WEBVTT 00:00:00.020 --> 00:00:04.860 Hello and welcome to Python Bytes, where we deliver Python news and headlines directly to your earbuds. 00:00:05.280 --> 00:00:09.560 This is episode 449, recorded September 15th, 2025. 00:00:10.330 --> 00:00:11.240 And I am Brian Okken. 00:00:11.460 --> 00:00:12.160 And I am Michael Kennedy. 00:00:12.640 --> 00:00:16.660 And of course, this episode, not of course, but this episode is sponsored by us. 00:00:16.690 --> 00:00:21.560 So please check out the stuff we offer you guys and everyone. 00:00:22.200 --> 00:00:26.200 Courses at Talk Python Training, pytestCourses at Pythontest.com. 00:00:26.520 --> 00:00:28.940 And there's also one at Talk Python Training too. 00:00:29.580 --> 00:00:32.160 And then Patreon supporters, we love you. 00:00:32.480 --> 00:00:33.380 Thanks for supporting the show. 00:00:33.980 --> 00:00:40.120 If you'd like to have a suggestion for the show or like to get a hold of us, we're on Bluesky and Mastodon. 00:00:40.440 --> 00:00:41.300 Links are in the show notes. 00:00:41.839 --> 00:00:49.040 And as normal, this is currently Monday at 7 a.m., but normally it's Monday at 10 a.m. 00:00:49.280 --> 00:00:54.860 If you'd like to watch the show live, you can go to pythonbytes.fm/live and catch it. 00:00:55.040 --> 00:00:57.080 And you can even interact with us while we're recording. 00:00:57.760 --> 00:01:01.920 Finally, I'd like to thank everybody that's joined up for the newsletter. 00:01:02.200 --> 00:01:13.840 You can go to pythonbytes.fm and join the newsletter, and you get all the links that we talk about in the show, plus extra information on how to digest that, even if you're not listening at the time. 00:01:14.060 --> 00:01:15.180 We'll send that right to your inbox. 00:01:15.780 --> 00:01:18.360 So lots of cool stuff to talk about today. 00:01:19.140 --> 00:01:20.100 Let's kick it off, Michael. 00:01:20.480 --> 00:01:24.600 Yeah, let's kick it off with a little FOSS news, free and open source software. 00:01:25.200 --> 00:01:29.500 And this is, I don't know, maybe the story is kind of the opposite of FOSS. 00:01:29.940 --> 00:01:37.240 Giant trillion dollar company interacts with Monopoly court, but the consequences may fall down to open source software. 00:01:38.000 --> 00:01:53.320 And Judge Mehta from the Google antitrust monopoly case has ruled on the remedy for Google being a monopolist who abused their privilege in ad search, in ad space, in search and so on. 00:01:53.560 --> 00:02:04.140 So there were many different possibilities floated, even by the prosecution, like, you know, you probably need to sell Chrome and you may need to divest from Android, things like that, which would have been wild. 00:02:04.920 --> 00:02:09.060 I know that OpenAI and gosh, was it Cursor? 00:02:09.399 --> 00:02:11.160 Might have been Cursor or Anthropic. 00:02:11.460 --> 00:02:14.900 I can't, one of the, one of the AI companies was like, oh yeah, we'll take it. 00:02:15.080 --> 00:02:15.620 We'll buy. 00:02:16.180 --> 00:02:18.200 And I know perplexity threw their hat in there. 00:02:18.260 --> 00:02:18.900 We'll buy Chrome. 00:02:19.400 --> 00:02:23.080 One of them offered to buy Chrome for like 10 times their, their valuation. 00:02:24.080 --> 00:03:22.180 they're like we're gonna fire divestors we don't care but it turns out none of that is going to be that happened in an alternate reality or i don't know were you ever a fan to choose your adventure books brian yeah they went to page 11 instead of page 42 on this one and so there will be no selling of chrome there will be no divesting of android and the reason this is even on the show although all these things are fun to imagine is mozilla and firefox exist almost solely in their current form because some of the bad behavior of Google, right? Google, one of the big problems was Google's paying Apple something like 26, 20, 20-ish billion dollars a year just to say, yeah, make, make Google the default in Safari search on iPhone, right? But the other one was, for reasons I don't still fully understand, although I think they might not be 100% goodwilled, they are also paying 85% of Mozilla's bill every year to exist. 00:03:22.560 --> 00:03:22.860 Wow. 00:03:23.200 --> 00:03:25.460 Like they're 85% of Mozilla's revenue. 00:03:25.900 --> 00:03:31.580 So one of the remedies could be, you can't be paying all these companies to like make you the default search engine. 00:03:31.780 --> 00:03:36.240 Same reason, that's what Mozilla was getting is like default search engine on Mozilla equals Google. 00:03:36.540 --> 00:03:38.960 Here's $100 million a year or whatever it was. 00:03:39.300 --> 00:03:39.600 Okay. 00:03:40.060 --> 00:03:40.340 Go ahead. 00:03:40.580 --> 00:03:42.840 No, I was just looking at the notes. 