WEBVTT 00:00:00.020 --> 00:00:12.900 Hello and welcome to Python Bytes, where we deliver Python news and headlines directly to your earbuds. This is episode 488, and it is Tuesday, July 14, 2026. I'm Michael Kennedy. 00:00:13.180 --> 00:00:15.120 And I'm Calvin Hendryx-Parker. Happy Bastille Day. 00:00:15.500 --> 00:00:27.760 Yes, indeed. Happy Bastille Day to you. And this episode is brought to you by us, so check out our things. Courses over at Talk Python. I have many in the works, competing for what comes next, maybe Rust. 00:00:28.110 --> 00:00:30.540 I think a Rust course is going to be next out the gate. 00:00:30.720 --> 00:00:33.440 So we'll talk more about that at some point soon. 00:00:33.800 --> 00:00:40.020 If you have a project and you would like some expert help for the really hard projects, reach out to Six Feet Up. 00:00:40.240 --> 00:00:42.080 You can subscribe to our newsletter. 00:00:42.700 --> 00:00:44.240 Got some nice announcements there. 00:00:44.800 --> 00:00:50.460 Follow us on all the socials and watch us live or the replays on YouTube. 00:00:50.640 --> 00:00:53.160 Every episode has a little YouTube thumbnail. 00:00:53.290 --> 00:00:54.520 If you click it, guess what you get? 00:00:54.740 --> 00:00:55.760 YouTube, YouTube version. 00:00:56.320 --> 00:00:59.500 So we appreciate the people who are part of the live show, but definitely not required. 00:00:59.900 --> 00:01:05.239 And with that, Calvin, I would say, I want to talk about the Eno Such blog. 00:01:06.120 --> 00:01:07.760 Yes, who do you trust? 00:01:08.140 --> 00:01:08.520 It's interesting. 00:01:08.680 --> 00:01:13.720 This is actually a pair of blog posts that both came out last week on the 7th. 00:01:14.140 --> 00:01:17.140 One from William Woodruff, that's this blog right here. 00:01:17.450 --> 00:01:20.220 And the other one's actually from Brett Cannon, which I'll show here in a second. 00:01:20.480 --> 00:01:24.460 But they're both around trusted publishing and who you should trust. 00:01:25.300 --> 00:01:34.500 What I think is important to understand, and I think what they both emphasize in their posts, is that trusted publishing is not for you. It is for the machines. So if you're not familiar 00:01:34.600 --> 00:01:48.459 with trusted publishing, it is PyPI's OIDC-based auth scheme. So you can now, it's used by other groups too, NPM, RubyGems, Crates, and Nougat. They basically want to replace long-lived API 00:01:48.480 --> 00:01:59.720 tokens with short-lived auto-scoped credentials tied to the CI/CD machine identity. So this may sound really great and like you could trust everything coming out of these tools, but really 00:01:59.900 --> 00:02:11.760 what it's about is making sure places the software is coming from and going to is trusted. And you can actually see that in this post, William goes into why you shouldn't trust trusted publishing 00:02:12.160 --> 00:02:22.700 alone. And then the other post is actually from Brett Cannon, but that was actually quite interesting, which is how to publish to PyPI using GitHub Actions securely. And he actually 00:02:23.060 --> 00:02:27.740 mentions Zizmor, how do you pronounce that? Is that the right pronunciation of the tool? 00:02:28.200 --> 00:02:33.060 So William wrote the tool mentioned in here, and he also mentioned in his own post. 00:02:34.000 --> 00:02:45.080 All right, the security episode I just did with the Python security folks, Mike Fiedler and Seth and Juanita, they called it Zizmor, I believe? 00:02:45.220 --> 00:02:46.340 We'll go Zizmor. We'll go Zizmor then. 00:02:46.920 --> 00:02:47.520 I believe. 00:02:48.340 --> 00:02:51.000 Mike would know, and Mike will correct me momentarily. 00:02:51.040 --> 00:02:51.340 I'm sure. 00:02:51.340 --> 00:02:53.360 But you've got to trust me. 00:02:53.480 --> 00:02:56.060 I mean, speaking about trust, you've got to trust that I'm remembering this correctly. 00:02:56.060 --> 00:02:57.080 Right, they're remembering correctly. 00:02:57.720 --> 00:03:03.300 But this is a tool that will help you clear, make the CI/CD actions happy. 00:03:03.340 --> 00:03:16.040 So once you make Zizmor happy, it's going to reduce scopes of tokens and double check for things like the old API tokens being in existence and what their scopes may be. 00:03:16.520 --> 00:03:23.420 It'll also make sure that when you do a checkout in GitHub Actions, the credentials that are used for that checkout act as you. 00:03:23.700 --> 00:03:27.360 And then you can make sure that things like those remain short-lived. 00:03:27.600 --> 00:03:30.040 And so that those don't persist throughout the rest of that run. 00:03:30.800 --> 00:03:38.060 Keeping the building of your package or your software separate from the publishing of your package and the software. 