0:00

Hello and welcome to Python Bites

0:02

where we deliver Python news and

0:04

headlines directly to your earbuds. This

0:06

is episode 425 recorded March 24th

0:08

2025. I am Brian Ockin. And

0:10

I'm Michael Kennedy. We are excited

0:12

to announce that this episode is

0:14

brought to you by posit connect.

0:16

Listen to their segment later in

0:18

the show. You can also connect

0:20

with us and if you go

0:22

to our show notes and... Visit

0:24

all the links we're on, Blue

0:27

Sky and Mastida, mostly lately. And

0:29

if you'd like to actually thank

0:31

you for listening, but also if

0:33

you'd like to watch us live,

0:35

you can check us out at

0:37

PythonBites.fm. slash live, usually Mondays at

0:40

10am, but it's spring break this

0:42

week, so we're doing a little

0:44

early. And then finally, if you'd

0:46

like to have all the links

0:49

that we talk about, delivered plus

0:51

extra information delivered right to your

0:53

inbox. and sign up for the

0:55

newsletter. We don't spam you, we

0:57

just send you the newsletter. What

1:00

do you got for us

1:02

today, Michael? I want to

1:04

kick off with a Twitter

1:06

thread of all things. And

1:08

this one's pretty interesting. So

1:10

Armin Ronaker, original creator of

1:12

Flask, put out a question.

1:14

And it got a lot

1:16

more. attract a lot more

1:18

attention than I would imagine

1:20

it justified. And I just

1:22

I found it it's so

1:24

simple and I just found

1:26

it pretty interesting. Okay. The

1:28

question is, if you use Python, but

1:30

you don't use UV, why? You don't

1:33

use it yet, why? So I got

1:35

like 215 responses, almost a thousand likes,

1:37

things like that. So I want to

1:39

just call out some of, I'm not

1:41

going to go read all 215, top

1:43

level. He says, well, I'm going to

1:45

call it some of the ones that

1:47

I thought were interesting. And just get

1:49

your thoughts, Brian, as well. So for

1:52

example, someone said, I'm not going to

1:54

also give attributes because there's a bunch,

1:56

it'll take forever. So someone said, I

1:58

get it replaces PIP and PIP. and

2:00

V, but should I use it instead

2:02

of the built-in Python dash M, V

2:05

and V, V and V, right? Yes,

2:07

yes, you should. Yes, yes, you should

2:09

also do that. The reason I'm calling

2:12

this out, the reason I'm calling this

2:14

out, the reason I'm making this a

2:16

topic is there were a lot of,

2:19

I think a lot of misconceptions or

2:21

a lot of people are like what's

2:23

UV? VE&V is, well, what if you

2:26

don't have Python? Python, V and V,

2:28

we'll just say error, no Python, right?

2:30

UV says, oh, well, why don't we

2:33

just get you the latest version of

2:35

Python? Or you can say, 312, or

2:37

311, or whatever version you want, or

2:40

specific minor version, right? Or build version,

2:42

you write, 131, instead of 3132, whatever,

2:44

right? So it frees you from even

2:46

having to worry to worry about having

2:49

Python, Two new computers, I got a

2:51

new laptop and a new Mac Mini.

2:53

And I have not installed Python on

2:56

it. And I'm actually, was a little

2:58

bit bugged when something I had to

3:00

brew installed, installed Python. I'm like, darn

3:03

it, you're messing up my flow, right?

3:05

I don't want any Python in here,

3:07

other than virtual environments. And I want

3:10

those to be managed by UV. And

3:12

the thing is, it's two seconds. Like

3:14

with Pi, E and V. What it

3:17

does is it downloads and it builds

3:19

and it builds and it, which if

3:21

you don't have it can take. 10

3:24

minutes or more. With UV, it takes

3:26

two seconds, right? It's a thing of

3:28

beauty. Well, it's just ran into the,

3:31

the build, a problem with the build

3:33

recently because if the, there's, there's, there's

3:35

issues, especially on Windows, when if you've

3:38

got your dependencies mucked up somehow, it

3:40

can cause you to have to recompile

3:42

some stuff, and, and then you have

3:45

to make sure they have the right

3:47

like compiler libraries on your computer and

3:49

stuff like that. So that's a mess.

3:52

Someone says, but I need Python installed

3:54

to make. Python programs? I don't know,

3:56

yes, but see previous comment. But I

3:59

mean it is installing it, using UV

4:01

to install, like it isn't installing it

4:03

like you normally think, but behind the

4:06

scenes UV is putting it on your

4:08

computer. This is an interesting one. I

4:10

want to hear your thoughts on this.

4:13

I'm anti this only because my workflow,

4:15

not against the person who made the

4:17

comment, because I don't use it because

4:20

it places the VENV folder inside of

4:22

your project. And I can't run executables

4:24

from there due to corporate policy, many

4:27

such use cases. No idea why Estril

4:29

doesn't address this with more urgency. And

4:31

someone responded, that sounds like a bad

4:34

corporate policy. Also it does, you can

4:36

configure it to do that. Yeah, exactly.

4:38

There's an environment variable. You can set,

4:41

have it, put it in somewhere else.