00:03:43.900 --> 00:03:58.120 you wrote down that the google can keep paying mozilla to make google but only if it's if the deals are are not exclusive i don't understand how i believe what that means is um they let's say 00:03:58.660 --> 00:04:08.479 you know i'm a fan of like start page and kagi yeah i think they can't like say we'll pay you 100 million dollars and no one else can pay you anything to be like privileged or something along 00:04:08.500 --> 00:04:19.239 those lines okay because like only one person can be the default but yeah that's true maybe somebody could be like top on the list or something yeah i think something like that yeah i don't i haven't 00:04:19.239 --> 00:06:01.860 read the whole ruling right um the idea is that would be a totally reasonable remedy as part of this like selling chrome which maybe is a little less reasonable but still yeah i'm here for it it was so interesting okay um but a lot of people would a lot would have lost a lot of revenue like mozilla right exactly like i don't care if apple is revenue they can suck it up they they make more money than anyone and they're gonna get a bunch of my money when they ship new computers and ipads in a month or two so it's gonna be fine i'll backfill that for them but mozilla 85 the revenue you as a company with hundreds of employees you can't lose 85 of your income and not get it would just make mozilla fold i think yeah or maybe just focus i agree that it would probably make them fold in the short term maybe in the long term it just makes them focus on providing products so anyway this is there's a bit of a long conversation around just this but um i think you know i wrote i wrote the hey on the unsolicited advice from mozilla giving him like five business ideas that should be not just depend on google yeah um not that they necessarily need to take him or that it was like a big deal but there are certainly things that they could do and they should have started doing five years ago and maybe now is better than never. Because this is not Google promises to keep paying Mozilla's bill forever. This is the law will not make Google stop paying them. This deal is up for renewal in 2026. So it still could get canceled. I mean, when I said, I'm not sure about the motives of Google, if Google is paying them just because, hey, we definitely want Google to be the top search thing in your browser, Google could decide, you know, the market share is getting down and down and down. 00:06:01.880 --> 00:06:03.660 And maybe it just doesn't make sense to pay that anymore. 00:06:04.200 --> 00:06:04.500 It's done. 00:06:04.940 --> 00:06:05.220 Yeah. 00:06:05.520 --> 00:06:07.740 And that just seems icky also. 00:06:07.880 --> 00:06:14.580 It does seem like Google's just keeping Mozilla around so that they are not the, just to keep competition there. 00:06:14.940 --> 00:06:17.100 Just to not look like a monopolist, right? 00:06:17.420 --> 00:06:17.520 Yeah. 00:06:17.680 --> 00:06:17.860 Exactly. 00:06:18.420 --> 00:06:20.560 But now they've been punished as a monopolist. 00:06:20.840 --> 00:06:22.400 The fines have been determined. 00:06:22.540 --> 00:06:25.020 There are no more consequences, right? 00:06:25.480 --> 00:06:26.620 So do they still have to pay? 00:06:26.640 --> 00:06:30.160 I mean, no one's going to say, well, look, they're not a monopolist because of Firefox. 00:06:30.330 --> 00:06:32.760 They are legally declared to be a monopolist. 00:06:33.120 --> 00:06:36.720 So that check, that guard by doing that is gone. 00:06:37.800 --> 00:06:45.640 The market share of Mozilla is tweaking down enough that maybe it just, they are like, you know, it's not even worth like the vendor registration that we do. 00:06:45.740 --> 00:06:46.900 Like it could be a huge problem. 00:06:47.020 --> 00:06:51.120 It could also be that Google wants to show some goodwill and do this, but I don't know. 00:06:51.220 --> 00:06:52.220 It's still tenuous. 00:06:52.400 --> 00:06:53.500 This is not like, oh, they're fine. 00:06:54.740 --> 00:06:55.100 Interesting. 00:06:56.880 --> 00:06:58.580 - It's a super interesting thing to follow. 00:06:59.580 --> 00:07:01.260 We'll see where it goes, but yeah, 00:07:01.500 --> 00:07:05.480 they should probably hustle and start getting some things that are not being paid. 00:07:05.940 --> 00:07:22.040 And let's put a bow on it, but final thought is, I think you're right on point that it's icky and it's icky because when Mozilla makes choices about how to like defend users, like for example, you know, that fingerprinting, like not the ad blocking, but that creepy stuff where you ask, well, what fonts are installed? 00:07:22.260 --> 00:07:25.780 this JavaScript runtime has this speed behavior. 00:07:26.040 --> 00:07:27.480 So that means it's actually this person. 00:07:27.880 --> 00:07:30.720 It's down to like two people in the world and only one person visited our site. 00:07:30.800 --> 00:07:33.900 So we can track them like not just as good as a cookie. 00:07:34.340 --> 00:07:37.520 Like those kinds of things that would be lovely to have. 00:07:37.880 --> 00:07:38.020 Yeah. 00:07:38.300 --> 00:07:40.800 I actually don't know if Firefox has that or not, but I don't think so. 00:07:41.060 --> 00:07:42.060 I think Brave might. 00:07:42.320 --> 00:07:46.960 Anyway, there's exact opposite incentives like for Firefox. 00:07:47.220 --> 00:07:51.120 One is the users would be better off if they were actually a private browser that did that kind of stuff. 00:07:51.320 --> 00:07:56.080 Two, Google would be more upset with them and less likely to pay them their salary. 00:07:56.520 --> 00:08:01.140 That just puts them in a, hey, we could be for the users or we could be for our job. 00:08:01.900 --> 00:08:02.300 I'm not a fan. 00:08:02.580 --> 00:08:03.040 What a morning. 00:08:03.280 --> 00:08:10.540 We're going to go from lawsuits to trove classifiers. 