00:03:38.200 --> 00:03:52.560 So keeping those kinds of concerns separate from one another helps you make sure that these kinds of attestations can be made and have some level of assurances that you're doing the right things, that things are going to the right places, 00:03:52.790 --> 00:04:07.500 that no one's exploiting something they shouldn't be. So it's a little bit of like least privilege wrapped around the ability to establish trust between the machines that are building and the machine that would be publishing it. If you want to see an example of this, 00:04:07.920 --> 00:04:09.680 You can look at the Zismore package. 00:04:10.260 --> 00:04:15.740 You'll see this is a Williams package that does the kind of work you need to make sure that you're doing the right practices. 00:04:16.359 --> 00:04:19.120 You'll see this verified details over here on PyPI. 00:04:19.579 --> 00:04:26.300 Again, this does not mean that this package is of a known quality, or you should definitely include it as a dependency into your system. 00:04:26.800 --> 00:04:39.720 It just means that these are verified and have come from trusted sources, that the person who is listed here as maintainer has some level of control or has exercised some proof of ownership over these pieces, 00:04:40.320 --> 00:04:45.640 you'll also notice in PyPI that there are unverified details like the homepage and documentation in this specific case. 00:04:45.720 --> 00:04:54.320 So you can look for those as levels of quality that you want to check into when you want to look into including a package like this. 00:04:54.700 --> 00:04:57.780 But don't confuse trusted with trustworthy. 00:04:59.280 --> 00:05:05.860 Make sure you have human judgment in the loop about the package you're getting ready to bring in to your potential project. 00:05:06.370 --> 00:05:13.160 This is no substitute also for securing your own CI pipeline or actually vetting the packages that you install. 00:05:13.420 --> 00:05:19.840 So stay diligent out there, but this is another great way to keep things tidy, secure, exercise some least privilege. 00:05:20.320 --> 00:05:23.260 If something does happen, it can help contain the blast radius. 00:05:23.590 --> 00:05:32.740 If something were to sneak into your CI pipeline and try and exploit something that it shouldn't, hopefully you've narrowed down the scopes for the OIDC tokens so that they couldn't do those actions. 00:05:33.100 --> 00:05:33.880 Sounds good to me. 00:05:34.220 --> 00:05:34.320 Yeah. 00:05:35.160 --> 00:05:38.460 William Woodruff is, by the way, from the Astral team. 00:05:38.880 --> 00:05:39.000 Yes. 00:05:39.580 --> 00:05:40.980 And uv and all that. 00:05:41.180 --> 00:05:43.300 So they've thought a little bit about this. 00:05:43.640 --> 00:05:43.820 Yeah. 00:05:43.940 --> 00:05:47.700 So, I mean, table stakes, just use this tool no matter what. 00:05:48.280 --> 00:05:54.220 It's going to give you at least some level of clarity around your current state of operations for the CI/CD pipeline. 00:05:54.620 --> 00:05:54.780 Yeah. 00:05:54.920 --> 00:06:00.820 I asked at the end of the Talk Python episode was about there's a whole security track at PyCon. 00:06:01.120 --> 00:06:06.000 I said, let's go through that, all the talks there and kind of like talk about the arc that that's telling about Python security, right? 00:06:06.300 --> 00:06:06.400 Yeah. 00:06:06.820 --> 00:06:11.420 And at the end, I said, what is one thing people can do to be better at Python security? 00:06:11.590 --> 00:06:13.820 I think two out of three people said Sismar. 00:06:14.060 --> 00:06:14.200 Yeah. 00:06:14.430 --> 00:06:14.900 Just use it. 00:06:15.220 --> 00:06:15.340 Yep. 00:06:15.760 --> 00:06:24.720 And Brett Cannon's post here is a really just a straight walkthrough of the three or four things you should just be doing out of the box by default. 00:06:25.120 --> 00:06:26.040 So follow his instructions. 00:06:26.270 --> 00:06:28.820 He knows a lot of this stuff, you know, better than we all do. 00:06:29.220 --> 00:06:34.660 I trust Brett and Brett will make sure you get trusted publishing in place, but don't just trust it blindly. 00:06:35.110 --> 00:06:35.260 Okay. 00:06:35.800 --> 00:06:35.960 Amazing. 00:06:36.340 --> 00:06:36.900 Michael, what you got? 00:06:37.260 --> 00:06:42.140 I want to talk about what I thought would be just a small little extra at the end. 00:06:42.140 --> 00:06:42.840 Oh, hey everyone. 00:06:43.360 --> 00:06:46.980 New release, point release, Jupyter Notebooks and JupyterLab. 00:06:47.360 --> 00:06:47.440 Nope. 00:06:47.720 --> 00:06:47.820 Nope. 00:06:47.980 --> 00:06:48.920 It is a lot. 00:06:49.100 --> 00:06:52.220 And let me just jump to the end here and I don't need to see a pop up. 00:06:52.380 --> 00:06:52.720 That's fine. 00:06:53.639 --> 00:06:57.480 It says, as large as minor releases get, basically. 00:06:58.860 --> 00:06:59.720 So Jupyter Lab. 00:07:00.000 --> 00:07:01.160 Pumped right up against the ceiling. 00:07:02.000 --> 00:07:02.260 Exactly. 00:07:02.380 --> 00:07:06.380 They're like, we almost called it a new release, but we didn't want to update the docs dropdown. 