4:43

Now, one of the things that made

4:45

me never ever adopt Conda, is if

4:48

you would go and create a Conda

4:50

environment for a data science project, it

4:52

puts it into some hidden folder. inside

4:55

of your profile instead of inside your

4:57

project. So I would go to a

4:59

project. I'm like, did I, you know,

5:02

a couple months later, I'd be like,

5:04

did I ever make a virtual environment

5:06

for this? Maybe. But then you got

5:09

to say, conda activate and name the

5:11

virtual environment. I'm like, what did I

5:13

call it? You know what I mean?

5:16

They're like, it drives me crazy. And

5:18

with it being inside the project folder,

5:20

you can look, yep, there's a virtual

5:23

environment there, I'm good to go. or

5:25

a regular LS or whatever, that, hey

5:27

look, it does have a virtual environment.

5:29

All right, I'm too lazy to switch

5:32

from high E&V and PIP. I mean,

5:34

it seems like a you problem. There's

5:36

lazy, like, I'm not going to walk

5:39

two miles to the store, and then

5:41

there's, I can't be bothered to put

5:43

the word UV, instead of PIP in

5:46

my command. Well, and like, I'm lazy,

5:48

that's why I am using UV, because

5:50

it's it's. speeds up everything and I

5:53

can get on with my day faster.

5:55

Yeah. Matthew Rockland of desk and coiled

5:57

had a really interesting comment once on

6:00

talk by thought that basically like so

6:02

much in programming happens because we're leveraging

6:04

the natural inclination for programmers to be

6:07

lazy. Let's set it up so they

6:09

could do this so they have to

6:11

do less work and do it. All

6:14

right, I'll keep going quicker now. Trust

6:16

issues, what if they do a bait

6:18

and switch, right? It's VC backed, what

6:21

if they go, oh, all of a

6:23

sudden, two comments there. One, it's open

6:25

source, and I'm sure a lot of

6:28

people have forked it. Although there are

6:30

still some potential issues, somebody's got to

6:32

run the Python Bill standalone for some

6:35

of the features. But there's that. And

6:37

they've already said that they won't, right?

6:39

Charlie Marsh said, like, our business model

6:42

does not hinge upon charging for UV.

6:44

It's kind of an open core model.

6:46

Okay, but Google used to say, don't

6:49

be evil. So, yeah, just saying. Brian,

6:51

I totally hear you. But did you

6:53

look, when they said it, like around

6:56

their back, they had their fingers crossed.

6:58

You gotta double check these things. And

7:00

no, I'm just kidding. I know they

7:03

did. Actually, yeah, and when VCs get

7:05

involved, and it's all well and good

7:07

when there's lots of money, things get

7:10

tight, then it could potentially. Well, and

7:12

also, we'll switch to something else then.

7:14

We'll switch to UV fairly quickly if

7:17

it becomes bad. Yeah, we'll switch to

7:19

VU. I don't know. So I have

7:21

a reason, but when you run out

7:24

of good reasons. All right, masochism. Another

7:26

one. Many times I've tried similar tools

7:28

and always come back to PIP and

7:31

PIP tools, aka PIP Compile. Those they

7:33

just work. I don't want something cool.

7:35

I know, but adding the word UV

7:38

in front of each of these commands,

7:40

like UV, PIP Compile, UV, PIP, install,

7:42

it's not a huge risk to roll

7:45

that back. Delie care. And we've tried

7:47

the other ones also, and they were

7:49

not as fun. Exactly. I tried this

7:52

week and was expecting UV install requests

7:54

instead of UV ad. And I brought

7:56

this one up. I switched back to

7:59

PIP Envy. I thought that was interesting

8:01

because there's sort of two. workflows with

8:03

UV. One says treat it like poetry

8:06

or hatch and let it manage a

8:08

project and that's the UV ad style

8:10

right it manages the lock file and

8:12

so on. They were expecting UV install

8:15

requests if they typed UV pip install

8:17

requests they would have been carrying on

8:19

with the regular workflow and I do

8:22

think it's worth shouting out that there's

8:24

like two unrelated workflows for for projects

8:26

and one is sort of the project

8:29

manager poetry hatch style. and one is

8:31

the more low-level PIP style. Yeah, yeah,

8:33

the, okay, but the, the UV is,

8:36

like, you can kind of flow between

8:38

the styles and it all just works,

8:40

whereas in, uh, in the other one,

8:43

you, like, you kind of had to

8:45

buy into it or not, but, yeah,

8:47

it's all or nothing, yeah. Another one,

8:50

really quick, we partially use it, we

8:52

will transition when pin-the-the-bot support is available.

8:54

and use the requirements.xt file compiled by

8:57

UV, you've got dependable support. For me,

8:59

dependable support is a net negative. I

9:01

hate dependable so much. I've tried to

9:04

turn it off on some of my

9:06

projects and I can't get it turned

9:08

off without turning off all the security

9:11

features at GitHub. So I've just had

9:13

to set up mail rules that say

9:15

delete anything from dependable. It says bump,

9:18

bump this thing. Because I would wake

9:20

up on Monday and I'd have probably...

9:22

2530 separate messages from dependable about saying,

9:25

hey, guess what? You got to update

9:27

for this. You got an update for

9:29

that. Oh, dependable. Please stop doing this.

