0:00

Hello and welcome to Python bytes where we

0:02

deliver python news and headlines directly to your

0:04

earbuds. This is episode 385 recorded

0:07

may 27th 2024

0:10

they get there right? Yeah, I am Brian

0:12

Auchin and I am Michael Kennedy This week's

0:14

episode is sponsored by mail trap more about

0:16

them later. If you want to connect with

0:18

us, we're all on Fostin

0:21

on the show and Michael and I

0:23

were fasted onions Pasta donians.

0:25

Yeah So those links are on

0:27

the show notes Also in

0:29

the show notes or at python bytes.fm

0:32

You can go to live if you want

0:34

to join us join the audience. Usually it's

0:36

Tuesdays at 10 a.m. Pacific

0:38

time You can also

0:41

use that link to get older videos too. If

0:43

you want to watch what watch what happens live

0:46

then finally if you'd like

0:48

to Receive an email

0:50

after a couple a day or two after

0:52

the show to find out all of the

0:54

links and And to have a

0:56

link to the show But make sure if you listen to this

0:58

and you want to make sure that you don't miss

1:01

any links and be able to go Check

1:03

those out later. You can become a friend

1:05

of our show and and we'll send

1:07

you those emails So you had announcement before we

1:09

get started? I do well I just wanted to

1:11

say Brian the reason that we're not just doing

1:13

this like a normal time We're just kind of

1:15

late and I got some messages from folks in

1:17

places like New Zealand like awesome This is a

1:19

perfect perfect time. So there's always a perfect time

1:21

somewhere in the world, but you've got this thing

1:23

going on tomorrow And

1:25

then for the rest of the week, I'm

1:27

going to the sink called the giant loop ride

1:30

This is off-road rally that's at

1:32

a hot springs Maybe

1:34

300 other off-road riders out in the southeast

1:36

desert of Oregon and riding the mountains and

1:38

the desert and open bar Live

1:41

music all sorts of shenanigans going on. Anyway, somebody

1:43

oh, yeah that that swimming pool is a hot

1:46

springs By the way, how crazy is that if

1:48

anyone is listening out there and it's also going

1:50

just come say hi That's all I'm saying but

1:52

also why we're doing this in the middle of

1:54

the night for some

1:56

places So I would

1:59

say If you have a podcast that kind of runs

2:02

late, sometimes it'll make you tired, wouldn't you think? You have

2:04

to rest a little? Yeah.

2:07

Let's talk about PostGrest, as in

2:09

PostGrest REST API. So this one

2:11

comes to us from Mark Little.

2:13

So thanks, Mark, for sending this

2:16

over. It's a good one. This

2:18

is basically a web

2:20

server, web front end for your database.

2:23

So you've got PostGrest, and you might

2:25

think, all right, well, I would just

2:27

like to expose the tables as endpoints

2:30

with ability to do

2:32

queries over HTTP endpoints. You can somehow

2:34

pass the query over. You want

2:38

to have security, things like that. So this is an

2:40

app that you can run, and it

2:42

automatically, with zero API work for

2:44

you, turns your PostGrest instance

2:47

into a somewhat

2:49

restricted API, basically.