00:07:08.660 --> 00:07:19.100 So it says this Jupyter Lab release includes 68 new features, 97 bug fixes, 38 documentation improvements, 95 contributors, and 171 maintenance tasks. 00:07:19.540 --> 00:07:19.960 Very minor. 00:07:21.080 --> 00:07:25.060 I would just call this, like, you know, consider it a pretty big release, let's say. 00:07:25.340 --> 00:07:26.220 So there's a couple of things. 00:07:26.400 --> 00:07:33.920 And I think probably the most efficient way is to just pull up this little diagram that they've annotated of Jupyter's JupyterLab. 00:07:34.260 --> 00:07:35.840 By the way, you can test this out online. 00:07:36.880 --> 00:07:39.980 And there's a test it in JupyterLite. 00:07:40.100 --> 00:07:41.220 Are you familiar with JupyterLite? 00:07:41.540 --> 00:07:41.820 No. 00:07:42.100 --> 00:07:46.180 So if we go to JupyterLite, available in JupyterLite. 00:07:46.180 --> 00:07:47.760 Let me click that and just see what happens. 00:07:48.140 --> 00:07:49.540 I'm glad that that opened a new window. 00:07:50.100 --> 00:07:57.620 So JupyterLite is basically a WASM WebAssembly powered Jupyter and uses local storage and stuff for it. 00:07:57.720 --> 00:08:00.660 So like I'm pulling up here and there you have it. 00:08:00.800 --> 00:08:03.240 Like full on JupyterLab, but running JupyterLite. 00:08:03.500 --> 00:08:04.480 Like in a browser, right? 00:08:04.760 --> 00:08:05.160 In the browser. 00:08:05.580 --> 00:08:06.300 That is pretty sweet. 00:08:06.680 --> 00:08:07.020 Yeah, yeah, yeah. 00:08:07.160 --> 00:08:09.620 Like full client side, nothing but just a download there. 00:08:09.900 --> 00:08:10.880 Well, I think that's a smart move. 00:08:11.660 --> 00:08:12.040 Yeah, yeah. 00:08:12.100 --> 00:08:13.020 So you can test this out. 00:08:13.260 --> 00:08:17.000 Bring your Jupyter notebook to your data wherever it lives. 00:08:17.400 --> 00:08:17.780 Exactly. 00:08:18.180 --> 00:08:19.420 It's all powered by Pyodide. 00:08:19.700 --> 00:08:21.200 So you can test these changes out there. 00:08:21.220 --> 00:08:22.280 But let me pull up the little diagram. 00:08:22.740 --> 00:08:28.120 And it says you have a new activity bar options, like files or extensions or whatever. 00:08:28.480 --> 00:08:29.460 There's a breadcrumb. 00:08:29.640 --> 00:08:33.340 So like on the left in JupyterLab, you've got a file browser. 00:08:33.440 --> 00:08:38.219 So you can open Python or IPYNB files. 00:08:38.599 --> 00:08:40.580 And now there's a breadcrumb that you can edit. 00:08:41.180 --> 00:08:47.520 macOS should take this to heart, please, someday, maybe, that you can just type up there. 00:08:48.320 --> 00:08:49.580 And it'll say, you're in this directory. 00:08:49.620 --> 00:08:51.240 and you put your cursor there, and guess what you can do? 00:08:51.510 --> 00:08:54.180 You can go to a subdirectory or another folder. 00:08:55.040 --> 00:08:56.100 Very cool, so that's there. 00:08:57.300 --> 00:08:59.660 I guess it was missing that it didn't show the creation date. 00:08:59.930 --> 00:09:03.440 It only showed modified in Files before, so now you can figure out the creation dates. 00:09:03.760 --> 00:09:09.120 You can move the sections in that whole left panel, and I guess the right one around as well. 00:09:09.480 --> 00:09:12.700 You can take stuff out of the sidebar, like widgets out of the sidebar. 00:09:12.900 --> 00:09:17.240 That's kind of cool, and make them more top level, so you can create a feeling that you might like this. 00:09:17.680 --> 00:09:20.660 Calvin, you can create this like custom layout here. 00:09:21.040 --> 00:09:22.180 Yeah, I do like my tiles. 00:09:22.640 --> 00:09:27.700 Yeah, it's got a kind of a VS Code like dropdown when you hit play for the debugger. 00:09:28.030 --> 00:09:29.680 So that kind of drops in right in the middle. 00:09:29.840 --> 00:09:32.620 So it's easy to understand your debugging and chase that around. 00:09:33.280 --> 00:09:35.800 You can add hotkeys from the editor. 00:09:36.170 --> 00:09:39.500 It has a cell number count in the debugger. 00:09:39.660 --> 00:09:42.600 You can like filter out your different sources and interact with the kernel. 00:09:43.160 --> 00:09:47.240 And you can go forwards and backwards in history of your edits. 00:09:47.640 --> 00:09:56.880 So one of the things that's kind of chaotic, you know, chaotic good, I guess, of character of Jupyter is that you can edit and execute things in different orders. 00:09:57.440 --> 00:09:57.520 Yeah. 00:09:58.920 --> 00:10:02.440 It's mind bendy and will mess most new people up for sure. 00:10:02.540 --> 00:10:03.160 Yes, exactly. 00:10:03.480 --> 00:10:07.180 It's like, oh, go to is bad, but maybe you could just manually go to at whim. 00:10:07.700 --> 00:10:13.160 Anyway, one of the features is you can jump forwards and backwards in terms of last edits. 00:10:13.680 --> 00:10:18.340 So maybe you edited cell seven and then cell 22 and then 20. 00:10:18.760 --> 00:10:20.780 So you can jump around in that through like some UI. 00:10:21.240 --> 00:10:23.280 So it's like a history order of like where you've been. 00:10:23.640 --> 00:10:25.140 Yeah, it's like a back forward button. 