0:00

Hello and welcome to Python Bytes, where we deliver Python news and headlines directly to

0:04

your earbuds. This is episode 414, recorded December 16th, 2024. I'm Michael Kennedy.

0:12

And I'm Brian Okken. And this episode is brought to you by us, all of our things, books, courses, stuff like that.

0:18

We have many things for you to get better at Python. Check them out, links at the top of the

0:23

show. And we are now pretty active on Blue Sky. Brian, thanks for dragging me over. And I want

0:30

to point out that you can go to my profile and click on Starter Pack. And there's a bunch of Python

0:35

people. If you click that, you follow me, you follow Brian, you follow the podcast, plus something

0:40

like 60 other noteworthy Python people like Samuel Colvin and others. So that's a real quick and easy

0:47

way to jump in there and kind of get going on that.

0:50

Starter Packs are pretty cool. Starter Packs are a really cool growth hack for Blue Sky. Yeah.

0:56

Yeah. Blue Sky is fun. I'm enjoying it over there. I'm wondering if the surge of momentum is starting

1:03

to fade a little bit, but we'll see. I know people are excited and I think it's a great place. So

1:08

I encourage people to check it out. If you would like to get every week, all the show notes and

1:14

links and everything delivered to your inbox, even if you don't happen to listen that week, which I

1:19

don't know why that would happen, Brian. That would be a big mistake.

1:21

That'd be weird. It would be very weird. But even then, if you go to Python by set of M, click on newsletter,

1:28

enter your information. Well, then Brian will send you a handcrafted, artisanal version of

1:33

the show notes. So very awesome. I did say heavy, Django heavy, didn't I, Brian?

1:38

Yeah. Let's start. Let's start with Django.

1:41

Well, we're going to start with a small item, but I think it's going to affect me right away. So Jeff

1:47

Triplett announced he's got a new project to shorten Django-admin to just Django. And what we're

1:56

talking about isn't the admin section of Django. It's just the command. So the thing, like on the command

2:02

line, I have it up in like a tutorial. So like, like for instance, the Django tutorial, there's,

2:08

there's, there's a bunch of Django admin command line things that you have to run like start project

2:15

or, and a bunch of other stuff too. So if you type Django-admin and some stuff like that,

2:21

and you've ever thought, why can't I just type Django to do this? Jeff has thought that too. So that

2:27

this new project just makes it so that you, you just pip install it with your stuff. And then you,

2:33

you don't have to use Django admin anymore. You can just use Django for, for the, the command line

2:39

thing. So the, and the idea, I mean, it's a, it's a great idea. And he does say in his, we've got a

2:46

blog post announcing it that he would like to see this in the normal Django, but the Django-admin won't

2:53

go away anytime because it's already been there for 20 years or more. So, but it'd be great if it

2:59

was just like this, because why not? Yeah. That'd be amazing. I don't see why you couldn't have them

3:04

both. And I think, I mean, what other commands do you type Django on the command line and do stuff for

3:09

most people, right? Yeah, exactly. You just take it. Yeah. So exactly. And you can have multiple entry

3:15

points in the Django package when you install it, which will give you both commands as you see fit. So

3:21

yeah, let's do it. Yeah. Also I've been doing a lot of Django lately and I appreciate just,

3:28

it's half the characters. If I counted them, he mentioned it's half the characters. I'm like,

3:33

really? Yep. It's not just half the characters, even more significantly, it's the dash, right? Which

3:38

requires like a incantation on your keyboard a little bit. Yeah. The different, yeah. It's not as fast to

3:44

type. So yeah. Thanks Jeff. So I would like to ask you, Brian, having done some Django lately,

3:50

have you seen any unicorns? Yeah. No. Because I have. I saw a Pegasus, but not a unicorn. Yes. The

3:57

Pegasus. That is like a unicorn, but I don't believe it has a horn or as much magical powers. Okay.