2:51

Cool, right? Yeah. Now,

2:54

obviously, these things, I don't know, I always have mixed feelings

2:56

when I talk about these things. On

2:59

one hand, if you just say, I would like,

3:01

say, my view front end or something, or even

3:03

my Python code to just have access to the

3:05

database as an API, just read the crud things

3:07

of all the tables, you're good to go. This

3:11

is what you want, right? On the

3:13

other hand, sometimes I want to have more specific

3:15

things. Instead of just, here's the user table, I

3:17

want you to go through the create new user

3:19

API, a new user

3:22

API endpoint that uses, I don't

3:24

know, argon2 for memory hard, password

3:27

hashing and other checks. Every

3:30

time I come up against these things, I'm like, oh, they're really

3:32

cool, but at the same time, I

3:35

kind of want a little more specific control over

3:37

how it works. So if

3:39

you're just like, I just have a database, I just

3:41

want to be able to expose it real quick, let's

3:43

go, PostGrest, sort of thing. Cool, right? So

3:46

the way it works is it's written in

3:48

Haskell, which you may, and it's 4% Python,

3:50

so I guess there's a little Python in

3:52

there, but it doesn't really matter what it's

3:54

written in. It's just a thing you run,

3:56

and Then you talk

3:58

to it from your Python. From your javascript

4:00

or whatever it is. Okay, so there's a

4:03

a couple of things you care about. One

4:05

you probably care about performance. Ah, I'm not

4:07

super familiar the performance of Haskell, but I

4:10

believe it runs on top the Jv I'm

4:12

sorry if I get around by think so

4:14

which makes it pretty quick to zillow do

4:16

two thousand requests per second on the hook

4:19

who free tier which is pretty fast actually

4:21

I that's a pretty wimpy server and two

4:23

thousand requests? the second is pretty high end.

4:26

Three reasons for that: Haskell and as Britain

4:28

used in the Warp http server. A

4:30

Compiling compiled languages lightweight threads. I also

4:32

to him out warp so and Learn

4:34

but also to think it delegates as

4:36

much as it can to the database.

4:38

so it's like a really thin layer

4:41

over the database as I was describing

4:43

Socialize and J spawns and a son

4:45

response directly and sequel. Or as does

4:47

the data validation, the authorizations combining real

4:49

count and retrieve off on the to

4:51

post in a single. To me all

4:53

those different things are happening in the

4:56

database. not in this thing right Fps.

4:58

Finally it I used as the. Has

5:00

your instead of it's Sql sequel

5:02

is that ass Sql library which

5:04

means it's database connections are buying

5:07

a protocol for talking to it

5:09

and as stateless for horse on

5:11

skilling. Also very important for security

5:13

you don't want to just go

5:15

and the whole thing is read

5:18

right on. The internet is just

5:20

kill of little bobby tables and

5:22

find it still uses I Jason

5:24

web tokens and basically what you

5:26

do you create a user in

5:29

the database. And then he said

5:31

authentication like this user can only read

5:33

and they can only read these tables

5:35

are they come right at that table?

5:37

I don't know and then that's what

5:39

you can do when you authenticate over

5:41

this right? So basically leverages database users

5:43

in their room. implicit permissions to control

5:45

what you can do with a pie

5:48

is waste of version that a pie

5:50

which can enter seen as the database

5:52

changes you could keep it stable it

5:54

seems like and a document itself to

5:56

open a p I was x pretty

5:58

neat. so if you have a post

6:00

database you like that we really just

6:02

need rewrite of data and maybe even

6:04

an internal app and you just want

6:07

to have it resembled. check this out

6:09

and co co I didn't know that

6:11

turkey ahead of free to A anymore

6:13

but it is no less. So sick

6:15

Alice your back you know last the

6:17

maybe and from previous times a be

6:19

as a sad statement it must be

6:21

from previous times cause I have this

6:24

was updated a month ago but it

6:26

didn't I think that's are just whatever

6:28

that wnd free to was. that's. Giving

6:30

you a sense of what you

6:32

can do because cooking meat. I'm

6:34

gonna talk about a sink iowa

6:36

little bit but so there's a

6:39

desert go called the From take

6:41

a video on how Python a

6:43

Cinco Rick how Python a Cinco

6:45

works recreating from scratch. I'm in

6:47

a to a killer like this

6:49

this sir this idea because where

6:51

else love is ago but sometimes

6:54

it just feels magical into sort

6:56

of walk through some of the

6:58

some how it might might. Be

7:00

working in a in the background

7:02

is chemical so it goes through

7:04

ah, generate generators or view facts

7:06

about the amount lube Tux was

7:08

sleeping than yielding to await and

7:10

await. They think I'll end it.