00:08:11.020 --> 00:08:12.620 This will wake everybody up. 00:08:12.920 --> 00:08:17.540 Are you telling me that pytest-check has been declared a monopoly pytest plugin? 00:08:18.460 --> 00:08:20.080 No, I just wanted to pick one. 00:08:20.120 --> 00:08:35.400 So I wanted to pick a project on, so I've got, we've got pytest-check, one of the projects I work on, that is, I'm just to have the PyPI page up, because I wanted to show what some of the extra stuff, because it's relevant to Trove classifiers. 00:08:35.820 --> 00:08:43.880 So there's some meta information in the PyProject.toml, or maybe there was like a setup CFG or something like that. 00:08:44.000 --> 00:08:45.060 Anyway, don't use that. 00:08:45.180 --> 00:08:46.080 Use PyProject.toml. 00:08:46.340 --> 00:08:46.560 Anyway. 00:08:47.780 --> 00:08:48.740 Things like the license. 00:08:49.480 --> 00:09:18.980 on the left-hand side of uh of pi pi you can see things like the license uh who the maintainers are um the author require what python version it requires um and if it provides any extra installers i didn't know it did that that's pretty cool i know i knew that my package did i didn't know that pi pi said which extras you can install i didn't know that either that's interesting yeah um i below that there's extra things like i've got on here i've got that the development status is stable so so you can go ahead and use it. 00:09:19.700 --> 00:09:20.880 It's part of the pytest. 00:09:21.060 --> 00:09:23.420 It's a pytest plugin, so I have the pytest framework. 00:09:23.940 --> 00:09:25.200 Intended audience is developers. 00:09:25.460 --> 00:09:27.600 I don't know if that's important or not, but sure. 00:09:27.900 --> 00:09:28.960 And which Python versions? 00:09:29.160 --> 00:09:32.260 It's Python 3.9, 10, so I'll test it all the way up. 00:09:33.000 --> 00:09:33.600 3.15? 00:09:34.100 --> 00:09:34.940 I was going to take that out. 00:09:35.260 --> 00:09:37.080 But anyway, I haven't tested it. 00:09:37.100 --> 00:09:37.340 Leave it. 00:09:37.640 --> 00:09:40.120 It's only like three or four more weeks, and then it'll start to make sense. 00:09:40.240 --> 00:09:40.620 Just leave it. 00:09:41.140 --> 00:09:42.880 Yeah, I haven't tested 3.15 yet. 00:09:43.100 --> 00:09:46.000 But okay, so and then what topic? 00:09:46.240 --> 00:09:46.640 It's utility. 00:09:46.960 --> 00:09:48.200 all this extra stuff. 00:09:48.480 --> 00:09:49.640 Why do we care about that? 00:09:50.000 --> 00:09:55.540 Well, those things are listed in pipi.org slash classifiers. 00:09:56.140 --> 00:10:02.340 And there's a big list of these that you can put in there, even which CUDA version you're using. 00:10:03.220 --> 00:10:03.400 Interesting. 00:10:04.560 --> 00:10:06.580 A lot of them are legacy and don't make sense. 00:10:06.780 --> 00:10:10.140 But they have to be exact matches in your classifier list. 00:10:10.460 --> 00:10:11.380 Why am I bringing this up? 00:10:11.380 --> 00:10:16.100 Because I'm bringing it up because I'm blanking right now. 00:10:16.260 --> 00:10:21.080 Adam Hill wrote a project recently called troml, T-R-O-M-L. 00:10:21.410 --> 00:10:26.020 And troml provides a list of potential classifiers that could be added to a Python package. 00:10:26.380 --> 00:10:28.600 And actually, I think that's the original thing. 00:10:29.470 --> 00:10:32.700 It was to provide a potential list of classifiers. 00:10:33.180 --> 00:10:35.920 Now, if you're okay with it, you can say, go ahead and update it. 00:10:36.060 --> 00:10:39.660 It'll update your PyProjectDommel file with the right classifiers. 00:10:40.070 --> 00:10:42.360 And this is what they look like in code. 00:10:42.800 --> 00:10:46.660 There's a classifier list, and it's a string. 00:10:47.400 --> 00:10:52.960 It has to be exact spelling, exact capitalization with the spaces and everything. 00:10:53.560 --> 00:10:55.180 So you don't really want to do this by hand. 00:10:55.280 --> 00:10:59.700 I usually copied and pasted it, but now you don't have to. 00:10:59.720 --> 00:11:06.320 You can just run troml against your project, and you just give it the directory of your project, and it'll find your project troml file. 00:11:06.760 --> 00:11:08.660 And there's a couple arguments. 00:11:08.800 --> 00:11:12.480 You can say suggest, and it just suggests what you should put in. 00:11:12.700 --> 00:11:16.140 But if you add fix, it'll just change it. 00:11:16.440 --> 00:11:21.160 And if you want to just check stuff, you can say troml check. 00:11:21.320 --> 00:11:21.700 It'll check. 00:11:21.880 --> 00:11:22.440 That's really nice. 00:11:22.920 --> 00:11:23.080 Yeah. 00:11:23.900 --> 00:11:28.240 And that might be a good one for CICD or something, a pre-commit to check. 00:11:28.580 --> 00:11:32.420 So really cool that I don't have to think about that too much anymore. 00:11:32.560 --> 00:11:35.780 I can just run troml on my project, and it'll work. 00:11:35.840 --> 00:11:43.560 So the final thought on this is if you're like, why do we even care about classifiers? 00:11:43.880 --> 00:11:44.500 This is ridiculous. 00:11:46.120 --> 00:11:52.540 We have a testing code episode from 2023 that I'm linking to where I talk about to Brett Cannon. 00:11:52.780 --> 00:11:55.740 Because I, frankly, I just wanted to stop using him on my project. 00:11:56.160 --> 00:12:00.280 So I got Brett on the phone and said, hey, Brett, why do we need to care about these anymore? 