4:03

I'm not entirely sure of the mythology of Pegasi versus unicorns, but the magical reactive component

4:11

framework for Django is the Django unicorn. This is pretty neat. I just learned about this. Let's see,

4:17

is it, it's, it's not super new. I just, it's super new to me. Okay. So the idea here is that it's a

4:24

little bit like a JavaScript front end framework, like react or something, but you can avoid using

4:32

it, right? You can avoid writing your own JavaScript front end. Instead, you can just pip install Django

4:36

unicorn, add it as an app. And then you've got to include it scripts and so on. But somewhere down

4:42

here, you can use these unicorn attribute modifiers in your template, right? You can say unicorn submit.prevent

4:49

add, add, and then instead tie that to, if I don't scroll too quickly, you can tie that to a model

4:55

called task. And if you hit escape, what does it do? It'll change, like replace the task text with that,

5:02

this thing right here, right? And you can just add a button when you click this, call the function add

5:08

and so on. And then what you do is you go and you create like a form object and an item that maps to it.

5:15

And then it just automatically wires together the creation of, of these, let's see, where's the

5:20

example. So yeah, not enough here on the little example on the homepage for me to totally know

5:25

exactly how it works. But basically when you interact with this UI element, it maps to rest functions

5:32

implemented by the unicorn thing automatically, right? So you can just include some, it's a little bit

5:39

like HCMX. You just include some magical text on there and it'll call back to the server. But the

5:44

difference is it'll handle it on the server as well. Right. So pretty neat. Yeah. So if it says,

5:49

is it magic? Sort of feels like it, it progressively enhances normal Django views with the initial render

5:57

being server side rendering, which is like I said, like HTMX as well, depends how you use it, but it

6:03

can be. So it's good for SEO. It just the pure HTML content is there. It's not just like angle

6:09

brackets everywhere. When you view source, it binds the elements you specify automatically and makes the

6:14

Ajax calls on its own when it needs it. And it comes back and it updates the DOM when the HTML changes

6:20

or with the HTML changes. So yeah, just write normal Django type stuff. And it takes it, it says,

6:25

it also has other features, form validation, redirections, dirty states, partial updates, polling,

6:31

etc. And then down here somewhere, maybe it's in the docs. It says, here's what you might do instead.

6:37

Right. You might have to use reactor. You might have to use this other thing. Unfortunately, I think

6:43

that's here on this page. It says what else you might have to use. Yeah, there's better examples on

6:48

their landing page. It gives you sort of a comparison to view, react, etc. What is notably lacking here is

6:56

HTML, which like I said, is kind of like it. But nonetheless, if people are doing Django and they

7:01

want something like viewer react, but they don't really want to do JavaScript, this could be a pretty

7:05

interesting thing to check out 2400 GitHub stars, pretty decent. And this is and this is a fairly basic,

7:11

like low learning curve to just try to it might be enough before you jump into something else. So

7:18

Yeah, absolutely. And by the way, real time follow up on your, your item. This is what I thought when I

7:24

heard about the Django admin, Pat Decker just says alias Django equals Django dash admin. And that's

7:29

exactly I'm like, yep, I have so many things like that. Like that's a huge long command. That is

7:34

alias to two or two letters. Okay, fine. I could do that also. But but it's nice. I mean, why why impose

7:40

the longer version on everybody until they get either think about it or you know, a lot of people are new

7:46

when they take these tutorials, they don't know that they can do that kind of stuff.

7:49

Right. Or you might be like on Windows, and I have no idea how to alias anything on Windows.

7:53

So batch files, it's all batch files. All right, over to you.

7:59

Oh, what am I talking about? I wanted to talk about testing a little bit. I had fun time reading

8:06

this article from Ned Batchelder called testing some tidbits. So this is sort of a fun thing. So he posted,

8:13

he just posted, like, let's say you had this, you wanted to, he wanted to look at different ways to

8:19

check to see if a string only had zeros or ones in it, and nothing else. And there's a lot of ways you

8:26

could do this. He presented, let's see, one, two, three, four, five, six different ways in a post on

8:33

both Mastodon and Blue Sky. And then he got a whole bunch of replies saying, like other ways to do it.