00:10:25.540 --> 00:10:29.240 Now if I could just have something similar in Slack to know where the heck I've been in Slack. 00:10:29.520 --> 00:10:30.040 That's fair. 00:10:30.420 --> 00:10:31.020 That's totally fair. 00:10:31.700 --> 00:10:33.100 Let me just flip through, see what else there is. 00:10:33.180 --> 00:10:34.060 There's a bunch of other things. 00:10:34.070 --> 00:10:35.300 Like there's only some of them, right? 00:10:35.420 --> 00:10:37.020 That's just the stuff they had annotated. 00:10:37.200 --> 00:10:41.400 The terminal is a little more, how shall we say, coding agent friendly. 00:10:41.800 --> 00:10:43.700 So you can open stuff in the terminal. 00:10:44.740 --> 00:10:46.380 Oh, we should get nerd fonts. 00:10:46.450 --> 00:10:48.300 I don't know if you can put nerd fonts in there or not. 00:10:48.960 --> 00:10:51.560 But there's a way they say it. 00:10:51.700 --> 00:10:52.480 That's not it. 00:10:52.960 --> 00:10:53.020 Yeah. 00:10:53.370 --> 00:10:57.100 So the terminal no longer traps keyboard focus, which is interesting. 00:10:57.420 --> 00:10:58.200 And here we go. 00:10:58.340 --> 00:10:59.120 This is the one. 00:10:59.440 --> 00:11:04.300 Pressing shift enter in the terminal now inserts a new line without executing the current line. 00:11:04.620 --> 00:11:08.880 This matches the behavior expected by certain terminal applications. 00:11:09.260 --> 00:11:10.540 I wonder which ones those are. 00:11:11.340 --> 00:11:11.960 It's a... 00:11:12.040 --> 00:11:15.200 Pi, Claude Code, et cetera, et cetera, right? 00:11:15.440 --> 00:11:27.100 The annoying thing is that some of the web UI versions of those tools uses like command enter or control enter or shift enter instead of a standard shift enter. 00:11:27.440 --> 00:11:33.320 I had to fix that for some folks around here who wanted the consistency between OpenAI and Anthropic tooling. 00:11:33.760 --> 00:11:34.100 That's fair. 00:11:34.440 --> 00:11:34.560 Yeah. 00:11:35.220 --> 00:11:36.800 So I think I'm going to leave it there. 00:11:36.880 --> 00:11:40.100 There's a bunch more features, as you can see, and I kind of called out and so on. 00:11:40.140 --> 00:11:45.160 But yeah, it's if you notebook and especially if you do stuff with Jupyter, a lot of changes there. 00:11:45.480 --> 00:11:45.840 People can. 00:11:46.040 --> 00:11:49.900 I mean, it looks like they got a ton of new features, like a lot of like creature comforts. 00:11:50.440 --> 00:11:55.440 That's pretty exciting for folks who are living day in and day out in the Jupyter notebook, Jupyter server world. 00:11:55.620 --> 00:12:01.160 But let's talk about another AI agent that got released very, very recently. 00:12:01.300 --> 00:12:02.800 This is probably within the last two weeks. 00:12:03.200 --> 00:12:08.760 The fine folks at Hugging Face have released Tau, which is two times pi. 00:12:09.480 --> 00:12:14.040 and it is a agent built mimicking a lot of pies. 00:12:15.699 --> 00:12:22.280 I think they were inspired by the fact that how minimalistic pie was, but they wanted pie to do some other things. 00:12:22.850 --> 00:12:29.220 And most of it being written in Python, watching it and teach you how it is working under the cover. 00:12:29.250 --> 00:12:39.280 So part of the goals with this coding agent is actually to have small readable layers, have it output back to you what it is working on, what it's doing. 00:12:39.900 --> 00:12:51.420 A lot of that gets kind of, if you've ever used Claude Code or even Pi, you'll see a lot of thinking dot, dot, dot, or, you know, conjugulating, whatever the verb of the randomizer put in there for you. 00:12:51.880 --> 00:12:54.280 But it kind of hides what's going on under the covers. 00:12:54.400 --> 00:12:55.800 But here, this is actually the opposite. 00:12:56.000 --> 00:12:57.500 They're trying to know hidden machinery. 00:12:57.840 --> 00:12:59.660 Every moving part is put on the page. 00:13:00.200 --> 00:13:02.740 I think it's kind of, it's pretty, I tried it out yesterday. 00:13:03.520 --> 00:13:04.880 The models become streams. 00:13:04.920 --> 00:13:17.340 you basically want to build an agent loop. So tau, it's going to be hard not to say pi, tau is made up of three pieces, the tau AI, the tau agent, and the tau coding. And any of these 00:13:17.500 --> 00:13:30.179 pieces are able to be swapped out. So the idea is you could make your own front ends for the tau coding part of this, build your own TUI, CLI, shell scripts, tools, et cetera, that are on the front 00:13:30.200 --> 00:13:40.660 of that. What's nice is every part is visible. If you've used one of the things I liked about Claude Cowork is it on the sidebar inside of Claude Cowork, it shows you like the to-do list it's 00:13:40.760 --> 00:13:55.719 putting together for the actions it's going to take. It shows you which skills have been picked up and used, which connectors got pulled into the context window. This