9:32

And if they would just go, your

9:34

project has 17 updates. Okay, that would

9:36

be kind of helpful. 17 times your

9:39

project has an update is not helpful.

9:41

That said, if you use UV space,

9:43

pip compile and requires not TXC, you

9:46

have dependable support. All right, I'll leave

9:48

it with this one from Jared. seeing

9:50

a whole lot of Stockholm syndrome and

9:53

the replies to this question. Okay, interesting.

9:55

A lot of people were sort of,

9:57

you know, they were tied to the

10:00

things that were not working well or

10:02

they were tied to the... things like

10:04

well this is how we do it

10:07

and I know it sucks it's not

10:09

better but this is how we do

10:11

it yeah yeah the one thing that

10:14

I would like to okay there's there's

10:16

a reason that you really get a

10:18

lot of the benefit of UV if

10:21

you globally install it you do the

10:23

curl install or something right that that

10:25

is and also using having letting it

10:28

install Python for you with all the

10:30

other Python dependencies dependencies you can use

10:32

your own custom like a custom IPI,

10:35

like your own in-house IPI. With that

10:37

part, the curl install and the installing

10:39

the binaries, Python binaries, I don't know

10:42

if you can do that. So I

10:44

think that that still might be a

10:46

decent reason is because you're using a

10:49

firewall in the company. And that part

10:51

isn't working right. But you can still

10:53

pip install UV and I have done

10:55

time tests and it's still faster to

10:58

use the use UV even if you

11:00

pip install it So you just don't

11:02

get all the benefits. Yeah, absolutely awesome

11:05

Yeah, anyway, you got it already have

11:07

a project You can't start you got

11:09

a python you got a V and

11:12

V and then then you go from

11:14

there. Yeah, anyway, which is fine because

11:16

mostly that's what you're doing with life

11:19

You're not like constantly creating a new

11:21

project. I love our audience and Henry

11:23

Shrin Just shout it out depend about

11:26

just added support last week for UV

11:28

lock How about that? Do you know

11:30

how out of date that Twitter threat

11:33

is so last week? So last week?

11:35

Well, it also the lot of the

11:37

knowledgeable people probably didn't respond because it's

11:40

on X Yeah, yeah, it's true. I

11:42

mean, there's it's a surprisingly amount of

11:44

surprise and amount of interaction over there

11:47

for for what it is. Yeah. Okay

11:49

over to you. Okay, let's talk about

11:51

UV a little bit Okay, we haven't

11:54

even run that yet. So I want

11:56

to talk about this cool project from

11:58

Tim Hopper. Love Tim. Kind of miss

12:01

seeing him. He's one of the fellows

12:03

that he used to run into at

12:05

conferences all the time. and haven't been

12:08

to conferences lately. So, hi Tim. Anyway,

12:10

he put out the Python developer tooling

12:12

handbook, and it's a in progress thing,

12:15

but he's been working on it for

12:17

a while. Yeah, Tim Hopper, great guy.

12:19

Anyway, the Python developer tooling handbook, and

12:22

it's walking through a lot of the,

12:24

basically the new models for doing things.

12:26

There's, I don't know. says that I

12:29

guess he does comment on a lot

12:31

of these things, UV poetry, flit, sub

12:33

tools hatch. But I was noticing right

12:36

away he's doing tutorials and how-to's and

12:38

explanations and kind of a cool and

12:40

then some references and kind of a

12:43

cool way to set up some documentation.

12:45

But like let's say, publish your first

12:47

Python package or let's do a create

12:50

your first... Python project. And instead of

12:52

doing a lot of the history, he's

12:54

just popping to you to the right

12:57

answer. It's a UV and it. And

12:59

I mean, actually a lot of the

13:01

other in it, like initializing projects, I

13:04

haven't liked before, even like the flit

13:06

in it, I didn't like, but UV

13:08

in it is pretty darn good. They're

13:11

doing a good job. And anyway, so

13:13

using. and then adding dependencies and I

13:15

love this like let's skip the all

13:18

the well you could do it nine

13:20

different ways just tell me the way

13:22

to do it or at least a

13:25

way it's opinionated but it's a good

13:27

opinion and publishing your first package talking

13:29

about setting it up anyway cool resource

13:32

I hope to see it grow a

13:34

lot so anyway like it encourages people

13:36

to test it encourages people to use

13:38

test pipi i first before they go

13:41

somewhere else although be careful because a

13:43

there's a lot of projects that don't

13:45

use test pipi so and and they

13:48

clean that one out every once in

13:50

a while so even if a package

13:52

is available on the test pipei-i-i, it

13:55

might not be available on the big

13:57

pipei-i-i, so. Yeah, yeah. So you're like,

13:59

oh, it uploaded, so that means there's

14:02

no name conflict, like, no, no, no,

14:04

no, search, search pipei. Yeah, but one

14:06

of the things I wanted to bring

14:09

up also is, let's see, right at

14:11

the top. It says, this is not

14:13

a book about programming Python. Instead, the

14:16

goal of this book is to help

14:18

you understand the ecosystem of tools used

14:20

to make Python development easier and more

14:23

productive. So it's going to grow. I

14:25

commented about some of the testing and

14:27

he said he's going to add some

14:30

stuff about testing as well. He said

14:32

the handbook is structured according to the

14:34

DIA taxes framework, which... Okay, didn't know

14:37

anything about this, tutorial, how to explanation

14:39

and reference. So I'm not going to

14:41

link to it, but there's a link

14:44

here on the on his project about

14:46

it. And I'm going to have to

14:48

read more about this. This is interesting.