00:12:00.680 --> 00:12:02.480 And he gave me a nice discussion. 00:12:02.840 --> 00:12:05.000 It's a half an hour discussion linked to that. 00:12:05.460 --> 00:12:05.600 Okay. 00:12:05.640 --> 00:12:06.500 Yeah, very nice. 00:12:07.480 --> 00:12:08.280 Anyway, classifiers. 00:12:10.340 --> 00:12:10.620 Love it. 00:12:10.920 --> 00:12:12.460 Yeah, and those things do look tricky. 00:12:12.800 --> 00:12:18.400 You almost feel like they should be an enumeration in some library like pip or something. 00:12:18.780 --> 00:12:23.160 You use a pip colon Django, pip colon Django find out to, you know what I mean? 00:12:23.740 --> 00:12:24.720 Well, yeah. 00:12:25.580 --> 00:12:26.260 One of the things I want to bring up. 00:12:26.260 --> 00:12:27.160 But it's by Project Toml, right? 00:12:27.380 --> 00:12:29.180 You can't run code in there. 00:12:29.820 --> 00:12:40.960 One of the things I wanted to point out that I was surprised when I first wrote this is I was putting licenses in my PyProject Toml classifiers. 00:12:41.559 --> 00:12:46.300 And because that used to be the only way you could get them in PyPI to show up. 00:12:46.560 --> 00:12:47.860 But you don't have to do that anymore. 00:12:48.279 --> 00:12:49.820 You don't have to list the licenses. 00:12:50.660 --> 00:12:51.620 So a little bit of news. 00:12:52.100 --> 00:12:52.400 Okay. 00:12:52.820 --> 00:12:53.120 Very nice. 00:12:53.680 --> 00:12:53.840 All right. 00:12:53.960 --> 00:12:54.860 What you got for us next? 00:12:55.480 --> 00:12:56.280 What is next? 00:12:56.320 --> 00:12:57.180 Well, let's see. 00:12:57.680 --> 00:12:58.080 pqrs. 00:12:59.540 --> 00:13:05.200 pqrs. I'm sure it's spoken out. It's not, it has no vowels. So if you're in data science, 00:13:05.920 --> 00:13:55.960 you, and let's say intersect two things. If you're in data science and you deal with large files, like large CSVs, right? Like let's say you've, you've got a 500 megabyte and a rather 500 gigabyte, I don't know, huge, huge CSV file for your project. And you start out your day by going to your notebook and you say pandas read csv or something equivalent of that right you're you're just directly working with the csv i guess even if this was like a web app might it still makes sense but the i think the real common usage scenario is someone is going around and working with their their data files right why do you keep it a csv there are better formats parquet for example which this applies to this pqrs is a command line tool for inspecting parquet files These are compressed, optimized binary versions of basically CSV files, right? 00:13:56.320 --> 00:14:04.100 I guess they're slightly different in the sense that like columns have types, like there can be an integer column, a little more like Excel, but Excel is hard to work with, right? 00:14:04.320 --> 00:14:05.260 Excel files are hard to work with. 00:14:05.560 --> 00:14:06.000 So CSV. 00:14:06.480 --> 00:14:08.120 Well, why might you keep it as CSV? 00:14:08.320 --> 00:14:11.500 Well, you want to be able to actually look at the file and see what the heck it is. 00:14:11.510 --> 00:14:15.440 These are like compressed binary things you can't look at unless you install this. 00:14:15.750 --> 00:14:18.600 And then you can just kind of browse around through the files. 00:14:18.900 --> 00:14:29.480 So you can go to your parquet file and say, show me the top entries by doing parquet.cat or head or merging them or do all sort of row count, sample some items out of it. 00:14:29.700 --> 00:14:38.120 So if you were avoiding working with these parquet files or something else, I think parquet is kind of one of the top ones out of the PyArrow project, out of the Arrow project. 00:14:39.000 --> 00:14:40.140 This is really cool. 00:14:40.410 --> 00:14:41.980 And it gives you back a little bit of that. 00:14:42.150 --> 00:14:47.280 Let me just poke around this binary file as if it were a CSV text file. 00:14:47.580 --> 00:14:51.420 So how about some reasons to use it? 00:14:51.920 --> 00:14:52.760 So I think there's two reasons. 00:14:52.850 --> 00:14:55.900 I think size and speed, both of which are relevant. 00:14:56.500 --> 00:14:57.880 You work with large data files, right? 00:14:58.160 --> 00:15:08.940 So in the show notes, I've got some things written down, like a typical 200 megabyte CSV file might be as small as 20 or 50 megabytes as a Parquet file. 00:15:09.779 --> 00:15:12.460 That's like a 5x smaller size. 00:15:12.920 --> 00:15:14.720 We're talking a bunch of huge data files, right? 00:15:14.960 --> 00:15:15.600 Smaller size, good. 00:15:15.740 --> 00:15:22.540 I mean, I guess if they're changing a lot, you know, a Git diff or whatever, but I mean, putting 500 gigs into a Git, it's going to hate it a lot. 00:15:22.880 --> 00:15:28.220 That I've never put a large file into Git, but it does complain about it. 00:15:28.660 --> 00:15:29.740 At least Git have complaints about it. 00:15:30.240 --> 00:15:32.900 And then the other thing is speed, right? 00:15:33.060 --> 00:15:37.600 So if you fully load like a big, obviously it depends on what the content is and so on. 00:15:37.660 --> 00:15:44.200 But on standard use cases, it also might be two to 10 times faster to read the Parquet file. 00:15:44.620 --> 00:15:46.260 And it's small, smaller. 00:15:46.780 --> 00:15:48.220 So really nice, right? 