one goes a step further and shows you what tools are currently active in the current window. If there's anything else 00:13:55.740 --> 00:14:07.260 interesting down through here. One thing I wanted, I'll just, let me just show it. I've got it running over here. I will unshare this bit here and I'll put my terminal in here. This will be 00:14:07.500 --> 00:14:13.900 obviously live real-time potential danger, but it's kind of pretty. I'm here for it. I'm here for it. 00:14:13.900 --> 00:14:22.300 I saw many people say, oh, you should never do demos. I'm like, you know what? No, that makes it real. Let's do it. Yeah. Seeing on the left-hand bar here, you've got details about the session, 00:14:22.420 --> 00:14:27.040 what provider you're currently running, what level of thinking, number of tools, number of skills. 00:14:27.480 --> 00:14:40.840 I think this is important because a lot of this gets hidden under the covers when you're using Claude Code. Unless you go and investigate and kind of ask it specifically what it's using in its current context window, I like having this available because then you could see something get out of 00:14:40.960 --> 00:14:50.660 hand. If there's all of a sudden brought in 20 tools and the context window is exploding and you're not sure why you're getting back poor results, this would surface that sooner for you. 00:14:51.060 --> 00:14:52.600 You can see what skills are currently loaded. 00:14:52.900 --> 00:14:55.260 Actually, that skill I wrote as part of this session. 00:14:55.640 --> 00:15:02.240 So I came in and I said, let's build a memory system for my AI agents. 00:15:02.540 --> 00:15:07.980 And so the memory system, then I kind of asked it for an explanation about what a memory system would be. 00:15:08.480 --> 00:15:09.060 And you can see it. 00:15:09.200 --> 00:15:15.600 It's really more like reading a textbook than it is having it code up the just the code for you. 00:15:15.700 --> 00:15:17.380 So it really gives you full examples. 00:15:17.620 --> 00:15:20.220 I'm getting a little bit of notebook vibes. 00:15:20.320 --> 00:15:25.780 actually it's got a little bit of like a storytelling UI. That's the goal is for it to tell 00:15:25.830 --> 00:15:37.160 you as it goes teach you even here it doesn't need to output all this to do its work it's outputting all this to help you understand and that was me asking it about dictionaries but then I went in 00:15:37.220 --> 00:15:47.279 and was like okay from the dictionary let's let's start a new here it is the memory piece right here we've got the idea of semantic memory episodic memory and procedural memory and so it explains 00:15:47.300 --> 00:15:53.200 each of those terms to you, what the goals are, you know, what the memory retrieval loop would be. 00:15:53.850 --> 00:16:04.220 And this was just the explanation. At some point, then I said, okay, great. Build me a simple memory system that understands episodic, semantic, and procedural memory. So then it goes into planning. 00:16:04.760 --> 00:16:17.860 It looks at what's already inside the system, created the whole first couple of modules, created some documentation, and then it shows me how to run it. At some point, I'm like, oh, updateagents.md to make sure we're always using uv. So if you want to check out Tau, 00:16:18.290 --> 00:16:26.560 you can uv install, tool install tau-ai, and that will get you the Tau CLI ready for you to use. 00:16:27.320 --> 00:16:39.160 It is, again, very early days. It is only a couple weeks old, but I'm a learner at heart, so I love the idea of having this explained to me so that I understand what's going on. I can make 00:16:39.160 --> 00:16:48.840 better decisions about the code and I can have fun playing at the same time. So this actually built the skill, installed the skill into the system, and now is using its own skill for memory. So again, 00:16:49.040 --> 00:16:58.840 that essence of Pi where you build your extensions, you make your coding environment the way you want it to be. That's what they're trying to take, but also with more of the explainer going on behind 00:16:58.860 --> 00:17:13.000 the scenes too. So pretty cool. It actually reminds me a little bit of Hermes, which I talked about last week in the sense that it's kind of self-improving and you can sort of change around the pieces honestly this inspires me a little bit more than Claude Code the terminal version well and 00:17:13.060 --> 00:17:27.260 that's the piece i didn't do yet is like in the docs and talk about it is take this to the next level by establishing what your harness looks like and then you create a loop you add the looping part to it to schedule and do work so i was one step away from i've got memory now i can now create the 00:17:27.300 --> 00:17:42.140 loop and have it set it loose on tasks yeah very neat very cool i see you got yeah you've got gpt 5.5 yeah have you tried 5.6 soul or any of these i have not yet no i haven't had a chance to um i i did 00:17:42.170 --> 00:17:47.420 the most basic of basic thing i said slash login open ai and that's what it gave me like let's go 00:17:48.580 --> 00:17:53.720 perfect well it definitely definitely looks really cool yeah so thanks to the fine folks over hugging 00:17:53.960 --> 00:17:58.700 face they're doing still doing great work over there year after year love it yeah also i was 00:17:58.760 --> 00:18:11.680 gonna say like you said it has a kind of like just getting started vibes and so on and i would be a little bit, not concerned, but just hesitant to go all in on something that's like a side project of someone's, but hugging face. 00:18:12.180 --> 00:18:16.240 Yeah, and it's already got like 1500 GitHub stars. 