14:51

I've never heard about this. So I

14:53

had not heard of it either, but

14:55

it does look interesting. Anyway, and some

14:58

great information about setting up. setting a

15:00

rough to check your code and things

15:02

like that. So I've been so excited

15:05

about UVV and V and UVP and

15:07

stuff like that that I kind of

15:09

forget that I'm using UV rough all

15:12

that are using rough all the time

15:14

also and it's kind of bundled into

15:16

there. So I'm curious while we're talking

15:19

like kind of transition to rough for

15:21

a second. I was listening to that

15:23

interview you did recently. We were you

15:26

talking about I think you interviewed Charlie

15:28

recently. And I there's I like I

15:30

like how detailed rough can be but

15:33

this is one of the things that

15:35

I think I'm I'm rough can you

15:37

can go in the pie project tomal

15:40

or there's rough tomal also now and

15:42

I kind of like I think I'm

15:44

in my transition to using the rough

15:47

tomal or try to because I end

15:49

up having like a decent amount of

15:51

little tweaking the rules a little bit

15:54

and I'm not really tweaking the project

15:56

I'm just just tweaking the the the

15:58

the linting rules so are you using

16:01

do you use rough and where do

16:03

you put your rules? I love rough

16:05

and I'm all about the rough dot

16:08

tomal yeah it's yeah it definitely takes

16:10

more than a page of small font

16:12

on my screen okay I don't want

16:15

to you know you could reasonably have

16:17

a pie project that tomal that's smaller

16:19

than the rough tomal so you know

16:21

I don't want to put those together

16:24

yeah and I loved his the the

16:26

Charlie comment about the rules of the

16:28

Even he doesn't turn them all on,

16:31

that would be silly. So I tried

16:33

it. I tried turning all the rules

16:35

on. And there's some rules that are

16:38

inconsistent. There's like two or three rules

16:40

that are conflict with other rules. So

16:42

you have to decide, you know, if

16:45

you turn them all on, there's actually

16:47

a built-in conflict because a couple of

16:49

the rules conflict. But it gives you

16:52

really great detailed information about how, anyway,

16:54

how to fix it. Okay. Yeah, another

16:56

benefit from using a rough tumult over

16:59

pie project. is if you're using Docker

17:01

or any, basically any continuous integration style

17:03

thing where it looks at, well, if

17:06

any of the source files change, I

17:08

need to rebuild or redeploy, restart your

17:10

web apps, your APIs, or whatever, you

17:13

can make changes to your rough tumult,

17:15

like real code changes, like real code

17:17

changes, so, like real code changes, so...

17:20

I know it's not a huge deal,

17:22

but that's like a plus one in

17:24

the column of separating those things. Yeah,

17:27

so if you're just doing a handful

17:29

of rules, it might be fine in

17:31

your project. Toml, but if you're experimenting

17:34

with turning on more and more rules,

17:36

then yeah, getting it separate. I do

17:38

look, it's it's both surprising and cool

17:41

that that by default not all of

17:43

them are on, because a lot of

17:45

linters in the past they defaulted to

17:48

all on. and you'd try to use

17:50

the tool and it would just like

17:52

blow up with errors. But rough does

17:55

not. It's got some good standard ones

17:57

and then you have to go. turn

17:59

them on. So anyway. Yep. One more

18:02

piece I saw in the toolbook here

18:04

that it was being added as with

18:06

a UV ad. I put rough into

18:09

my system using UV tool install rough

18:11

because just globally and then have a

18:13

rough tumult for each project to configure

18:16

it. So that way you don't have

18:18

to have updates to your project or

18:20

make sure you remember an install. It's

18:23

just like a globally available tool. Yeah.

18:25

Yeah. I should do that. I don't

18:27

know why I don't. Anyway,

18:29

well, should we talk about the

18:31

wonderful people at Pozt that are

18:33

helping us today? Absolutely. Okay, do

18:36

it. This episode of Python Bites

18:38

is brought to you by the

18:40

folks at Pozt. Pozt has made

18:42

a huge investment in the Python

18:44

community lately, known originally for our

18:46

studio. They have been building out

18:48

a suite of tools and services

18:50

for Team Python. Today, I want

18:52

to focus on hosting your Python-based

18:54

data science workloads. This includes dashboards,

18:56

reports, interactive web apps, all the

18:58

way to custom flask and Jango

19:00

apps. Their service is positive connect.

19:02

Positive Connect makes it easy for

19:04

data scientists to share work built

19:06

with Python code. If you have

19:08

a streamlet app, dashboard, plotly interactive

19:10

plots, a fast API service, a

19:12

fast API service, or even a

19:14

quarto report. Just give positive connect.