7:12

Just kind of like the format

7:14

that this is written into. the

7:16

sort of get your get a

7:18

good mental model of how they

7:20

Cinco works. So this is some

7:22

and I guess if you're highly

7:24

recommend. This article was written looks

7:26

really fun, the colors and nice

7:28

to hate is Oregon's. Sucker for

7:30

good web pages he's read but

7:32

dumb but this is good. an

7:35

egg they skimmed it briefly were

7:37

already in like is chemicals. through

7:39

our we're facing works is exists

7:41

or spheres had to magical sometimes

7:43

but it is pretty magical and

7:45

at the heart of it as

7:47

generators shade so enraged themselves feel

7:49

super magical and on you can

7:52

understand how those where I can

7:54

assemble okay whoa that's hate and

7:56

a partition. the work in the

7:58

new just running in. It's

8:01

yes fragment of the generator Nazi go but

8:03

yeah it seems pretty wild is so can

8:05

walk through a little steps and then his

8:07

shows what the to has I felt the

8:10

final bit and then it rewrites the whole

8:12

thing in actually think I have so the

8:14

weekend see what it looks like and such

8:16

great along with that when I noticed the

8:18

links that some there was some camera go

8:21

from there was a comment also about if

8:23

you want to like learned a sicko from

8:25

trying to recreate or from scratch. There was

8:27

also a i do have easily talk from.

8:30

The From: I think it's from Fica.

8:32

Yep, I can India from Toy nineteen

8:34

I hadn't watched this. I still haven't

8:36

watched the Spurs my list. Now I'm

8:38

build your own A Sink. From

8:42

do Beasley awesome instructor services has

8:44

could videos was only will lead

8:46

to that I'm so some some

8:48

learning is Cinco are some good

8:50

stuff indeed What what else is

8:52

good is our sponsor Maelstrom All

8:54

right us Talk about of this

8:56

have said is sponsored by Mail

8:59

trap an email. Delivery platform that

9:01

developers love, an email sending solution

9:03

with industry best in Linux as

9:05

Ncp and Email A Pie as

9:07

well as as the case for

9:10

major programming languages and Twenty Four

9:12

Seven Human Support. Try for free

9:14

at the mail trap.i'll make you

9:16

think you mail traps. Let's move

9:19

on us and little higher order

9:21

salad Ryan Sir Perfect. Follow on

9:23

from what you are talking about.