00:15:48.540 --> 00:15:48.740 Okay. 00:15:49.320 --> 00:15:51.380 Less memory usage, the way that it's read. 00:15:51.540 --> 00:15:53.220 Like there's a lot of advantages to having this. 00:15:53.440 --> 00:15:54.960 And Michael Booth out in the audience. 00:15:55.880 --> 00:15:56.240 No, sorry. 00:15:56.780 --> 00:16:04.600 Bart Williams points out that Parquet files also preserve D types, which is great for dates, date times, and US zip codes. 00:16:04.760 --> 00:16:23.860 indeed meaning like it knows it's a string and doesn't parse 0 1 7 7 2 as a 1 7 7 2 as a number right which is a big big problem and then um michael booth points out hey you could use duckdb for exploring perky files yeah i think this is more like i'm just cruising through the system and i 00:16:23.860 --> 00:16:29.180 would do a cat or just a tail a file just to see what the heck it is you know yeah that's pretty 00:16:29.160 --> 00:16:37.560 cool anyway pqrs check it out if you work with uh csv and other data cute name too it is do you 00:16:37.560 --> 00:16:55.080 know what time it is it's almost time for a new version of python checking out the python 314 or the pi python version um it's um um it it's coming up so looking at the dates i was looking at the PEP 745, which shows the dates. 00:16:55.380 --> 00:16:59.000 It looks like the release candidate is expected. 00:16:59.300 --> 00:17:03.580 The third release candidate is expected on the 16th, just tomorrow. 00:17:04.000 --> 00:17:04.079 Wow. 00:17:04.819 --> 00:17:10.380 And then the final is expected October 7th, but at least probably sometime in October. 00:17:11.559 --> 00:17:14.040 So why am I bringing this up? 00:17:14.100 --> 00:17:16.560 I'm bringing this up for a couple of things. 00:17:16.640 --> 00:17:19.540 I was going to talk about what's new in Python. 00:17:19.720 --> 00:17:21.439 So I'm going to link to it anyway. 00:17:21.680 --> 00:17:24.020 There's a what's new page with all the things. 00:17:24.300 --> 00:17:25.319 We've talked about some of these already. 00:17:26.040 --> 00:17:28.380 I'm not going to, there's a free thread. 00:17:28.500 --> 00:17:32.460 It is officially supported, but it's been in there for already. 00:17:33.440 --> 00:17:34.300 Template strings are neat. 00:17:34.900 --> 00:17:36.180 Anyway, cool list of new things. 00:17:36.280 --> 00:17:44.880 But what do I really want to talk about today is I want to talk about that I really wanted pytest-check, which we were already talking about, and the rest of my projects. 00:17:45.220 --> 00:17:46.300 I thought, you know what? 00:17:46.580 --> 00:17:48.820 I'm not completely testing them yet. 00:17:49.040 --> 00:17:49.940 I am testing them. 00:17:50.120 --> 00:17:54.660 I added them to talks recently, so I would test everything locally. 00:17:54.940 --> 00:18:01.720 But I wasn't testing 3.14 in CI, and I thought, now's the time to turn that on. 00:18:02.120 --> 00:18:03.060 But I had trouble. 00:18:03.160 --> 00:18:06.560 I forgot the magic incantation of, what do I do? 00:18:06.560 --> 00:18:13.520 Do I just add 3.14 to my GitHub actions list or the Python matrix? 00:18:13.920 --> 00:18:14.460 Tried that. 00:18:14.640 --> 00:18:15.080 Didn't work. 00:18:15.600 --> 00:18:24.240 So I asked people on Master on Bluesky, and I'm always grateful for everybody that answers things rather quickly. 00:18:24.310 --> 00:18:25.280 Got a bunch of cool answers. 00:18:26.780 --> 00:18:32.060 A few people suggested that I forgot the, what was the answer? 00:18:32.260 --> 00:18:35.680 I forgot the use pre-releases. 00:18:36.720 --> 00:18:47.380 And in particular, Ed Rogers said, hey, there's a great article from Hugo about running 3.14 in CI. 00:18:48.000 --> 00:18:48.960 So I'm going to link to that. 00:18:49.320 --> 00:18:53.060 It's called Free Threaded Python on GitHub Actions. 00:18:53.340 --> 00:18:55.280 So that doesn't say how to run 3.14. 00:18:55.560 --> 00:18:56.600 It's about free threading. 00:18:56.880 --> 00:18:58.780 But hidden in here is the secret. 00:18:59.220 --> 00:19:00.720 The secret is down here. 00:19:01.160 --> 00:19:06.560 And there's a bunch of great examples of YAML files and workflow files. 00:19:06.820 --> 00:19:12.320 But the magic keyword is right down in here that says allow pre-release is true. 00:19:12.720 --> 00:19:12.920 That's it. 00:19:13.000 --> 00:19:14.240 You really, that's all you got to do. 00:19:14.360 --> 00:19:16.520 So you got to say allow pre-release is true. 00:19:16.980 --> 00:19:24.040 And of course, while you're there, it's good to check to make sure that you have a current setup Python version. 00:19:24.360 --> 00:19:27.020 And I believe this is already out of date. 00:19:27.060 --> 00:19:30.400 I believe that we've got maybe the, I don't know, check them. 00:19:30.460 --> 00:19:37.840 So I think we're up to V5 and V6, but it's good to check on your actions, the checkout action and Python action. 00:19:38.680 --> 00:19:41.140 But at least you have to be up to five, I think, for this to work. 00:19:41.460 --> 00:19:43.200 Anyway, pre-release is true. 00:19:43.440 --> 00:19:52.700 And then while you're there, you can check it out and say, hey, I am going to test against 3.14, but maybe I should test against the free-threaded version also. 00:19:52.940 --> 00:19:56.440 And all you have to do is add a 3.14T to make that work. 00:19:57.220 --> 00:19:57.720 All right. 