19:16

The code it needs to maintain

19:18

the asset and Connect automatically does

19:20

the rest. Connect will manage your

19:22

APIs and serve your interactive apps.

19:24

And if desired, update your ports

19:26

and dashboards on a scheduled basis.

19:28

That's right. No more need to

19:30

explain to the stakeholders why the

19:32

dashboard or plot stopped updating last

19:34

week. You get to focus on

19:36

your data science and leveraging your

19:39

skill set while Connect makes you

19:41

look good, keeping your code running

19:43

in private. With Connect, you get

19:45

a private URL on your Connect

19:47

server. ensuring that your asset is

19:49

continuously available to your shareholders. You

19:51

can also control which users have

19:53

access to the asset. Let positive

19:55

connect handle... the delivery and Devops,

19:57

including your, involved in sharing your

19:59

work. You focus on what you

20:01

do best. So if you work

20:03

on a data science team, you

20:05

owe it to you and your

20:07

org to check out posit Connect.

20:09

Visit PythonBites.f.f.com/Connect today and get a

20:11

three-month free trial to see if

20:13

it's a good fit. That's Python

20:15

bytes.fm.com/Connect. The link is in your

20:17

podcast players show notes. Thank you

20:19

to Posit for supporting Python Python

20:21

Python. Absolutely, thank you posit. That's

20:23

sort of a data science theme,

20:25

right, Brian? Let's go and talk

20:27

about something from Adam Chains. And

20:29

I know we talked about Ruff,

20:31

but what about Black? So Black

20:33

is still a super important, super

20:35

influential thing. I imagine with Black.

20:37

Without Black, maybe Ruff doesn't exist.

20:39

So I want to come up,

20:42

I want to point out a

20:44

tool called Blackin Docs. So if

20:46

you have documentation, in that documentation,

20:48

maybe... you've got some code blocks,

20:50

some Python code blocks, and you

20:52

would like to format them. I

20:54

don't know how well your Markdown

20:56

editor formats code, but I'll tell

20:58

you what, I love my Markdown

21:00

editor, which is Typora. It does

21:02

not, it does not have a

21:04

reformat according to rough style feature

21:06

whenever I say tick tick tick,

21:08

Python, types and stuff, right? This

21:10

does. And so you can, you

21:12

can run this against all sorts

21:14

of different things. So it says

21:16

documentation, documentation. What it will do

21:18

is it works on restructured text

21:20

or it works on markdown files.

21:22

Okay. For example, I'm working on

21:24

that Python and production book from

21:26

Talk Python, which is like 250

21:28

pages of Markdown. There's a crap

21:30

ton of tick tick, tick, tick,

21:32

Python, something or other in there.

21:34

Oh, yeah. I could run it

21:36

against that. It doesn't have to

21:38

be documentation. Anything that's restructured text

21:40

or markdown that has Python, you

21:42

know, full-sized code block sections, you

21:44

can just hit it with that.

21:47

Oh cool, that's cool, right? Yeah.

21:49

You can also install a pre-commit

21:51

hook. It comes with one. And

21:53

then when anybody commits, it'll automatically

21:55

format that. And I think, does

21:57

it do doc strings as well?

21:59

Let's see. I believe, yeah, you

22:01

can run it on Python files

22:03

to reformat doc strings. Yeah. So

22:05

doc strings and basically mark down

22:07

and restructure text. That's pretty awesome,

22:09

right? Yeah. So there you go.

22:11

You can even turn it off

22:13

with some comments, like, no. Please

22:15

don't do it here. And so

22:17

on. So just to be clear,

22:19

I mean, it's probably obvious, but

22:21

it's not, it's not, it's, it's

22:23

formatting the code blocks within your

22:25

Markdown files to black style. There

22:27

aren't black styles for Markdown. Yes,

22:29

exactly. Okay. Although, that's an interesting

22:31

idea as well. I actually wouldn't

22:33

mind a, something like black for

22:35

Markdown because it tries to mean,

22:37

I do headings with the. pound

22:39

and pound pound and and but

22:41

mark down allows the underscore the

22:43

like the the dash line line

22:45

underneath it wouldn't and I hate

22:47

that. Yeah I do. So it's

22:50

like somebody came from a structured

22:52

text and stuck it in there.

22:54

Let's make it easier to convert

22:56

our restructured text. No. No please

22:58

don't do that. We're here because

23:00

we don't like structured text. Don't

23:02

you understand what's happening? All right

23:04

well anyway. Thanks Adam for that.

23:06

That's a cool looking project. Yeah.

23:08

I'm going to stick with some

23:10

data science themes for a little

23:12

bit longer. So I want to

23:14

talk about reinventing notebooks as reusable

23:16

Python programs. Have you heard of

23:18

this, Muramo, Murimo thing? Merimo, yes.

23:20

I just had the founder of

23:22

Merimo on Talk Python. Okay. Yeah.

23:24

Yeah, it's not out yet, but

23:26

it's been, you know, been recorded

23:28

and available on YouTube. Okay, this

23:30

is sort of interesting. I'm like,

23:32

oh, okay, somebody else is reinventing

23:34

notebooks or something. But this is

23:36

pretty cool. So this is, what

23:38

are we looking at here? What

23:40

we're looking at is, I'm gonna.

