:28.380

All right, how about now?

00:03:28.380 --> 00:03:29.680

What are you working on day to day?

00:03:29.680 --> 00:03:33.880

- Yeah, so I work at a company called Quansight.

00:03:33.880 --> 00:03:39.240

We do consulting mostly around open source,

00:03:39.240 --> 00:03:41.840

open source Python, open source data science stack.

00:03:41.840 --> 00:03:47.120

I work about 50% of my time on SymPy

00:03:47.120 --> 00:03:50.320

as part of a grant, which we might talk about later,

00:03:50.320 --> 00:03:51.880

I think.

00:03:51.880 --> 00:03:56.520

And I also work on different consulting projects.

00:03:56.520 --> 00:03:59.800

Actually, right now I'm part of a project

00:03:59.800 --> 00:04:04.000

called the Data APIs Consortium,

00:04:04.000 --> 00:04:09.000

which is trying to standardize the array API libraries

00:04:09.000 --> 00:04:16.840

for different Python array libraries, the APIs for those.

00:04:16.960 --> 00:04:19.560

like NumPy and X-Ray and those types of things?

00:04:19.560 --> 00:04:23.920

- NumPy, PyTorch, Jax, CuPy.

00:04:23.920 --> 00:04:26.280

So the idea is that all these libraries

00:04:26.280 --> 00:04:29.080

have a very slightly different APIs

00:04:29.080 --> 00:04:32.720

and it makes it hard to write code against that goes,

00:04:32.720 --> 00:04:34.440

that just works against all of them.

00:04:34.440 --> 00:04:37.880

So what we're doing is creating a standard API

00:04:37.880 --> 00:04:40.400

that all of these libraries will target.

00:04:40.400 --> 00:04:43.440

And so you can write code that will just,

00:04:43.440 --> 00:04:45.080

will work with NumPy,

00:04:45.080 --> 00:04:49.240

but you can also just, instead of importing NumPy, you can import Coupy,

00:04:49.240 --> 00:04:52.120

and that same code will run on the GPU, for example.

00:04:52.120 --> 00:04:56.120

Oh, that's fantastic. Yeah, that's a really good, ambitious goal.

00:04:56.120 --> 00:05:00.360

It sounds so easy, I'm sure it's pretty challenging.

00:05:00.360 --> 00:05:04.680

Yeah, well, I think we're being successful. We just released the first version of the

00:05:04.680 --> 00:05:11.560

specification, and we've got several libraries that are implementing it already,

00:05:11.560 --> 00:05:17.880

including NumPy, Coupy, several others. So it's actually, I think it's being a successful project

00:05:17.880 --> 00:05:23.720

and should really help push the Python data science ecosystem forward.

00:05:23.720 --> 00:05:27.400

Yeah, it sounds like definitely a good step. Andrej, how about you? How'd you get into

00:05:27.400 --> 00:05:32.120

programming in Python? You know, probably in high school. I was trying to figure out when I

00:05:32.120 --> 00:05:38.840

got into it. I think in high school, I don't know. I don't remember how I found it, but it was early

00:05:38.840 --> 00:05:45.320

2000s I'm guessing, it was before NumPy, I think there was NumArray, I think, it was

00:05:45.320 --> 00:05:52.040

the array library in Python. And during my undergrad, I was studying physics, I wanted

00:05:52.040 --> 00:05:58.520

to play with mathematical formulas in Python, so that's when I started SymPy. And then I

00:05:58.520 --> 00:06:03.240

used Python, you know, pretty much probably close to 20 years now, or over 20 years.

00:06:05.240 --> 00:06:07.240

And I cannot say I'm an expert.

00:06:07.240 --> 00:06:09.400

You know, it wasn't a good weekend.

00:06:09.400 --> 00:06:11.960

I don't know.

00:06:11.960 --> 00:06:17.400

A lot of stuff I'm not that great at, you know, but the basics I would say I know pretty well.

00:06:17.400 --> 00:06:17.900

Yeah, sure.