00:19:57.960 --> 00:19:58.520 Yeah, that's very cool. 00:19:59.380 --> 00:20:02.520 And yeah, thanks to everybody that helped me out on that one. 00:20:03.200 --> 00:20:03.380 Indeed. 00:20:03.610 --> 00:20:04.180 Yeah, that's super cool. 00:20:04.260 --> 00:20:06.980 And pretty soon you can drop the pre-release thing, right, in a month. 00:20:07.520 --> 00:20:09.200 Well, but you don't really have to. 00:20:09.370 --> 00:20:12.620 It can just be in there to test stuff. 00:20:12.800 --> 00:20:12.880 Yeah. 00:20:13.360 --> 00:20:13.520 Exactly. 00:20:13.800 --> 00:20:14.640 It's not going to hurt anything. 00:20:14.900 --> 00:20:18.120 And you'll need it for 3.15 right around the corner. 00:20:18.640 --> 00:20:19.000 Exactly. 00:20:19.050 --> 00:20:21.760 So you're almost ready to test your Trove classifier. 00:20:22.180 --> 00:20:24.660 I mean, who doesn't want to do like some deadlocks on GitHub Actions? 00:20:24.820 --> 00:20:25.020 Come on. 00:20:25.260 --> 00:20:25.880 Put the T in there. 00:20:27.200 --> 00:20:29.080 I don't think we're in danger of deadlocks. 00:20:29.500 --> 00:20:38.520 I think we're in danger of a lot of race conditions and data corruption because deadlocks assume that people are already writing locks using thread, thread locks, just using them wrong. 00:20:38.910 --> 00:20:39.860 I think no one's using them. 00:20:40.620 --> 00:20:40.880 Yeah. 00:20:41.310 --> 00:20:43.020 So I don't know. 00:20:43.220 --> 00:20:48.720 So I turned on testing on free threaded for pytest, but I don't know if it makes a difference because it's a pytest plugin. 00:20:48.960 --> 00:20:50.000 And I don't think that. 00:20:50.720 --> 00:20:51.140 Yeah. 00:20:51.300 --> 00:20:52.600 It's probably fairly stateless. 00:20:52.730 --> 00:20:52.840 Yeah. 00:20:53.220 --> 00:20:56.100 Well, I don't think pytest utilizes the free threading yet. 00:20:56.240 --> 00:20:59.820 It'd be kind of cool if it did, so we could test multiple things at once. 00:21:00.140 --> 00:21:01.060 But I think it's a single test. 00:21:01.060 --> 00:21:05.420 Yeah, that's true, because it's still subject to the gil, unless it runs it in multiprocessing. 00:21:05.760 --> 00:21:13.980 Yeah, I mean, you can run them in parallel, but traditionally people use Xdist, pytest Xdist, which splits them off into separate processes. 00:21:15.180 --> 00:21:15.420 All right. 00:21:15.820 --> 00:21:16.980 Now are we ready for extras? 00:21:17.480 --> 00:21:18.900 Yeah, we're at extras now. 00:21:19.000 --> 00:21:19.260 All right. 00:21:19.880 --> 00:21:24.180 For something completely different, let's talk about an article by Hugo van Caminande. 00:21:26.400 --> 00:21:26.580 Yeah. 00:21:27.620 --> 00:21:36.880 So you spoke about PREC last time, which is a alternate, faster UV-enabled pre-commit, right? 00:21:37.200 --> 00:21:41.000 We talked about pre-commit in different ways intersecting with uv, and that was one of them. 00:21:41.480 --> 00:21:47.500 And said, hey, by the way, I also wrote, by the way, it comes from the episode, I'm getting the BIOS flavor. 00:21:47.700 --> 00:21:48.960 It's got to be one of the better titles. 00:21:49.460 --> 00:21:49.820 So good. 00:21:50.419 --> 00:21:58.600 Anyway, the article is Ready PREC Go, talking about how he's using Prec for pre-commit type of things. 00:21:58.840 --> 00:21:59.860 So super cool. 00:22:00.220 --> 00:22:00.980 People can check that out. 00:22:01.500 --> 00:22:07.420 And this was not on my follow-up earlier and my extra earlier, but now I realize I should throw this on you. 00:22:07.620 --> 00:22:10.540 So remember I said that it was quite a bit slower 00:22:10.710 --> 00:22:15.140 to run my little URL crunching utility in Python 00:22:15.630 --> 00:22:17.420 instead of, I can't remember what we were talking about. 00:22:17.440 --> 00:22:20.760 You're talking about something, oh, the startup cost of Prec actually it was. 00:22:21.120 --> 00:22:25.900 So I'm like, yeah, even my little super silly utility is kind of slow. 00:22:26.110 --> 00:22:29.580 Like it takes almost a second to just start and do its little thing. 00:22:30.000 --> 00:22:31.420 Well, well, Michael learns a lesson. 00:22:31.980 --> 00:22:41.680 The lesson is that was something that came out of the Talk Python code base, which has like packages installed that have a whole bunch of stuff in there. 00:22:41.830 --> 00:22:56.540 And I think somehow loading up that sub package of that whole code base was loading like almost all of the talk Python stuff as it just parsed, doing all the imports to find its way to what it needed to like put a few pieces together. 00:22:57.220 --> 00:23:01.000 And I may have broken that out and it may run way, way, way, way faster now. 00:23:01.400 --> 00:23:11.540 So lesson, keep it, if you put it in its own little virtual environment and you make it do its own, if it's like a one or two line utility or one or two file utility, maybe that makes a big difference. 00:23:11.800 --> 00:23:12.400 At least it does here. 00:23:12.580 --> 00:23:15.160 So yeah, I thought that was actually really interesting that it makes a big difference. 00:23:15.560 --> 00:23:17.660 like 10 times different startup speed. 00:23:17.900 --> 00:23:18.480 Wow. Yeah. 00:23:18.680 --> 00:23:20.880 Yeah. Yeah. So there I've learned a lesson. 