23:42

down to some code examples is

23:44

that a notebook doesn't have to

23:46

look like a notebook anymore in

23:48

like Jason form or something it

23:50

can look like Python or at

23:53

least with with the Merrimo environment

23:55

and I'm not quite sure you

23:57

probably know more than I do

23:59

then it's can you just open

24:01

these up into into notebooks or

24:03

do you have to do something

24:05

special to get them open so

24:07

it's a different execution environment than

24:09

saying Jupiter Space Lab or Jupiter

24:11

Space Notebook. It's a different runtime.

24:13

Okay. But they're the real server.

24:15

Okay, got it. But I really

24:17

like that. It's just, it's just,

24:19

it's just, well, it's not just

24:21

Python code. It's extra stuff also.

24:23

Looks like they add some decorators

24:25

to make their, their environment understand

24:27

what it is. Like there's app

24:29

dot cell and at app dot

24:31

function. And. probably other stuff too.

24:33

But some of the benefits of

24:35

this are like incredible. So the

24:37

some of the benefits are that

24:39

you can version with get easier.

24:41

Well you can do that with

24:43

Jason too, but the diffs don't

24:45

really make a lot of sense

24:47

often. The diffs are all over

24:49

the place. You can test with

24:51

pie test and that one. I'm

24:53

like really that's what that's been

24:55

one of the problems because in

24:58

the past that testing is a

25:00

kind of an issue or at

25:02

least it has been long time

25:04

ago when I tried it. Maybe

25:06

they've solved it other ways but

25:08

testing even with doc test and

25:10

some of the other other stuff

25:12

in there embedding sequel and mark

25:14

down like all the sort of

25:16

stuff that you could normally do

25:18

with Python you can now still

25:20

do with these and and that's

25:22

it's just kind of cool. I

25:24

wanted to check this out and

25:26

try it out. This is available

25:28

for everybody to run right. Yeah,

25:30

yeah, it's a goal for people

25:32

to run. You can download it,

25:34

run it locally. They also have

25:36

some some cloud options and it's

25:38

really neat. There's super nice ways

25:40

to host it. You know, you

25:42

just say. run it as service.

25:44

One of the things that's pretty

25:46

interesting here, let me steal the

25:48

stage for a moment, is this

25:50

all runs inside of, in your

25:52

browser, client side, so that's one

25:54

of the big differences with, say,

25:56

Jupiter versus this, is it's powered

25:58

by web assembly and pie oxide.

26:01

Okay. Pieodide, rather, sorry. And so,

26:03

for example, this one that I

26:05

pulled up, I'll put it in

26:07

the show notes, is the F1

26:09

driver career explorer. And look how

26:11

nice this looks Brian. Like in

26:13

terms of relative to compare that

26:15

to Jupiter, for example, I think

26:17

it's got like a really beautiful

26:19

style, you know, you can check

26:21

out. This is old, I got

26:23

to put Ferrari in here for

26:25

Hamilton. I only got Mercedes and

26:27

McLaren, but it was powered by

26:29

say DuckDB Web Assembly, for example,

26:31

and Efron Data. And yeah, it's,

26:33

you know, it's got little explorer

26:35

sort of things and. It just

26:37

feels really polished. It also has

26:39

a collaborative edit sort of thing

26:41

like Jupiter notebooks. I'm sorry like

26:43

Google Dogs rather. So you can

26:45

like pair up on them. There's

26:47

a lot of neat stuff about

26:49

this. Okay, but this is okay.

26:51

It's still cool. I was but

26:53

I'm misunderstood. I thought maybe it

26:55

was just a new way to

26:57

run Jupiter notebooks. But yes, but

26:59

with a decent amount of new

27:01

features. I mean, yeah. Yeah. Well

27:04

I guess the thing I'm missing

27:06

then or not understanding is is

27:08

there a relation between this and

27:10

Jupiter notebooks is can can you

27:12

use any of the the rest

27:14

of the Jupiter environment or is

27:16

it? I believe so. I'm trying

27:18

to remember now like how much

27:20

integration there is between those but

27:22

yeah I think so. Okay cool

27:24

I'm I'll investigate more so anyway

27:26

I'm still pretty excited about it

27:28

looks pretty neat. I'll have to

27:30

play with it a bit so.

27:32

Yep, absolutely. And it's also another

27:34

thing that it really tries to

27:36

solve is they are trying to

27:38

solve the, oh you know what

27:40

I said it had a Google

27:42

Docs integration, not exactly, but what

27:44

they're trying to solve is the

27:46

real challenges of integrating notebooks with

27:48

get. So for example, with notebooks,

27:50

there's, they're really Jason files, and

27:52

then they've got the input blocks

27:54

and the output blocks. So if

27:56

you've run some code and it's

27:58

dependent on anything that might change,

28:00

you'll get like huge, instead of

28:02

getting a nice merge, you'll get

28:04

two conflicts of like giant embedded

28:07

blobs and so on. Yeah. The

28:09

other thing that this really addresses

28:11

around notebooks is notebooks allow you

28:13

to run them in any order,

28:15

which is kind of insane. Like

28:17

I could run cell one, cell

28:19

two, cell three, cell three, cell

28:21

three, cell four. change cell two

28:23

run cell four again to see

28:25

what the output is but maybe

28:27

that the changes would have made

28:29

cell two be different or cell

28:31

three be different you know and

28:33

you can like skip over you'll

28:35

have like using I have little

28:37

numbers on the side of which

28:39

order it was running to be

28:41

like one seven eight nineteen fourteen

28:43

or something you know as you

28:45

go down you're like oh this

28:47

so this is reactive in the

28:49

sense that every variable you create

28:51

is like wrapped, it's not technically

28:53

wrapped in a reactive behavior thing,

28:55

it's they use, they parse the

28:57

dom every time you run anything.