00:18:17.060 --> 00:18:18.020 I think I like it. 00:18:18.120 --> 00:18:18.460 It's beautiful. 00:18:18.820 --> 00:18:19.740 It does look nice. 00:18:19.800 --> 00:18:21.400 It shows me all the information I want to see. 00:18:21.880 --> 00:18:22.380 That's important. 00:18:22.560 --> 00:18:25.060 I'm about the experience of my workstation. 00:18:25.360 --> 00:18:26.700 So this definitely falls into that. 00:18:27.180 --> 00:18:36.560 Yeah, and I think the other thing is I really like how it shows you and what it's doing and how it's thinking because it's so easy to just make these things become an easy button, turn your brain off and go. 00:18:36.830 --> 00:18:39.760 But on the other side, it can be an incredible learning thing. 00:18:39.800 --> 00:18:40.420 You're like, what? 00:18:40.980 --> 00:18:42.880 Okay, I don't make, that is actually new to me. 00:18:43.000 --> 00:18:43.880 Tell me what that does. 00:18:43.960 --> 00:18:44.740 Why are you doing this? 00:18:44.810 --> 00:18:45.320 You know what I mean? 00:18:45.420 --> 00:18:47.980 If you don't know what's happening, how do you know to ask that? 00:18:48.200 --> 00:18:49.820 Well, it's funny because I asked about memories. 00:18:50.380 --> 00:18:54.060 And so it actually looked into its memories and obviously says user prefers uv. 00:18:54.480 --> 00:18:58.680 And then I love to execute Python from uv and I created a simple memory system. 00:18:59.060 --> 00:19:01.440 So it's already using its own memory system they built. 00:19:01.720 --> 00:19:07.320 with two prompts. That's awesome. Yeah. Super cool. Very cool. All right. If you're going to have a 00:19:07.800 --> 00:19:13.200 some kind of agent, you probably want to run it in the background. So let's talk about Django. 00:19:13.660 --> 00:19:28.319 I just had a really cool episode with Carlton Gibson over on Talk Python about, he basically rewrote Django's async documentation saying it's gotten really out. It was really out of date and it had a lot of warnings like, oh, you shouldn't do this or this won't perform that well. And they're 00:19:28.340 --> 00:19:42.900 actually like, you know what? That's not true. The stuff is really good now. You should probably use async in Django, but our docs tell people they should be weary. So let me rewrite that. And part of that conversation, we talked about Django's new task framework, which is part of version 6, 00:19:43.440 --> 00:19:57.160 6.0. So, so often you have got to think like, all right, well, I'm working on some requests and maybe it's slow. So I could use async and await on it potentially, but a better way often is like people don't actually even need the response. 00:19:57.590 --> 00:19:58.720 They don't need to wait on this. 00:19:58.840 --> 00:20:00.840 Like I'm going to send you a reset email. 00:20:01.500 --> 00:20:04.660 Email already is oddly asynchronous and non-verifiable. 00:20:04.850 --> 00:20:09.080 So what are you going to, you're going to send an email and like what, wait for a pixel to open before you respond? 00:20:09.140 --> 00:20:10.000 Like you can't do that, right? 00:20:10.000 --> 00:20:11.940 You just got to just send it and let it go. 00:20:12.110 --> 00:20:18.080 And so a lot of times people would do something like Celery or RabbitMQ or some other mechanism to run. 00:20:18.270 --> 00:20:19.220 Which is heavy, yeah, very heavy. 00:20:19.480 --> 00:20:19.940 Yeah, exactly. 00:20:20.220 --> 00:20:28.980 It's like you've gone from running like a single worker and a platform as a service to Now you've got a DevOps and a virtual private network for your servers to live in. 00:20:29.540 --> 00:20:31.280 It's like a huge jump, right? 00:20:31.740 --> 00:20:35.960 So this task framework kind of attempts to solve that in a framework-y way. 00:20:36.260 --> 00:20:41.380 What you can do is it's not exactly a task execution story. 00:20:41.540 --> 00:20:49.180 It's more of a way to like, I guess, like an adapter or a facade type of design pattern that says, here's how you program tasks in Django. 00:20:49.280 --> 00:20:54.900 the task backend that you stick in it means maybe it still goes to Celery or maybe it runs in a background thread. 00:20:55.160 --> 00:20:55.540 I don't know. 00:20:56.400 --> 00:20:57.040 We'll figure that out. 00:20:57.100 --> 00:20:58.960 You configure the thing in different ways. 00:20:59.240 --> 00:20:59.480 All right. 