9:25

A son Go parallel programming. It's

9:27

basically scales the waiting now. this

9:30

this is gonna be a dated statement

9:32

i believe but at the moment python

9:34

is not super good at doing computational

9:37

parallelism right but you know what is

9:39

gp use as the jeep use are

9:41

awesome at doing pillow programming so your

9:43

typical python program has one thread or

9:45

seen on t the most thing which

9:48

they have one thread he skelter seabees

9:50

maybe that sixteen my feeling were gaming

9:52

pc as that my him to macbook

9:54

pro a thing as ten thread You

9:57

know, there's way more it's get a

9:59

lot more work done if you scale

10:01

those out, right? But if you scale

10:03

their GPUs, they have sixteen thousand three

10:06

hundred eighty four threads things like that,

10:08

right? Incredible. Yeah, I

10:10

ran across this thing called

10:12

Bend be endy and

10:14

it comes from higher order company Which

10:17

has a couple of it's like a runtime for

10:19

Bend and then has been the language And

10:22

it's all about taking things and running

10:24

them in parallel and making them Basically

10:26

be Python so it says with bend

10:28

you can write parallel code for multi-core

10:30

CPUs and GPUs Without being a C

10:33

or CUDA respectively expert what ten years

10:35

experience and it feels just like Python

10:37

So they give us this example. Here's

10:40

a function called some not the best

10:42

example because it shadows a built-in but

10:44

whatever Let's get roll with it,

10:46

right? And so it

10:48

has def sum ticket depth and there's some

10:51

value that it starts with and then adds

10:53

them all up, right? So so sum all

10:55

the numbers from zero till two to the

10:58

end and to the depth right as a

11:00

switch statement that once it gets to

11:02

the end stops and then Sort

11:05

of recurses down as it would

11:07

and there's nothing that looks parallel about

11:09

this whatsoever But what

11:11

happens is this code actually runs

11:13

in parallel potentially on those thousands

11:15

of GPU cores Oh,

11:17

wow wild, right? Yeah, so

11:20

the way it does that is There's

11:23

a picture here Is

11:25

it breaks these things up into what are

11:27

called? Computational graphs

11:30

instead of work and then it just it

11:32

goes in processes them to see if it

11:34

if that part of the graph can be

11:36

parallelized Which is kind of

11:38

nuts? So you don't even

11:41

write the parallel code. It just looks at

11:43

your code You go so you that loop

11:45

that we just saw or that recursion type

11:47

of thing We just saw that actually could

11:49

be done recursively by just changing the parameters

11:51

or something, right? So pretty

11:53

wild and so yeah, they have

11:55

this HBM highly massively parallel runtime

11:58

that achieves near ideal speed as

12:00

a function across the available course. And then

12:03

Ben is the programming language that runs in

12:05

parallel powered by this thing. Looks

12:07

like Python. I don't

12:09

know how compatible, say, it's standard

12:11

library is or things like that. And

12:13

then you have this thing called kind, which

12:16

is a parallel proof checker. I have no idea about that. It

12:19

just says it's coming soon. So this is available on GitHub. And

12:22

you come over and you can find the language. And

12:25

guess what? It has 15,000 stars. I'd never even heard of it because

12:28

it's pretty popular. Written in Rust, not

12:30

that that's super relevant, but it's like

12:32

powers a language that looks like Python.

12:34

That's pretty cool. I'm cautiously

12:37

optimistic about that. It could

12:40

be good. So it looks

12:42

like the real implementation for it is here,

12:46

although it looks pretty short. I don't know.

12:48

I've dated it a couple days ago. But

12:50

here it says, I don't know. We'll have

12:52

to see how large of a

12:54

thing is it. I guess that's just some

12:56

pretty big scroll bars. So maybe the whole

12:58

run time is here. I mean, the number

13:00

of files might be lower, but could have tons

13:02

of stuff in there. Yeah, that's true. Anyway,

13:06

is it going to be a thing? I'm not sure. Maybe,

13:08

maybe not. But it's cool to see people

13:10

being creative with this kind of stuff. Yeah,

13:12

actually, it sounds pretty cool. I

13:17

guess simplify the way you get your code ready

13:19

for massive

13:21

parallel running. If it can just look

13:23

at your code and go, oh yeah, I got it. Yeah,

13:26

if it can look and say this is

13:28

what's often referred to as embarrassingly parallel, there's

13:30

like no state shared. You could

13:32

just take the pieces and break them up and run them on cores. Yeah.

13:36

If it just finds those automatically and does

13:38

that, which is what it claims to do.

13:40

Yeah, that's great. That's really great. Yeah, that's

13:42

really great. All right. Next

13:45

up, I've got a regular expressions.

13:47

So we're getting all sorts of

13:49

deep topics today. We are. We

13:51

sure are. I

13:54

ran across, I was looking at Reddit. And

13:57

so this Book is an older

13:59

book. The euro think it's year old

14:01

I'm scarred. the smartest way to learn

14:04

Python regular expressions and they don't know

14:06

really pets, other people's books or anything

14:08

on the show but I thought it

14:10

was kinda nice that it's a lean

14:13

pub thinks his. third they're doing their

14:15

own thing and his for get into

14:17

i think this can the but the

14:19

minimum price is free so is in

14:21

since they're allowing their own preserve a

14:24

to to throw few bucks array of

14:26

thing is they can because when had

14:28

it's good to have people. Teaching things

14:30

you I ran across on

14:33

Reddit. And or

14:35

the topic sound nice others to go

14:37

through an induction, a regular expressions they

14:39

do an application, a taco puzzle, learning

14:42

and stuff and then I skimmed the

14:44

rust belt. Looks like great stuff and

14:46

as a is some Python related thing

14:48

to teach regular expressions as group would

14:51

be good. And then I went watch

14:53

the video and I'm even more intrigued.

14:55

So the idea on this book is

14:58

that modern ways to learn or difference

15:00

in possibly. People my age might be

15:02

used to have just reading a book

15:04

and trying things out, but this is

15:07

this is set up visible video introduction

15:09

to the book. They're using the book

15:11

itself as a as a long form

15:13

teaching people but it's also. There's a

15:16

video is a video course attached to

15:18

his of you. but if Lake Erie

15:20

gets a book in your intrudes, there's

15:22

links to the different course chapters within

15:24

the book which is Kevin the idea

15:27

and then ah there's there's puzzle solvers

15:29

for each. each senior learning single

15:31

apply it in a puzzle environment

15:33

and they're using a thing called

15:35

as sin sinks tour at says

15:38

so awful name honestly i think

15:40

but if i nxt are but

15:42

dub but it looks like there's

15:44

a whole bunch other people using

15:46

a to and i've never tried

15:48

it but it's has some an

15:50

online trying as from code platform

15:52

i think so i'm i'm i'm

15:55

a i'm gonna try this out

15:57

omg i'm curious about the the

15:59

whole idea of reading a book

16:01

and clicking on it and doing that

16:03

as well. Yeah. That's, you

16:05

know, mileage very, very for me because I'm probably

16:07

going to download this onto a Kindle and I

16:09

can't collect on anything with the Kindle,

16:11

but at least the one I'm using. But

16:14

anyway, might be fun to go through. So. Yeah.