28:59

And they create relationships between cells,

29:01

say like, okay, cell 10 uses

29:03

something from cell 1 and cell

29:05

3. So if you run cell

29:07

3, it knows that cell 10

29:09

is out of date, or potential

29:12

will just rerun it for you.

29:14

So it keeps track of the

29:16

relationships, even. depending on the order

29:18

so that you don't end up

29:20

with like stale data and then

29:22

run another cell that pulls in

29:24

that stale data and so on.

29:26

So it's trying to really address

29:28

that as well. That's cool. Yeah,

29:30

that's one of the things that

29:32

threw me off when I first

29:34

started using notebooks or a little

29:36

bit. Yeah. That's a little weird

29:38

like why is it allowed to

29:40

just run this piece in in

29:42

from an exploratory perspective that that

29:44

makes sense like well that was

29:46

all expensive and I just want

29:48

to ask questions down here and

29:50

over here but from a reproducibility

29:52

perspective and debugging it's bad yeah

29:54

yeah so anyway I need some

29:56

something fun new to play with

29:58

with indeed yes it's pretty cool

30:00

so that's that's our item Do

30:02

you have any extras for us?

30:04

I do have some extras. Let's

30:06

see a couple real quick. Pike

30:08

on Taiwan 2025 is doing a

30:10

call for proposals. It's going to

30:12

be in the fall. And so

30:15

you've got until April 5th, anywhere

30:17

in the world. So if you

30:19

want to go to Taiwan and

30:21

present stuff there, they have a

30:23

decent number of English tracks and

30:25

talks and stuff and they're looking

30:27

for more. So check that out.

30:29

Also, a follow-up on the Hacker

30:31

News thing. Remember, Brian, I was

30:33

reading the Hacker News thing saying,

30:35

who's looking for a job? Who's

30:37

hiring? Well, Shinjitsu sent a semester

30:39

and said, I'm sure some other

30:41

Hacker News readers have already told

30:43

you this. No, they haven't. So

30:45

thank you, Shinjitsu. But if you

30:47

haven't, you kind of got this

30:49

exactly wrong. Like, oh, sorry. Ask

30:51

Hacker News, who is hiring is

30:53

a monthly post that asks employers,

30:55

what jobs they're posting. not people

30:57

asking which one and who wants

30:59

to be hired is a monthly

31:01

topic that has people who are

31:03

looking for jobs about themselves so

31:05

did we have that backwards there

31:07

was actually bad news yeah I

31:09

think that's what it's saying let

31:11

me see yes I think I

31:13

said who wants to be hired

31:15

was asked by the employers and

31:18

who is hiring was asked by

31:20

the people looking for a job

31:22

but it sounds like it's the

31:24

other way around so that would

31:26

make it less rosy so sorry

31:28

about that folks if that was

31:30

a little bit backwards oh okay

31:32

see Arata. Arata, yes, we have

31:34

to have an Arata. How about

31:36

you? I got a couple things.

31:38

First off, Picon U.S. announced that

31:40

there's a refund policy for international

31:42

attendees if you qualify for certain

31:44

things. And actually, I'm just embarrassed

31:46

as American that we have to

31:48

do stuff like this. But international

31:50

travel to the United States has

31:52

become more complex for many of

31:54

our community. Picon U.S. welcomes all

31:56

community members to Pittsburgh. But if

31:58

for some reason you can't come

32:00

because of very... problems. Read the

32:02

post for details, but they're a

32:04

fairly, I think a fairly cool

32:06

refund policy details. So anyway, if

32:08

you can't come for some reason,

32:10

check this out and maybe... I'm

32:12

thinking a lot about this. There's

32:14

been a lot of badness and

32:16

I'm not happy about it, Brian.

32:18

I'm not happy that our country

32:20

is treating people that way. It's

32:23

one thing... if you catch somebody

32:25

trying to sneak over the border

32:27

and I mean it's a political

32:29

debate on how lenient or aggressive

32:31

you want to enforce them but

32:33

when people especially from fellow ally

32:35

countries just try to come over

32:37

and they get put in jail

32:39

because they didn't have the right

32:41

kind of visa or something that

32:43

is that is some bad stuff

32:45

it's well you would just never

32:47

think that that was that was

32:49

the way the US treat people.

32:51

I know they've been a little

32:53

bit tough on immigration, but you

32:55

know Australia has, Canada as I

32:57

was detained at the Canadian border

32:59

for a long time and finally

33:01

was let in the Canada. And

33:03

it's, okay, it's a hassle. But

33:05

this is another level of jerkery

33:07

and yeah, I am glad to

33:09

see that Picon is doing this.