00:20:59.680 --> 00:21:01.220 So basically that's how it works. 00:21:01.540 --> 00:21:02.620 You set up your tasks. 00:21:02.840 --> 00:21:04.080 You define the backend. 00:21:04.180 --> 00:21:07.500 There's only one that comes with it and it's oddly unsatisfying to me. 00:21:07.860 --> 00:21:13.220 It's the immediate execution backend, which means like when you enqueue a task, it just blocks and executes. 00:21:13.260 --> 00:21:15.040 It goes, okay, it's enqueued when it's done. 00:21:15.980 --> 00:21:20.440 And really that's just for execution as a developer so you don't have to set up anything. 00:21:21.040 --> 00:21:26.820 I'll tell you what I wish was actually in there for development and for like 75% of all deployments. 00:21:27.100 --> 00:21:28.700 There's also the dummy backend. 00:21:29.880 --> 00:21:34.060 They missed a chance at like the null backend, like the null pattern. 00:21:34.220 --> 00:21:38.060 Like here's the thing you don't have to check for null or see if it's configured, but it just does nothing, right? 00:21:38.320 --> 00:21:39.560 So this is for like testing. 00:21:40.000 --> 00:21:42.580 So you can enqueue things and so on and like, you know what? 00:21:42.900 --> 00:21:43.920 No, nothing happened. 00:21:44.980 --> 00:21:50.700 Basically like a built-in way to say turn off tasks, which is, I would imagine, really just for testing. 00:21:51.480 --> 00:21:55.160 Anyway, so there's third-party backends, and there's an ecosystem here. 00:21:55.160 --> 00:21:56.320 It also has async support. 00:21:56.680 --> 00:21:58.060 Yeah, that's what I was going to say. 00:21:58.360 --> 00:21:59.340 Yeah, that's pretty neat. 00:21:59.800 --> 00:22:01.280 So a lot of cool stuff here. 00:22:02.180 --> 00:22:06.700 You can just put the at task decorator on a function, and it's now a task and so on. 00:22:07.020 --> 00:22:08.860 This is so nice compared to what it used to be. 00:22:08.960 --> 00:22:20.240 So Django has had this level of ability, like with Django channels and async and workers, backend workers, but the usability of it as a developer was always tricky. 00:22:20.660 --> 00:22:22.180 Like it wasn't very well published. 00:22:22.240 --> 00:22:23.860 It wasn't, it was hard to use. 00:22:23.860 --> 00:22:28.760 You had to do a little bit of, we did this back in 2020 when we first launched the Loudswarm platform. 00:22:29.420 --> 00:22:35.100 We had to use background async workers like that to establish connections, WebSocket connections to things like Discord. 00:22:35.700 --> 00:22:38.540 And it was some gymnastics to make it work. 00:22:38.700 --> 00:22:39.940 This looks so much nicer. 00:22:40.840 --> 00:22:49.200 So thank you all to whoever put in the effort to make this usability and the API, the usability as a developer for this to be so nice. 00:22:49.460 --> 00:22:50.660 Yeah, it looks really, really nice. 00:22:50.670 --> 00:22:52.880 And then the community stuff is good. 00:22:53.180 --> 00:22:59.740 Yeah, Loudsworn being your online conference stuff that you launched at a very timely time of 2020. 00:23:01.400 --> 00:23:01.580 Definitely. 00:23:02.080 --> 00:23:10.580 So for 75% of the people, I would recommend, I would point them at this project by Lincoln Loop called Django Tasks Local. 00:23:11.100 --> 00:23:16.940 And the idea here is it uses either threading or processing backends. 00:23:17.840 --> 00:23:22.860 No Celery, no Redis, no database, no configuration necessary, right? 00:23:23.060 --> 00:23:23.860 So this is really cool. 00:23:23.860 --> 00:23:27.760 You basically just plug this in, and then when you run it, it just runs in a background thread. 00:23:28.220 --> 00:23:31.180 And so it really does immediately return, and it really does do these things. 00:23:31.540 --> 00:23:39.820 The one thing that you've really given up here, I guess, depending on what you choose in Threadpool, until you do free-threaded Python, you're still subject to the guild. 00:23:40.240 --> 00:23:44.340 But most of these background tasks are not heavily computational. 00:23:44.960 --> 00:23:45.900 They're talking to a database. 00:23:46.320 --> 00:23:47.800 They're talking to a SMTP. 00:23:47.960 --> 00:23:48.620 There's some kind of weight. 00:23:49.080 --> 00:23:50.260 Yeah, it's not a lot of CPU. 00:23:50.480 --> 00:23:56.780 It's mostly probably IO weight, or you just need to be unbounded and get a quick response back to your end users. 00:23:57.360 --> 00:23:59.460 Or you have startup tasks. 00:23:59.720 --> 00:24:00.900 That was the thing we needed it for. 00:24:01.060 --> 00:24:02.460 We needed something to happen at Django startup. 00:24:03.120 --> 00:24:03.600 Oh, interesting. 00:24:03.890 --> 00:24:04.000 Okay. 00:24:04.540 --> 00:24:05.380 So you just fire that off. 00:24:05.620 --> 00:24:08.560 Yes, you could choose process backend if you really want like true separation. 00:24:08.790 --> 00:24:11.040 The one thing that you're giving up here is durability, right? 