16:18

Looks excellent. Excellent. Give us a

16:20

report when you get back. Sure. Yeah. See

16:23

how it went. Yeah. And, you know, not

16:25

enough regular expression material out there for just

16:27

deep, diving deep and really getting to getting

16:29

your head around it. So yeah, they're very

16:31

powerful if you're good at them. Yeah.

16:35

Big if. Those

16:37

are items. Do we have any extras?

16:39

I have one that's quite interesting and

16:41

I don't, somebody gave me

16:44

a bit of a tip towards this. I

16:46

don't remember who it was. So I think it was

16:48

a massadon, but thank you. And

16:51

sorry about forgetting who the credit

16:53

here on Wikipedia. We have the

16:55

Python conference, aka PyCon. And

16:58

what does it say about it? The Python

17:00

conference, also called PyCon, is the largest annual

17:02

convention for the discussion and promotion of the

17:04

Python language. Scroll down a little bit, Ryan.

17:07

See things like where was it in 2003?

17:10

It was in DC. Had 200 attendees. 2012

17:13

was in Santa Clara. Oh,

17:16

2016. Yeah, baby. Portland. Almost.

17:21

Almost. The record. Almost. Almost.

17:26

Almost. Almost. Two

17:31

more people beat that number, came

17:34

over that number from Cleveland, Ohio

17:36

in the before days. And

17:38

then COVID hit, knocked

17:40

it down and slowly building back. Like in

17:43

Pittsburgh, it was up to 2,500 again. So

17:47

what was that? It's still got

17:49

about 50% more to add or a

17:51

third looking backwards. It's down.