8:41

And, and one, it's kind of a fun, just like, like, how would I do this sort of thing? And there's a lot

8:47

of ways to handle it, which is fun. Anyway, so he, he also wanted to test how to do it. And since

8:53

there's testing in the title, I thought maybe he'd use my test or something. But yeah, no, he's got

9:00

like a set of good input that should be just all zeros and ones. And then a set of bad input that has,

9:06

you know, it's not all, and some of it's like a whole bunch of zeros and just one non zero. The

9:12

empty string is used. That would be considered good. And then I, even a thousand character,

9:17

a rain, a string with a thousand characters in it or 10,000, it's 10,000. Wow. So big, long string.

9:25

So, and then a whole bunch of different ways that he had, he has original checks, but plus a whole

9:30

bunch of others that other people used. And then his, he just runs through them. But that one of the

9:36

things I really loved about this is I learned some stuff about pytest or Python. I learned about

9:41

clean doc from inspect.clean doc. It's a way to strip out a white space that I usually use. What did I use?

9:50

I use, usually used a like text rapid D dent for something like this, but so I'm going to have to

9:58

run some testing to find out which one deals with stuff better clean doc. That might be better. Yeah,

10:04

that's cool. And then there's a partitioning. I don't use partition much, so I couldn't, I didn't

10:09

remember what that does. So partition, he was using partition to strip out comments. And what it does is

10:14

split a string on, on whatever you pass it in, in this case, the pound sign or, or hash or whatever

10:21

you want to call it. And then he used a, and then it splits that into three strings before the, before

10:28

the delimiter, what the delimiter was and after. And so this is, this is a way to just grab everything

10:34

before the comment, which is cool. And then he used, what else did I learn? Oh, I didn't understand what

10:40

this if not test is in here for first, but this was checking for blank lines, which makes sense or

10:46

stuff with just a comment. But then there's this eval, which I am so afraid of evals, but in this case,

10:54

you're writing the stuff it's from right here. So it's pretty safe, but eval the code and then passing

11:01

in a variable like the S variable into the code. So everywhere in this code, you're the S means the

11:08

string you're passing in, but this or G, what does or G do? And what he's doing is it's, it's a way

11:14

with eval to pass in imported stuff. So it imports, it'll import the regular expression and the counter

11:21

modules into the eval statement. So I didn't know you could do that. So I was thinking bitwise or

11:28

yeah. No, it's a way to get these imports in there, which is pretty cool. Anyway, that's fun.

11:34

And then he lists some other ways, but I was still frustrated that there's no pytest in here.

11:39

So I thought, how would I do this with pytest? So I just wrote up a quick blog post, actually just

11:46

praising Ned as well, because I learned some stuff and I like learning new things. But then what did I

11:53

do? I imported pytest, the same imports that he used, and then the, the same good and bad and tests.

12:00

But then I just split it up a bit different and used parameterization to write the test code. So

12:06

if anybody wants to use pytest instead, here we go. The excellent, the one fun thing that I had was this

12:13

10,000 character I'm using with parameterization. I was able to take the, the input as part of when you,

12:21

you do pytest dash fee, it prints out all this. It's the parameterization prints out, which is nice. But I

12:28

I didn't want 10,000 characters printed out. So I also got to use the ID function to shorten that up a

12:34

bit, shorted it to 20 characters. But anyway, testing.

12:38

Very nice. Yeah. I love the compare and contrast as well.

12:41

Yeah. Yeah. Awesome. Let's talk some trends, huh?

12:44

Okay. So this was originally just going to be an extra. The more I looked at them, like this actually could be a fun

12:49

conversation for you and me to have. So I wrote a article for the JetBrains blog called "The State

12:56

of Python in 2024." Okay.

12:59

I thought that'd be pretty fun. And there's eight key trends or whatever I pulled out. So I thought,

13:03

hey, you know, maybe, maybe that'd be fun to talk about the eight trends and get your thoughts on it.

13:08

All right. So let's do it real quick. So first of all, Python keeps growing. But it's interesting

13:15

that actually, if you look at the amount of other languages used along with Python, they're decreasing.