00:24:11.220 --> 00:24:17.800 Like if you deploy a new version and the task thing is grinding through some tasks and Docker says, time for a new one, goodbye. 00:24:20.160 --> 00:24:27.660 Yeah, well, you go into it knowing that I's wide open and you account for, I'm okay if it dies, but I know how to handle recovery. 00:24:27.960 --> 00:24:28.340 Yeah, sure. 00:24:28.780 --> 00:24:33.580 And how most projects, how much back end traffic do they really have? 00:24:33.860 --> 00:24:33.920 Right. 00:24:34.020 --> 00:24:36.440 They're sending an email because you said reset my email or something. 00:24:36.640 --> 00:24:41.680 But yeah, most most sites are, you know, a thousand people at your company use this internal thing. 00:24:41.740 --> 00:24:44.240 And that would be a big deal, you know, or something like that. 00:24:44.300 --> 00:24:44.400 Right. 00:24:45.260 --> 00:24:47.300 It's not it's not Netflix scale or. 00:24:47.840 --> 00:24:47.960 No. 00:24:48.040 --> 00:24:50.020 But when you need celery, you know, you need celery. 00:24:50.400 --> 00:24:54.840 And then you can get into more durable cues and retry patterns and those kinds of things. 00:24:55.300 --> 00:24:55.460 Right. 00:24:55.480 --> 00:24:56.920 Or temporal or something like that. 00:24:56.920 --> 00:24:57.060 Yeah. 00:24:57.440 --> 00:24:57.600 Absolutely. 00:24:57.940 --> 00:24:58.020 Yep. 00:24:58.400 --> 00:24:59.120 That's awesome. 00:24:59.450 --> 00:25:00.020 Yeah, cool. 00:25:00.150 --> 00:25:03.360 So check out Django Task and Django Task Local. 00:25:03.740 --> 00:25:04.060 All right, Calvin. 00:25:04.580 --> 00:25:06.440 I got one extra for you. 00:25:07.920 --> 00:25:18.500 The current release manager for Python 3.14, they've always been looking for fun little puns to do on the fact that this is the Py release of Python. 00:25:19.130 --> 00:25:22.120 But actually, it ended up becoming a fix in the dictionary. 00:25:22.640 --> 00:25:24.820 And Michael, it's not the dictionary you're thinking of. 00:25:25.040 --> 00:25:27.840 It is actually the Oxford Dictionary. 00:25:28.200 --> 00:25:36.580 They found a bug in the dictionary in its own citation usage of pi symbol from 1706. 00:25:37.220 --> 00:25:39.340 So they have upstreamed the fix. 00:25:39.420 --> 00:25:40.860 It's got approved and published. 00:25:41.180 --> 00:25:53.680 And so because of the messings around and playing with pi and the latest release of Python, we got stumbled upon an actual markup bug in their own dictionary and got it published. 00:25:54.060 --> 00:25:57.040 So thanks to Hugo for doing that for us. 00:25:57.300 --> 00:25:58.420 It is not the dict. 00:25:58.630 --> 00:26:00.320 It is the actual dictionary. 00:26:00.700 --> 00:26:01.280 Oh, okay. 00:26:01.470 --> 00:26:02.400 I didn't see that coming. 00:26:02.720 --> 00:26:02.940 Yeah. 00:26:03.220 --> 00:26:04.060 Oxford English Dictionary. 00:26:04.720 --> 00:26:06.260 Not quite a joke, but almost. 00:26:06.680 --> 00:26:06.880 Almost. 00:26:07.220 --> 00:26:07.280 Very cool. 00:26:07.860 --> 00:26:13.820 Very fun, though, to think that the Python community has gotten a fix upstreamed into the Oxford English Dictionary. 00:26:14.180 --> 00:26:16.539 Well, I sure hope they use Oxford commas when they... 00:26:19.280 --> 00:26:21.240 You're inciting some flames, Michael. 00:26:21.660 --> 00:26:22.020 Exactly. 00:26:22.050 --> 00:26:24.400 Like, rejected, improper use of comma. 00:26:24.860 --> 00:26:26.600 So I want to talk about for my one extra. 00:26:27.060 --> 00:26:28.000 Yeah, I think that's the only thing. 00:26:28.640 --> 00:26:30.220 I have other stuff, but I'm going to leave it with this. 00:26:30.520 --> 00:26:32.100 So I'm a big fan of Bunny.net. 00:26:32.320 --> 00:26:34.620 I mean, Cloudflare is all the rage, right? 00:26:34.800 --> 00:26:35.540 I've never used this. 00:26:36.080 --> 00:26:37.220 Bunny.net is awesome. 00:26:37.420 --> 00:26:38.380 The pricing is great. 00:26:38.600 --> 00:26:39.260 It's so easy to use. 00:26:39.580 --> 00:26:40.580 I use it for CDN. 00:26:40.680 --> 00:26:51.720 So, for example, when you get a download of Python Bytes, like the weekly episode, that comes through the CDN for, if I can pull it up, comes through the CDN of Bunny.net. 00:26:52.220 --> 00:26:58.380 But primarily, the reason I chose it, Ian Maurer recommended this to me, is because I want to use it for the courses. 00:26:58.960 --> 00:27:03.260 So for the course videos, one of the things that's a hassle is you can use CDNs a lot. 00:27:03.540 --> 00:27:11.700 There's a bunch of options there, but it's not very common that you can share private files through CDNs, which you can with bunny.net. 00:27:12.040 --> 00:27:13.080 So that's super cool, right? 00:27:13.180 --> 00:27:17.060 So for kind of like an S3 URL, you can sign it temporarily. 00:27:17.660 --> 00:27:21.400 Like this URL is good with this special link for 20 minutes. 00:27:21.860 --> 00:27:22.160 Play it. 00:27:22.420 --> 00:27:23.080