00:23:21.300 --> 00:23:43.600 Also just before I call it on my extras, I was listening to some security podcasts and apparently there was a super broad, although not super bad, supply chain attack at npm where some package, some maintainer of a package got phished by a, Hey, you've got to make sure your 2FA is up to date innpm. 00:23:43.620 --> 00:23:52.020 And they did, but that was actually logging into a site that was completely not that, which gave the hackers access to their npm repository, their package. 00:23:52.390 --> 00:24:04.740 And they replaced, they called it a one-function, one-liner dependency that 500 other packages depend upon that resulted in something like 2 billion downloads a month or a week. 00:24:04.810 --> 00:24:05.120 I don't know. 00:24:05.120 --> 00:24:06.300 It was like a lot of downloads. 00:24:06.650 --> 00:24:10.880 And they put in cryptocurrency stealers where the one line used to be. 00:24:11.060 --> 00:24:16.440 Just thinking about like, do you necessarily need a one line dependency that has other dependencies? 00:24:16.730 --> 00:24:19.080 Like, could you just maybe just vendor that in? 00:24:19.080 --> 00:24:25.300 So just, I definitely think not depending on too many things has many good advantages to you. 00:24:25.380 --> 00:24:39.840 And this age of AI, if you can ask AI to say, write me a single function that does this thing and it solves your use case, like maybe you don't need any dependencies and side benefit is like you code might start a little bit faster, but also people can't fish someone downstream and do bad stuff too. 00:24:39.920 --> 00:24:41.380 So anyway, very interesting. 00:24:41.980 --> 00:24:42.520 Starts much faster. 00:24:42.920 --> 00:24:42.980 Cool. 00:24:43.600 --> 00:24:44.100 All right, over to you. 00:24:45.740 --> 00:24:47.880 I kind of, I've got a lot of extras, 00:24:48.320 --> 00:24:49.020 so I'll try to get through. 00:24:50.780 --> 00:24:55.400 So Adam Hill, we talked about troml or yeah, troml. 00:24:55.830 --> 00:25:01.740 So Adam Hill also has another Toml utility and it's DJ Toml settings. 00:25:01.850 --> 00:25:11.420 I haven't played with this yet, but essentially it's, you can save your Django settings in Toml files, either into pyproject.toml or you can make a Django Tommel file. 00:25:11.860 --> 00:25:17.900 Anyway, he's got a new package called DJ Tommel settings that you can store Django stuff in Tommel. 00:25:17.930 --> 00:25:18.920 So that's that. 00:25:19.960 --> 00:25:23.780 A couple of people notified us about this, and I noticed it also. 00:25:24.220 --> 00:25:27.560 Somebody named Sam Henry Gold, did you see this? 00:25:28.440 --> 00:25:37.460 He wrote a little utility that allows you to detect. 00:25:39.120 --> 00:25:47.520 It allows you to detect the angle of your Python, or not your Python, your MacBook, the lid, and then react to it. 00:25:47.780 --> 00:25:49.560 So he's got a couple things that are fun. 00:25:49.740 --> 00:25:55.860 He's got a creaky door sound that you can play on your MacBook while you're opening and closing the lid. 00:25:56.080 --> 00:25:58.180 And then a, what did he call this? 00:25:58.840 --> 00:25:59.460 The theremin. 00:26:00.660 --> 00:26:02.180 It's a different kind of instrument. 00:26:04.860 --> 00:26:06.540 I saw this, thought it was fun. 00:26:07.020 --> 00:26:13.200 And then we got, somebody told us, it's called Lid Angle Sensor. 00:26:13.530 --> 00:26:14.480 And then- 00:26:14.480 --> 00:26:15.660 Brian Weber. 00:26:15.930 --> 00:26:16.060 Yeah. 00:26:16.210 --> 00:26:16.660 Yeah. 00:26:16.660 --> 00:26:22.540 So Brian Weber notified us that there's a version of this for Python already. 00:26:22.540 --> 00:26:26.380 This just came out like a week ago or so, called PyBookLid. 00:26:26.380 --> 00:26:29.820 So just directly from Python, you can control your book lid. 00:26:29.820 --> 00:26:32.200 You don't have to get anything else. 00:26:32.250 --> 00:26:33.060 All right. 00:26:33.060 --> 00:26:36.480 So who's going to be the first person that takes this to the Apple store? 00:26:37.520 --> 00:26:51.460 and tells them they need to oil their macbook lid at the genius bar i mean look listen to the thing and then if you put it too far like a crunch if it like goes oh my god you broke it yeah 00:26:52.100 --> 00:28:01.600 pretty funny don't do that okay uh what do we got not pie test check again i've already done that um no these are my oh i was just debugging my uh my stuff so uh the last thing i wanted to bring up was pie bay so pie bay is just around the corner also uh pie bay is coming in october october 18th of 2025 just so um middle of october if you are planning on if you're already in san francisco um or or the bay area or plan on being there in october or would like to go and um don't want to buy your own ticket so i encourage people to go buy your own ticket this is a really cool there's a lot of great people there talking we've got Guido Pamela Fox Christopher Lisa Jim there's there's tons of great speaker profiles I kind of wish I could go I'm not gonna be able to get there Burnett is going to talk about Python packaging a lot of great speakers anyway so check out PyBay and I've got a secret for everybody so if if you would like to go but for some reason would just like to get a free ticket, just let us know. 00:28:01.640 --> 00:28:02.480 So how do you let us know? 00:28:02.820 --> 00:28:08.600 I think that'd be cool to do a drawing from our email list. 00:28:08.880 --> 00:28:16.400 So go to pythonbiz.fm, sign up for our email newsletter, and we're going to draw how many tickets do we got? 00:28:16.840 --> 00:28:18.460 They gave us some tickets to give away. 00:28:18.880 --> 00:28:20.300 I've got one, two, three, four. 00:28:20.400 --> 00:28:22.320 I've got four vouchers, so four people. 