13:22

Right? So for example, in 2021, 40% of Python people did JavaScript plus Python. Now it's only 35.

13:28

If you look at Bash, 33% of people did. Now 29 do. And a lot of languages are going down like that,

13:35

except for Rust. So that's pretty interesting. And that's because so many people are coming into Python

13:42

from non-traditional programming languages like data science and other scientists and so on. You know,

13:48

GitHub just announced that, I think we covered this, that GitHub said that Python is the most popular

13:54

language on GitHub now. It's pretty awesome. But most of those people are coming from non-traditional

14:00

backgrounds, meaning they're just going into like Jupyter Notebooks or something like that, right?

14:05

Like this is their only Py programming language is Python.

14:07

Yes, exactly. Exactly. They're like, they just became programmers. So they're not doing the other

14:12

stuff. So it's interesting to see these like, hey, you're doing less JavaScript as a community,

14:17

which is kind of the opposite of what I would imagine. And very, very closely to what you said,

14:21

41% of Python developers have been working professionally in any language for less than two years.

14:27

Like almost everyone is new here. Yeah. And actually, I remember somebody talking about this, this, this question,

14:35

like maybe it was a different question, but like how long have you been a professional Python developer?

14:40

And a lot of people taking this survey don't think of themselves as developers. They think of themselves as just some other job. They happen to use programming also.

14:50

Yeah, exactly. So yeah. And then if you throw in three to five years at like, that's pretty much it.

14:54

Yeah. Trend three. Oh, let's go up to the years again. Where are we at?

14:58

Oh, just 11 plus. So we're just in one bucket of 13% of the people.

15:03

And even so that's only like over 10 years is basically only 13%, which is wild.

15:09

Yeah, that's really wild. Yeah, yeah, yeah. Where do people learn Python? They learn it on YouTube,

15:14

on our John codes, and coding. And if they listen to podcasts, talk Python to me, who's the number one out there.

15:21

Nice. But I'm sure Python buys us just under, just under. Trend four, Python two versus three,

15:29

that's over. We've raged on about the legacy Python long enough. There will be no 2.8 says Guido all

15:37

the way back in 2014. Right. But if you look at it, it's like asymptotically, like whoever's still

15:43

on Python two, they just, those people aren't leaving.

15:45

Yeah. It's what's surprising to me is the people that are still on that and are participating in

15:52

surveys. I think they just fully have checked out. I imagine a lot of those people, which is 6%

15:59

for this year that are still on Python two, that they're on some huge project. I know there's some,

16:05

they're not going to migrate the project. The project will not migrate, but those people would

16:09

very much like to, and they probably in their spare time work with him FastAPI and other modern things.

16:16

But when they go back to work, they're here. You know what I mean?

16:18

Yeah. Or at least some of their project is using that. Yeah.

16:21

Yeah, exactly. But still, let's keep going. All right. Flask, Django and FastAPI are all the top

16:29

three frameworks, which is nearly a dead heat, which is pretty interesting. We just talked a lot about

16:34

Django. But what's interesting is if you ask web developers who are Python people,

16:39

not just Python people, but we talked about there's a lot of people who don't consider themselves

16:43

like developers or whatever. But if you say, hey, web developers who are also Python people,

16:47

Django is used one and a half times as much as Flask or FastAPI.

16:54

Oh, yeah. So amongst the web developers, Django is clearly leading the pack. But amongst data scientists,

17:00

Flask and FastAPI are ahead because they're more about building APIs and getting their models online

17:05

and so on. So that's a pretty interesting difference, right?

17:08

Yeah. And I'm thinking there's a lot of data or data science stuff that doesn't have a back

17:13

database or anything or some huge thing. Yeah.

17:16

Yeah. Where do you host your stuff? It's all about the hyperscale clouds, apparently, which is

17:21

blowing my mind, actually. AWS, Google Cloud and Azure represent, gosh, how much did it say?

17:28

I wrote it out maybe at the top. Represent something like 78% of where people host their code. Yeah,

17:34

78% are on one of the three massive clouds. Wow.