00:06:17.900 --> 00:06:24.040

I, after my PhD in physics, I went to work at Los Alamos National Lab for about eight years

00:06:24.040 --> 00:06:25.880

as a computational physicist.

00:06:25.880 --> 00:06:27.320

And I use Python.

00:06:27.320 --> 00:06:30.520

I use a lot of Fortran, a lot of computational codes and a lot of C++.

00:06:30.520 --> 00:06:31.720

That sounds fun.

00:06:31.720 --> 00:06:34.760

There's a lot of Python stuff happening there at Los Alamos.

00:06:34.760 --> 00:06:39.720

Yeah, it is. The NetworkX, I think, library was started here.

00:06:39.720 --> 00:06:48.680

And then, very recently, half a year ago, I changed jobs. I now work at a company

00:06:48.680 --> 00:06:55.400

called GSI Technology, which is a hardware vendor as a compiler developer. And so at Los

00:06:55.400 --> 00:07:01.880

Alamos, I started this compiler for Fortran called LFortran. It's a compiler for Fortran,

00:07:01.880 --> 00:07:06.440

but it also allows you to use it interactively just like you would Python. It runs in a Jupyter

00:07:06.440 --> 00:07:14.280

notebook as well as a kernel. And at this new company, we are also, in addition to LFortran,

00:07:14.280 --> 00:07:19.880

creating a new frontend called LPython, which takes Python, but compiles it through the same

00:07:19.880 --> 00:07:25.160

pipelines, through the same intermediate representation and all the backends, all

00:07:25.160 --> 00:07:31.640

the code is the same. So effectively, it treats Python like Fortran. So the same speed and so

00:07:31.640 --> 00:07:32.920

and so forth.

00:07:32.920 --> 00:07:33.760

- Oh, wow.

00:07:33.760 --> 00:07:34.600

That sounds awesome.

00:07:34.600 --> 00:07:36.280

- It is awesome, yes.

00:07:36.280 --> 00:07:37.980

- Very, very cool.

00:07:37.980 --> 00:07:42.640

All right, well, super fun to be talking about SymPy

00:07:42.640 --> 00:07:45.320

with you all today, symbolic math.

00:07:45.320 --> 00:07:47.660

I wanna start this off though,

00:07:47.660 --> 00:07:50.160

with a write-up from Aaron

00:07:50.160 --> 00:07:52.800

that is maybe not what people would expect

00:07:52.800 --> 00:07:55.640

as the first topic to introduce SymPy.

00:07:55.640 --> 00:07:57.160

SymPy has been in the news a little bit

00:07:57.160 --> 00:08:00.080

for the wrong reasons recently, right?

00:08:00.080 --> 00:08:00.920

Right, Aaron?

00:08:00.920 --> 00:08:05.920

So it was even on the front page of Hacker News

00:08:05.920 --> 00:08:10.960

because this company, HackerRank, reached out to GitHub,

00:08:10.960 --> 00:08:16.200

sounds like through automated ways,

00:08:16.200 --> 00:08:19.800

and said, "We seem to see some math

00:08:19.800 --> 00:08:21.820

"that looks like questions we might ask,"

00:08:21.820 --> 00:08:22.960

or something like that,

00:08:22.960 --> 00:08:25.060

"so we need to completely ban Senpai

00:08:25.060 --> 00:08:28.380

"because they're stealing our interview questions,"

00:08:28.380 --> 00:08:30.080

or something like that, right?

00:08:30.080 --> 00:08:30.920

Give us the story.

00:08:30.920 --> 00:08:33.000

What kind of madness was this?

00:08:33.000 --> 00:08:36.720

>> Yeah, so we sort of just got this notice from GitHub

00:08:36.720 --> 00:08:39.640

that a DMCA takedown had been issued

00:08:39.640 --> 00:08:43.520

against one of the pages on our documentation.

00:08:43.520 --> 00:08:47.400

And the way the DMCA works

00:08:47.400 --> 00:08:49.520

and the way GitHub's DMCA policy works

00:08:49.520 --> 00:08:51.760

is when somebody issues a notice,

00:08:51.760 --> 00:08:54.700

they basically just have to take it down.