00:28:23.120 --> 00:28:32.980 We'll randomly pick before next week, we'll randomly pick four people and we'll email you and say, hey, do you want a ticket? 00:28:33.250 --> 00:28:34.420 And we'll give out some tickets. 00:28:35.080 --> 00:28:40.360 Also, it looks like it's a voucher for one individual ticket and a free conference T-shirt. 00:28:40.540 --> 00:28:41.480 So you get the T-shirt also. 00:28:41.840 --> 00:28:42.160 So that's cool. 00:28:42.410 --> 00:28:43.580 So when are we going to do that drawing? 00:28:43.680 --> 00:28:44.420 I'll probably do it. 00:28:44.540 --> 00:28:45.200 I'll do it next Sunday. 00:28:45.500 --> 00:28:47.000 So sign up this week. 00:28:47.320 --> 00:28:51.440 Sign up before Sunday, actually by the end of Saturday. 00:28:51.860 --> 00:28:53.640 And then I'll draw it next week. 00:28:54.290 --> 00:28:54.380 Sweet. 00:28:54.890 --> 00:28:55.700 Yeah, that's a fun conference. 00:28:56.020 --> 00:29:02.040 So, and it's a whole new set of people, too, since the last time I went of running it. 00:29:02.310 --> 00:29:02.900 So, yeah. 00:29:03.840 --> 00:29:03.960 Yeah. 00:29:04.620 --> 00:29:05.280 How about something funny? 00:29:05.560 --> 00:29:06.820 I've got something good for you. 00:29:07.160 --> 00:29:07.460 All right. 00:29:08.100 --> 00:29:11.980 Imagine, Brian, you are junior in high school or something. 00:29:12.160 --> 00:29:14.240 Now, I guess the joke, it's got to be a little bit older. 00:29:14.540 --> 00:29:16.720 College age, but decided not to go to college. 00:29:17.220 --> 00:29:22.660 You're on your first time meeting your potential newish girlfriend's father. 00:29:23.030 --> 00:29:25.660 They're a little upset that he decided to go into this computer thing. 00:29:26.060 --> 00:29:27.320 They don't know if it's going to work out. 00:29:27.660 --> 00:29:29.560 And they wish that he had gone to college. 00:29:30.060 --> 00:29:30.960 So you meet the father. 00:29:31.140 --> 00:29:34.560 And the father says, my daughter told me that you're a console developer. 00:29:34.800 --> 00:29:37.940 So PlayStation, Xbox, and just know what? 00:29:39.320 --> 00:29:42.680 The young guy wearing a hoodie says, I console.logster. 00:29:44.320 --> 00:29:46.940 You have exactly 10 seconds to get out of my house. 00:29:50.900 --> 00:30:02.580 yeah console developer love it that's what i got for you um of course comes from uh programming humor which is doing a good job with the memes yeah i think we need to like redo that one look um 00:30:03.100 --> 00:30:11.399 uh what's that uh some cli stuff right yeah like a print print to print debugger yeah very good so 00:30:11.660 --> 00:30:14.600 I heard you're a console log, a console developer. 00:30:14.900 --> 00:30:15.080 Amazing. 00:30:15.660 --> 00:30:16.420 Console.log. 00:30:18.180 --> 00:30:18.280 Yeah. 00:30:18.520 --> 00:30:19.600 I like that he is sir. 00:30:19.980 --> 00:30:21.220 I console.log, sir. 00:30:25.240 --> 00:30:25.360 Yeah. 00:30:26.419 --> 00:30:29.360 Do you pronounce parens in spoken English? 00:30:30.899 --> 00:30:31.700 I guess. 00:30:31.800 --> 00:30:32.240 I don't know. 00:30:32.580 --> 00:30:32.640 Yeah. 00:30:33.300 --> 00:30:33.720 Open close. 00:30:34.440 --> 00:30:34.560 Yeah. 00:30:34.660 --> 00:30:35.020 I don't know. 00:30:36.880 --> 00:30:37.180 All right. 00:30:37.820 --> 00:30:39.120 Well, awesome. 00:30:39.400 --> 00:30:40.200 Oh, wait. 00:30:40.260 --> 00:30:41.300 Real time follow up, by the way. 00:30:41.720 --> 00:30:43.600 Brian Weber says thanks for mentioning his extra. 00:30:43.700 --> 00:30:44.400 So very good. 00:30:44.900 --> 00:30:45.800 And thanks, Brian. 00:30:45.940 --> 00:30:46.460 And thanks to everybody. 00:30:46.600 --> 00:30:50.300 We had quite a show of people showing up at 7 a.m. 00:30:50.740 --> 00:30:51.120 Interesting. 00:30:51.680 --> 00:30:55.560 We got some from Sydney, Michael Booth. 00:30:55.900 --> 00:30:59.220 Hey, thanks for being live in a friendlier time slot than usual. 00:30:59.620 --> 00:31:02.020 So maybe we should do an early one once in a while anyway. 00:31:02.960 --> 00:31:03.680 Yeah, absolutely. 00:31:03.800 --> 00:31:04.300 Mix it up a little. 00:31:04.500 --> 00:31:05.560 I mean, how about this? 00:31:05.740 --> 00:31:07.360 How about we just get a random number generator? 00:31:07.440 --> 00:31:11.100 It generates a number between zero and 24 every day. 00:31:11.600 --> 00:31:12.660 Every week, we just hit that. 00:31:12.660 --> 00:31:14.360 And whatever hour that is, that's when we record. 00:31:14.560 --> 00:31:15.320 I hate that idea. 00:31:15.480 --> 00:31:16.100 It's 3 a.m. 00:31:16.460 --> 00:31:16.840 You hate it? 00:31:17.200 --> 00:31:18.340 I don't like that idea. 00:31:19.080 --> 00:31:20.200 I honestly hate it too. 00:31:21.220 --> 00:31:24.040 But I'm glad some people got to attend. 00:31:24.860 --> 00:31:25.120 All right. 00:31:25.220 --> 00:31:26.760 Well, thanks, everybody, for listening. 00:31:27.120 --> 00:31:28.760 And we will talk to you next week.