33:11

I apologize to the world. I

33:13

guarantee you, Brian and I are

33:15

not supporters of this policy. Of

33:17

the reef, we were supporters of

33:19

the refund policy. Yes, not the

33:21

cause of the refund policy. Yeah.

33:23

Right. Okay. Oh, on a slightly

33:26

more positive note, I finally finished

33:28

up restructuring the Complete Pie Test

33:30

Course Series. So, originally it was,

33:32

so this is over at Python

33:34

test.com. This originally was I had

33:36

a remaking the the pie test

33:38

book I wrote as a course

33:40

and it was one course but

33:42

that really the book was into

33:44

three parts and intended to be

33:46

read at different times during your

33:48

development process when you're first starting

33:50

out at pie test then when

33:52

you're trying to apply it to

33:54

projects and then when you're like

33:56

really good at it. and you

33:58

really want to like take off

34:00

and get great at it, there's

34:02

another there's another section. And I

34:04

had it in one course and

34:06

it doesn't write quite, I don't

34:08

think it really quote quite fit

34:10

how people use pie test and

34:12

learn pie test. So now it's

34:14

split up. So you can still

34:16

get the complete bundle and it

34:18

gets the three parts, but the

34:20

three parts learning the basics of

34:22

pie test is a separate course

34:24

now, you can just buy it

34:26

separate. And then you can grab

34:29

part two. and then part three,

34:31

but also part two is working

34:33

with projects, part three, the booster

34:35

rockets, like building, really building plugins

34:37

is why you'd probably grab this

34:39

and advanced parameterization techniques. However, even

34:41

if you didn't buy the other

34:43

ones, I like having them separate

34:45

now because if somebody's like, I

34:47

really want to build my own

34:49

plugin, you can just go check

34:51

out this one course instead of

34:53

taking all three of them. So

34:55

that's my big news there. The

34:57

other reason why the other structuring

34:59

restructuring that I did is I...

35:01

All the chapters are now a

35:03

single video per chapter, for the

35:05

most part. Chapters two and three

35:07

are a bit long, so I

35:09

split them up so they're about

35:11

20 minutes each. But the thing

35:13

that I really like about this

35:15

now is that in each video,

35:17

you can pop in and you

35:19

can change the speed. So you

35:21

can watch like... the whole like

35:23

I mean I'm usually a 1.25

35:25

or 1.5 speed kind of person

35:27

and it was bugging me that

35:29

you had to reset that setting

35:32

every three minutes like for a

35:34

three or five minute video now

35:36

now it's only like you can

35:38

only set it once for a

35:40

20 minute video so anyway works

35:42

better for me hopefully it works

35:44

better for everybody else too so

35:46

that's my that's my big extra

35:48

so yeah awesome congrats on getting

35:50

that all on redone that's great

35:52

yeah so how about a joke

35:54

boldly go And no one was

35:56

gone before. Yeah, so last week's

35:58

joke was around was a Star

36:00

Trek cling on theme. And we

36:02

had who was it? We had

36:04

somebody from Holgi on Mastodon said

36:06

you again, you probably already know

36:08

this but. There's this thing called

36:10

the Klingon programmers. So this is

36:12

over at Cornell.adieu. Top 12 things

36:14

likely to be heard if you

36:16

are a Klingon programmer. And I

36:18

kind of love this. So we

36:20

should probably zoom in a little

36:22

bit. Okay. We're not going to

36:24

read all 12, but number 12

36:26

is actually pretty good. Specifications are

36:28

for the week of, for the

36:30

week in timid. Do you want

36:32

to read any of these? Let's

36:34

see. This machine is a piece

36:37

of God. I need a dual

36:39

Pentium processor if I am to

36:41

do battle with this code. Yeah,

36:43

number four, a true Klingon warrior

36:45

does not comment his code. debugging.

36:47

Klingons do not debug. Our software

36:49

does not coddle the week. I

36:51

think the perfect... Oh, no, go

36:53

ahead. Klingon functions do not have

36:55

parameters. They have arguments and they

36:57

always win them. I love it.

36:59

All right, let's round out the

37:01

tie back to last week, which

37:03

was we will test in production.

37:05

Okay, yeah, so this is, what,

37:07

number one? Number one, do it.

37:09

Our users will know fear and

37:11

cower before our software. Ship it,

37:13

ship it and let them flee

37:15

like the dogs they are. Exactly.

37:17

This is good. This is a

37:19

reminder also to everybody that if

37:21

you think we probably have already

37:23

heard it, there's a decent chance

37:25

we haven't. So send it to

37:27

us anyway. We don't mind repeats.

37:29

So if you find something funny

37:31

or something like this. A lot

37:33

of credit, Ryan. Maybe too much

37:35

more than we deserve sometimes. Yeah.

37:37

Anyway. Well, thanks again for jumping

37:40

on the call with us and

37:42

enjoy doing another episode. And thank

37:44

you everybody in the... Everybody that

37:46

listens and everybody who shares this

37:48

with other people, we like to

37:50

grow our, grow our community. Yeah,

37:52

very much appreciate everyone. See y'all

37:54

later. Bye. Bye.