17:54

That's not what I want to talk about. I Want to

17:57

show you what is in this Wikipedia table right below

17:59

it. The next the Long

18:01

Beach be going to California. Let's go

18:03

going to the Beats gonna some warm

18:05

weather, go into a vacation spot. I

18:08

will bet you that this significantly beats

18:10

ah those numbers and because it's just

18:12

a destination Ip wanted to go there

18:14

to be there. a lump on lieutenant

18:17

I sit in the west coast. I

18:19

mean nothing is Pittsburgh but it's only

18:21

two hours away from such from or

18:24

was it cleaver Cleveland via an south

18:26

we did of Salt Lake City in

18:28

their the In Terms mission. Times with

18:30

Salt Lake City was still under a

18:33

lot of funk like was the first

18:35

one that came back and person after

18:37

covert so it it was kind of

18:39

funky. I went to the Twenty Twenty

18:41

Three one. I know that in the

18:44

i'm a a a Summer's gone the

18:46

record as saying it thinks it like

18:48

there's a lot more cool destination things

18:50

doesn't they Some places on the coasts

18:52

yeah and Midwest obviously but but think

18:55

doing it can the middle of the

18:57

country make sense of the people can

18:59

get. easy flights from anywhere in

19:01

the country to san by us and

19:03

a what are you optimizers for are

19:06

you i try to move it around

19:08

are you trying to make it as

19:10

centrally located a tiny make it as

19:12

cheap as possible he tried to make

19:14

it a destination all these different things

19:16

play into this and this is the

19:18

biggest funding for the ps f so

19:21

it's it's not just like a curiosity

19:23

right it it is the funding for

19:25

the pss to a significant degree is

19:27

interesting though the portland was way back

19:29

and twenty seventeen and this in twenty

19:31

twenty six as can be than next

19:33

time that it's gonna be on the

19:36

west coast i don't think that i

19:38

know that everybody in the east coast

19:40

thanks to salt lake is in the

19:42

west boots not really since nine the

19:44

same time zone brain sometimes smelled i

19:46

am pacific time let's go against anyone

19:48

is my access argot planned tentatively pipe

19:51

that python after pittsburgh's so next yourself

19:53

as wearing it's and round to but

19:55

then long beach california and i'm i'm

19:57

here for a brain also over always

19:59

me I'm curious about the numbers

20:01

of how many people showed up in

20:03

Montreal. It's cool

20:06

to see you can just look it up on Wikipedia. Yeah, it's

20:08

all right here. I've got

20:10

something completely random that I... Prince

20:16

of Persia was a video game from 89. I remember it. Oh,

20:20

not only was it a video game, it was

20:23

one of the best video games. It was so

20:25

good. Yeah, it was one of these cool racers

20:27

through all sorts of stuff. We

20:30

had a little runner thing and had to jump

20:32

over all sorts of stuff. It was a fun

20:34

game. And then there was a bunch

20:36

of different permutations of it.

20:39

And it's still going strong, I think. Still

20:42

stuff going on. Anyway,

20:44

a cool game, but I

20:47

ran across this. There's a book about it. So

20:50

Jordan Meckner is a person

20:52

that wrote it and originally

20:55

wrote the Prince of Persia game.

20:58

And he wrote a book called

21:01

Replay. It's a memoir of an uprooted

21:03

family. And apparently,

21:06

it goes back to 1938

21:08

when somebody in his

21:11

family, his parents

21:13

fled Austria to escape

21:15

the Nazi persecution. And

21:18

then they traveled to the US. Eventually,

21:20

they went to France. And then they

21:23

went to eventually... I think. Actually,

21:25

I'm just making this up. I don't know if

21:27

he ever made it to the US. I

21:29

think so. But then

21:32

he goes back eventually. Yeah,

21:34

because he goes back eventually

21:36

and works in France for

21:39

a little while. And

21:41

then this book about his life, his

21:43

family, and his life story, and about

21:45

the Prince of Persia. So I'm kind

21:47

of excited to read this. I

21:50

really like these looks inside the storytelling

21:52

of the history of some of these

21:55

famous games and famous platforms. Yeah,

21:58

they're very interesting. Yeah. The

22:00

day I wrote the like wrote the

22:02

music for for this of the original

22:04

video. Okay, is Chemical on a Runner

22:06

as sick as a new one of

22:09

them and of the older computers than

22:11

it is? so that's it as my

22:13

extra saw some. Let's close it out.

22:15

What? they. Were. Can't

22:18

joke at in a way cause

22:20

brain to We have this woman

22:22

developer said in hair pretty happy

22:24

and this boss comes in managers

22:26

as I don't like this bash

22:29

script you created. Brian Python instead

22:31

commands the boss imports a process

22:33

so process call worked out sh

22:35

well that's a day's work done.

22:39

A face at a half ads as

22:41

if you want me to write my

22:43

phone call is going from Python Moscow

22:45

Yeah good. Wouldn't have one of the

22:47

time sir. Mike Wary Early on earth

22:49

is not earlier. they were twenty

22:51

years ago so I'm I'm a

22:53

spy myself for a new job

22:55

and they've got all these builds

22:57

make script servers for C plus

22:59

plus codes and a camera. There

23:01

were there in their batch files

23:03

and in fashion all sorts of

23:05

stuff and in his had guinea

23:07

worm reread this cause is kind

23:09

of a mess so I rewrote

23:11

it in a combination of i

23:13

had have combination of bash pearl

23:15

python said whole bunch of things.

23:17

Whatever it wanted to to to

23:19

what I saw was an elegant

23:22

solution and the team was horrified

23:24

near like we're not learning that

23:26

me like images of my okay

23:28

so i rewrote role in python

23:31

and that was of for semi

23:33

I really grabbed Python for do

23:35

whatever I want as a shell

23:37

scripts language to go yeah limits

23:40

to suppress com said process that

23:42

cause shape and nice a don't

23:44

we've heard complaints that ago. After

23:48

Mothers' where I didn't hear no error handily

23:50

needed let's those his letter read Now it's

23:52

a good job and that comes we're Credit

23:54

Where credit's you. That comes from linux. And

23:57

book Linux Him witness to Twelve Six Michael

23:59

from another one. Watson as

24:01

always. Thanks everyone! To look

24:03

for listen,