00:08:54.700 --> 00:08:58.600

And GitHub basically took down

00:08:58.600 --> 00:09:04.480

the entire documentation site for about 12 hours.

00:09:04.480 --> 00:09:08.760

And so I've got a timeline of everything that happened here

00:09:08.760 --> 00:09:11.080

in this blog post.

00:09:11.080 --> 00:09:18.000

But basically, yeah, they had some company

00:09:18.000 --> 00:09:22.520

that they were working with to try to find their solutions

00:09:22.520 --> 00:09:26.280

on GitHub and issued DMCA takedown notices against them.

00:09:26.280 --> 00:09:30.720

And somehow this worth IT solutions company

00:09:30.720 --> 00:09:33.760

decided that our docs was one of these,

00:09:33.760 --> 00:09:36.440

and they just issued the notice.

00:09:36.440 --> 00:09:42.200

And so it ended up taking down our documentation, which

00:09:42.200 --> 00:09:44.320

immediately, like you said, it made it

00:09:44.320 --> 00:09:47.640

to the top of Hacker News and--

00:09:47.640 --> 00:09:49.320

>> It's too outrageous.

00:09:49.320 --> 00:09:55.620

>> Yeah, so the notice itself, it was completely ridiculous.

00:09:55.620 --> 00:10:00.620

Like there's like the stuff that they were claiming was

00:10:00.620 --> 00:10:05.820

their copyright is like probably not even copyrightable.

00:10:05.820 --> 00:10:07.980

Like if you look at the examples on

00:10:07.980 --> 00:10:10.820

on that documentation page that they took down

00:10:10.820 --> 00:10:12.740

they're just like simple math examples.

00:10:12.740 --> 00:10:16.900

- Yeah, it was like a X squared minus two solution is,

00:10:16.900 --> 00:10:19.820

you know, negative plus or minus radical two, right?

00:10:19.820 --> 00:10:22.300

Like that you shouldn't be able to copyright.

00:10:22.300 --> 00:10:26.360

I think that actually predates HackerRank Foundation.

00:10:26.360 --> 00:10:30.220

- I think it is HackerRank as a company.

00:10:30.220 --> 00:10:35.940

- Yeah, so the CEO of HackerRank,

00:10:35.940 --> 00:10:38.900

I guess noticed the buzz that was being made about it

00:10:38.900 --> 00:10:42.180

and retracted the notice.

00:10:42.180 --> 00:10:47.100

So the docs ended up going back online about 12 hours later.

00:10:47.100 --> 00:10:48.260

If they hadn't done that,

00:10:48.260 --> 00:10:51.820

we would have had to issue a counter notice,

00:10:51.820 --> 00:10:56.440

which would have kept the docs offline for about two weeks

00:10:56.440 --> 00:10:58.780

before that counter notice would have taken effect.

00:10:58.780 --> 00:11:01.700

And that's just because of the way the DMCA law works.

00:11:01.700 --> 00:11:06.300

I also explained how the DMCA law works in this blog post.

00:11:06.300 --> 00:11:09.920

If anyone's wondering like why would GitHub even do this?

00:11:09.920 --> 00:11:12.100

Like, well, they're kind of required

00:11:12.100 --> 00:11:13.980

to do this kind of stuff by law.

00:11:13.980 --> 00:11:15.820

There are some ways I think they could have improved

00:11:15.820 --> 00:11:18.460

what they did here, which I go over as well.

00:11:18.460 --> 00:11:19.300

But this DMCA law is-

00:11:19.300 --> 00:11:21.460

- It's probably like pointing those out.

00:11:21.460 --> 00:11:24.180

- Yeah, so I mean, the DMCA provides

00:11:24.180 --> 00:11:25.980

what's called safe harbor for places

00:11:25.980 --> 00:11:29.460

that host user generated content, basically.

00:11:29.460 --> 00:11:31.940

YouTube, the social media places,

00:11:31.940 --> 00:11:34.780

GitHub clearly has a ton of user generated content

00:11:34.780 --> 00:11:35.600

that's public.