17:37

Hetzner is showing up. Hetzner is there. Yeah. I mean, we put a dent in Hetzner.

17:41

We definitely did. We got one server there. Yeah. And also Heroku interestingly went down,

17:48

but I think that's because when they canceled their free tier, a lot of people stopped. They

17:52

decided to turn off their for move elsewhere. But Python Anywhere is going strong, actually,

17:57

above all the other small providers, right? DigitalOcean, Hetzner, Linode, and so on.

18:01

Python Anywhere. Interesting. Yeah. Yeah. That's pretty wild, right? All right. Two more. People prefer containers over VMs and

18:06

they prefer VMs over bare hardware. That shouldn't be a surprise, should it? I mean,

18:11

maybe the containers versus VMs, but certainly not just straight hardware. Straight hardware is not

18:16

even on the picture, by the way. It's like lower. And last trend of 2024 is UV takes the Python

18:22

packaging, takes Python packaging by store. Yes.

18:25

I think that's obviously it's, we've covered it a bunch of times. It's not just that it's fast.

18:30

It's that it combines a bunch of functionality from different tools into different places. Yeah.

18:35

It's super neat. It installs Python. It installs, it creates virtual environments. It manages projects,

18:42

if you wish. It updates. There's a lot of, a lot of good stuff there.

18:45

And anyway, I think a big part of the UV uptick is the UV team tried to make it so that you could use

18:52

it with your current workflow. You didn't really have to change your workflow.

18:55

Yeah. I think that is very much an important part of it. That's why I adopted it for sure.

18:59

So, yeah. So anyway, these are, these are my trends. Hopefully people find those interesting.

19:04

There's a lot of like writing and data to back that up in the link.

19:07

Right. Yeah. All right. Yeah. Yeah. Extra.

19:10

I've just got. We've come to extras.

19:12

I just got one extra toggling back to Django just for a moment is I noticed that Django admin has a

19:21

Dracula theme now under the Dracula theme.com. And I can't, I can't remember where I learned it from.

19:29

One of the maintainers posted this on, on, on blue sky, I think. But anyway, it looks great.

19:35

Great colors. And I also, because of this learned that there is a thing called Dracula theme.com that

19:43

contains like a whole bunch of different projects that all have Dracula theme. So.

19:48

Oh, nice. So let's see what I love it. Yeah. And it has cool bats. Oh no. No pad plus. That's funny.

19:56

Nice. VS Code. Cool. Yeah. Yeah. Obviously. Obviously. no visual studio. Oh, oh yeah. I guess people use visual studio still.

20:07

Um, anyway. Yeah. viewer on windows. Yeah.

20:09

Yeah. Jet brains. Of course. Nice. Cool. So yeah, I had no idea about this site. That's super cool.

20:14

That's fun. How about extras for you? I have a couple. Let's see here. First of all, my Zen browser experiment is

20:24

still going strong. I'm absolutely loving the Zen browser. It's based on Firefox, which I really like.

20:29

Okay. But that also means that you're limited to the Firefox limitations. Like I can't use it for our

20:35

live stream because our live stream only supports Chrome, which really means Chromium based browsers.

20:41

Right? So I, there's certain times that I'm not using it, but yeah, I'm enjoying people can check that out.

20:46

I feel, I talked to somebody and they're like, this is what Firefox should have built. This is like fire.

20:51

Why is Firefox not doing this? Yeah. You know, it's, it's a good question. I do think I honestly,

20:58

you know, as sort of a sidebar, I'm, I'm a little worried what's going to happen to Mozilla and

21:02

Firefox if the antitrust thing against Google goes through, right? Because if that goes through 90% of

21:09

Mozilla's revenue instantly gets declared illegal and cut off, not illegal for Firefox and Mozilla,

21:15

but illegal for Google. So they would have to stop. Right? Yeah. Maybe I'm not following that. So,

21:20

well, one of the big problems is Google has used their monopoly and their money and all that to buy

21:27

off locations to basically pay to either be the default or to prevent competition in different

21:33

ways. For example, they're paying something like $23 billion to Apple to be the default search engine,