00:11:35.600 --> 00:11:40.540

If you got the source code of Windows or something

00:11:40.540 --> 00:11:42.460

and you posted it up there, right?

00:11:42.460 --> 00:11:43.820

Theoretically, Microsoft might say,

00:11:43.820 --> 00:11:45.300

you know, we don't really want you hosting this.

00:11:45.300 --> 00:11:46.740

It's not supposed to be open source.

00:11:46.740 --> 00:11:48.100

Please take it down, right?

00:11:48.100 --> 00:11:50.940

So that's the idea.

00:11:50.940 --> 00:12:03.940

The way it works though is basically once there's a request to take it down, they just have to, here's the part that I think that got missed, decide that it's a valid request and then just take it down.

00:12:03.940 --> 00:12:12.460

Like they don't really try to negotiate or determine the right, the correctness of that assertion, right?

00:12:12.780 --> 00:12:21.380

Yeah, so they sort of have to stay like as a disinterested third party, I guess.

00:12:21.380 --> 00:12:28.060

And the reason is that in order for them to have the safe harbor, the safe harbor status sort of

00:12:28.060 --> 00:12:35.420

frees them from the liability on either side, if there is a copyright claim or if there's a lawsuit or something,

00:12:35.420 --> 00:12:38.860

they can just say, OK, well, we're just hosting the content.

00:12:39.140 --> 00:12:41.780

The law sort of allows them to do that.

00:12:41.780 --> 00:12:44.420

But in order to do that, they sort of have to just,

00:12:44.420 --> 00:12:47.020

if they get a DMCA claim, they have to just take it down.

00:12:47.020 --> 00:12:49.260

And then if there's a counter notice issued,

00:12:49.260 --> 00:12:50.780

then they can put it back up.

00:12:50.780 --> 00:12:52.220

And the DMCA has these,

00:12:52.220 --> 00:12:57.620

the DMCA sort of tells them that if they get a claim,

00:12:57.620 --> 00:13:00.380

they have to sort of take it down expeditiously.

00:13:00.380 --> 00:13:01.700

And if there's a counter notice,

00:13:01.700 --> 00:13:05.260

they have to put it back up in 10 to 14 days.

00:13:05.260 --> 00:13:08.940

So it's, there's a lot of parts of the law

00:13:08.940 --> 00:13:10.440

that they sort of have to do.

00:13:10.440 --> 00:13:13.940

There's some things I think that they could have done better.

00:13:13.940 --> 00:13:15.980

And there's also some things that they already do,

00:13:15.980 --> 00:13:18.060

which they don't have to do, which are nice,

00:13:18.060 --> 00:13:21.600

including they have a repo where they actually post

00:13:21.600 --> 00:13:24.660

every single DMCA notice that they've received,

00:13:24.660 --> 00:13:27.300

going all the way back to 2011, actually.

00:13:27.300 --> 00:13:28.140

- Interesting, that's good.

00:13:28.140 --> 00:13:31.040

- So you can actually see all the DMCA notices

00:13:31.040 --> 00:13:32.980

that have been issued against GitHub repositories

00:13:32.980 --> 00:13:37.980

on that repo, it's github.com/github/dmca.

00:13:38.600 --> 00:13:40.300

- Yeah, okay.

00:13:40.300 --> 00:13:42.880

You know, I wanna preface this by saying I'm not a lawyer,

00:13:42.880 --> 00:13:46.280

so I'm only speculating here,

00:13:46.280 --> 00:13:48.280

but one of the things that seems very,

00:13:48.280 --> 00:13:51.640

very much like it fell through the cracks,

00:13:51.640 --> 00:13:55.600

and this part I think is both maliciousness

00:13:55.600 --> 00:13:58.800

and laziness on the Worth IT company,

00:13:58.800 --> 00:14:02.420

as well as, you know, sort of either laziness

00:14:02.420 --> 00:14:06.760

or disinterest, I guess, from GitHub.

00:14:06.760 --> 00:14:08.440

I don't really know how to characterize it,

00:14:08.440 --> 00:14:13.400

But this company came and said, on your documentations,