21:39

not just in Safari on iOS, but in, if you go to Siri and you ask it a question, right? Like where does it,

21:45

it says, Oh, here, I'm going to search the web for you. Like, how do you think it's going to search

21:48

the web? Yeah, that's paid. That's okay. Right. But they're also paying, I don't remember the number,

21:54

but it's 90% of the revenue of Mozilla to support Mozilla. But I, you know, there's a lot of

21:59

thinking that that is to, to be the default search engine on Firefox. Right. But so that would make

22:05

it illegal for them to do that. Okay. It's the double whammy of like, they're kind of paying Mozilla to

22:10

keep Firefox around so they can say that there's competition. Yeah. Right. But if they can no longer

22:15

pay Mozilla to be the default search engine. Hmm. It's fishy though. If is there really competition

22:22

when you're only competitors paying you to be there? I know it's well, I mean, why is there a lawsuit?

22:29

Right? Yeah. So we'll, we'll see. Anyway. I think it's really cool. I'm enjoying it. Still a big fan of

22:35

Vivaldi as well, which is what I'm talking to you around on right now. Yeah. But also, did you know that

22:39

Microsoft has a browser? I heard of that. Oh, you know what? It's actually just based on Chrome.

22:43

Although Google doesn't pay for that one. I don't believe. Okay. I bet you get Bing as your default

22:48

search engine there. All right. refresh my desk setup. I just, I really, I'm enjoying this way more.

22:54

I set up a little separate table and a little separate computer. So I have a nice view. So I'm not

22:58

looking at like cameras and lights and junk all day. Oh, wow. Even if they're turned off,

23:03

I'm still just staring at a wall with like sound padding and stuff. And I enjoy it way more than

23:09

I realized. And I just want to encourage people. Like you're kind of frustrated with where you're

23:12

sitting, where you're looking while you're at work. Like maybe it's not that much work. It took a couple

23:16

hours. So I can look at trees instead of recording foam, which I encourage. All right. And while I was

23:21

sitting at this fun new desk, Brian, I added a really cool feature to our RSS feed. When I realized that,

23:28

there's an updated spec to RSS, which allows you to specify transcripts in subtitle format. So VTT,

23:36

web VTT or SRT files. Right. Okay.

23:39

So I added those to our set of transcripts to the website supplies to RSS feed. And now you get real

23:45

time follow along as you and I speak transcripts. Oh, wow.

23:50

And that cool. So you can just say, show the transcripts. And it's kind of like a Spotify or

23:54

YouTube music. It's just like follows along as we speak in the real transcripts. And I think you can

23:59

even search them. Although that's probably a per player type of thing. So this doesn't work in

24:04

overcast sadly, but it does work as far as I can tell in Apple podcasts and pocket casts. So at least

24:11

in those two, if you see a little transcript thing and you click it, that means it's going to follow

24:15

along in real time with our conversation. I wanted to have the little bouncy ball, like,

24:19

like, you know, the singalong movies when we're kids.

24:22

Boop, boop, boop, boop. Yeah, exactly. And, it works really good on an iPad. Like you can have

24:27

the, the view of the control and the art and everything. And then next to it, it has.

24:31

Oh, nice. Yeah. Yeah. But it's still also works on the phone. All right. that's it for my extras.

24:36

All right. Oh, I just wanted to add, it took five days to generate those transcripts with the computer

24:42

running 24 hours a day. So I regenerated these cause only, only maybe a third of our episodes had

24:50

VTT format. They had a different format. Just, it's like a pure text type thing that our website

24:55

understands and people can read. But VTT is more of us. It's like not JSON, but imagine like,

25:01

here's your JSON transcript to read, right? It would feel a little bit like that. Yeah. So I've always

25:06

thought of VTT is like some extra setting that is completely useless because I'd never use it. So,

25:12

um, so I'm glad that you are. You could use SRT. Those are the two that are supported here. Those are

25:16

like the two well-known subtitle formats or whatever. But yeah, I fired up. I have my old M1 Mac mini

25:22

laying around and I'm like, I'm just going to go over there and have it generate these for Python

25:27

bytes and talk Python five days later. It was done. And then I have a whole bunch of Python software

25:34

that goes through and like correct stuff. Like Ruff formatter is like R O U G H. I'm like,

25:41

no, not really. That's not what we meant to say. And it'll sometimes get your, you'll be Brian Aiken.

25:48

I'm like, Nope, we're going to fix that. And so on. Sometimes I'm Brian Aiken.

25:52

I don't know. Oh, we're good. Oh, I'm Aiken.

25:55

Yeah. So there's a lot of work that went in to make this happen. So hopefully people enjoy it. And

26:00

yeah, cool. Thanks for it. And I appreciate you keeping improving the back, the back end of Python

26:07

bytes. So yeah, you're welcome. No problem. Okay. You know what? If you were,

26:13

if you were working on software, you taught, you started this show off with testing, right?

26:17

Yeah. Well, if, if no, you start off with the jingle, but we did talk a lot about testing.

26:23

Second item. Anyway, if you were doing testing and you wanted to make sure your tests pass,

26:29

you have many options, Brian, you could not run the test, but if the tests are going to run and say

26:33

continuous integration, whoa, boy, you better make those tests pass or continuous integration is going to

26:38

fail. So Martin Bettman shared this joke with us on blue sky said, how about Volkswagen? You know,

26:48

they kind of gotten a little bit of trouble for writing their car software to say, yes, these diesels,

26:55

they're so clean. You should absolutely get clean diesel, not dirty diesel, get a Volkswagen. What

27:00

that really meant was while it's under test, it's going to change how it behaves to have a better

27:06

emissions test than not. Right? So this is riffing on that and says, Volkswagen, this is a thing you

27:12

can install into your apps is it detects when your tests are being run in a CI server and make sure they

27:17

pass. Okay, I got to check this out. That's hilarious. Yeah, it says why? Well, if you want your software

27:23

to be adopted by Americans, because that's where VW got caught for their cheating, test scores from the CI

27:31

server are very important. Volkswagen uses a defeat device to detect when it's being tested in CI server

27:37

will automatically reduce the errors to an acceptable level for the test to pass. This allows you to spend

27:43

less time worrying about testing and more time enjoying the good life as a trust, trustful software

27:48

developer. They even have a badge like a read me for your GitHub read me. Yeah, with a build passing with

27:57

a little Volkswagen symbol. That's great. Works for Travis CI, Circle CI, Jenkins, Hudson's Bamboo, Team

28:04

City, TFS, Visual Studio Online CI, GitLab, etc., etc., etc. And it defeats assert, tap, tape, and shy.

28:14

And any actual, any test actually that is set to exit code or throw an error. That's funny.

28:21

That's bad, right? I gotta check out and see all the ways they're doing it. Yeah.

28:24

I know. I was wondering like, how does this actually work? People don't really do this. It's a joke.

28:30

Well, so I've actually had cases where, especially during development, where I didn't want the CI to

28:38

jump out at the end of the test, even if there's failures. So there is a pytest feature where you

28:44

can change the exit code. It's a pytest custom exit code internal plugin or something.

28:50

pytest dash always exit zero?

28:53

Pretty much, yeah. To override the exit code so that you can debug the rest of the tool chain

29:00

or get data on something. But yeah.

29:02

That's the only time I've done that. But that's funny.

29:05

It's a little, the image or the logo of it or whatever is like a little Volkswagen bug from

29:11

the 70s that transforms into a transformer robot. I'm using.

29:15

Nice. There's 13 contributors to this. There's 15 releases. What is this?

29:20

How many stars?

29:23

Okay, okay. Let's find out. Oh, 15,000 stars.

29:27

People are using this thing. Oh my goodness.

29:30

Or just amused by it. It's been nine years old or something.

29:34

Yeah, it has been around for a while, but I ran across this and thought it's pretty funny.

29:38

So this is good. Yeah. Yeah. All right. Well, as always, thanks for being here. Thanks everyone for listening.

29:44

Thank you. Bye.