Python Bytes - #486 underscore-underscore-ghost-emoji

Episode Date: June 30, 2026

Topics covered in this episode: Free-threaded Python: past, present, and future django-admin-site-search Qwen 3.6 27B is the sweet spot for local development A large batch of PEPs are finalized Extras Joke Watch on YouTube Show Intro Sponsored by us! Support our work through: Our courses at Talk Python Consulting from Six Feet Up Connect with the hosts Michael: Mastodon / BlueSky / X / LinkedIn Calvin: Mastodon / BlueSky / X / LinkedIn Show: Mastodon / BlueSky / X Join us on YouTube at pythonbytes.fm/live to be part of the audience. Usually Tuesday at 7am PT. Older video versions available there too. Finally, if you want an artisanal, hand-crafted digest of every week of the show notes in email form? Add your name and email to our friends of the show list, we'll never share it. Calvin #1: Free-threaded Python: past, present, and future The GIL has prevented true multi-threaded parallelism in CPython since the beginning — multiple past attempts to remove it failed on performance grounds Sam Gross at Meta finally solved it; his work became PEP 703 and ships as free-threaded CPython today Python 3.13 was experimental with 20–40% single-threaded slowdown; 3.14 brought that to 0–10% Python 3.15 (October 2026) delivers a unified ABI — one extension binary works on both GIL and free-threaded builds Already >50% of the top PyPI binary wheels support free threading Wouters predicts free-threaded becomes the default between 3.16–3.20 (2027–2031), with the GIL eventually disappearing next decade Michael #2: django-admin-site-search via Adam Parkin A global/site search modal for the Django admin, by Ahmed Aljawahiry. Hit cmd+k anywhere in the admin and you get a command-palette-style search window, kind of like the one in VS Code. It doesn't just search one model's list page. It searches your entire site in one box: App labels Model labels and field attributes Actual model instances (your data) Two ways to search the instances: model_char_fields (the default): runs an __icontains across every CharField (and subclasses) on the model. Zero config, works out of the box. admin_search_fields: defers to each ModelAdmin's existing get_search_results(), so it respects the search_fields you've already set up. The part I like: it's permission-aware out of the box. Users only see results for the apps and models they actually have view permission on, so you're not leaking anything through search. Results appear as you type, with throttling/debouncing so you're not hammering the server on every keystroke, and it's full keyboard nav: cmd+k to open, up/down to move, enter to go. It's responsive, does dark and light mode, and it pulls Django's built-in admin CSS variables so it just matches whatever admin theme you're running. Under the hood it's Alpine.js, but bundled into static so there's no external CDN dependency. Setup is about what you'd expect: pip install django-admin-site-search, add it to INSTALLED_APPS, mix the AdminSiteSearchView into your AdminSite, and drop a few template includes into base_site.html. Supports Python 3.8 through 3.14 and Django 3.2 through 6.0, MIT licensed, and everything is overridable if you want to skip certain models, add TextField matching, etc. Calvin #3: Qwen 3.6 27B is the sweet spot for local development Qwen 3.6 27B is being called the first local model that genuinely competes as a general-purpose intelligence — benchmarks put it at roughly mid-2025 frontier level (comparable to GPT-5 / Claude Sonnet 4.5) Runs locally via llama.cpp; on an M5 MacBook Max with 8-bit quantization + multi-token prediction, it hits ~32 tokens/sec using ~42GB RAM 4-bit quantization gets it under 18GB, runnable on 32GB devices; Nvidia RTX cards run it even faster The dense 27B is recommended over the faster MoE 35B A3B — author prefers higher quality output over raw speed Privacy and reliability are the pitch: fine-tunable, can't be taken down, suitable for sensitive/proprietary data Author sees this as a stepping stone — frontier open-weight models like GLM 5.2 are now locally runnable with company-grade hardware, and smarter-still local models are coming Michael #4: A large batch of PEPs are finalized A bunch of PEPs went from accepted to final. 668, 687, 691, 699, 701, 703, 728, 770, 773, 829 But this wasn’t them making their way into CPython. It’s an admin sorta thing. (Thanks PyCoders) See the commit. Extras Calvin: More fun bling for your terminal this time - https://charm.land/ Michael: Follow up from pls, What the pls? Thanks Pito. Joke: BEMoji A production-grade utility and component framework built entirely on emoji class names via Jeff Triplett

Hello and welcome to Python Bytes, where we deliver Python news and headlines directly to your earbuds. This is episode 486, recorded June 30th, 2026. I'm Michael Kennedy. And I'm Calvin Hendricks-Barker. This episode is brought to by us. Check out six feet up if you want not just a consulting partner, but an impactful consulting for good partner. And if you're looking to do better at Python, data science, AI, check out Talk Python training. I have three courses in the works right now, which is interesting. So there's going to be a lot of new stuff to be sure to either keep listening here. Well, I'll tell you about it or sign up over there.

And speaking of signing up, by them by Soutfm, click on the newsletter. You can sign up there and get notifications every week as well as follow us on social media. We got all the links for the show, Calvin, and me. Now, Calvin, your first item is something has been sitting in a browser tab on my computer for at least three days. I'm like, oh, I need to read this. come on now. And I haven't gotten to it, so I'm really happy you're going to take care of that for me. Well, how could you not mention this? There was a talk at PyConn 26, and it's a perennial talk that has probably happened for the last two decades, starting with Larry Hastings. I didn't go

look up exactly what year he started doing the Gallectomy project, but the idea is fine free threading in Python, past, present, and future. So there's been a lot of discussion in the past. Like I said, there's over two decades of history going into solving this problem. This gets back into Python and the global interpreter lock and what we've done in the past to try and improve performance. So for those who are kind of new to Python, that a global interpreter lock really makes it difficult to run a single Python process and leverage multiple cores in your computers.

But every modern computer today has how many, even the Raspberry Pi's have many cores in them. So this would be a huge boost, we think, to Python to be able to take advantage of more computing power going on inside those CPUs. So basically, the Gill has prevented this true multi-threaded parallelism in C-Python since the very beginning. There's been multiple attempts. Many of them have failed on performance grounds. Once you remove this global interpreter lock, that overhead of managing I.O. memory, et cetera, all the pieces going back and forth usually adds in, you know, 100% or more additional performance load than giving you the performance you'd think you'd get by

leveraging multiple CPUs. We got free threading in what, Python 313 was experimental, and so it's still a 20 to 40% single threaded slowdown. 314 brought that from kind of zero to 10%. And so look for Python 315 coming in October to deliver basically a unified story here. They've made a huge amount of progress here. During the talk, it's interesting because this always brings out the OG Pythonistas, core developers in the room for these kinds of talks. Actually, one of the members of our team was in that talk at Pygon. Larry Hastings got up to the mic and Thomas Woosers basically, they both kind of played off each other on how this all came about. And the punchline was the things that made this possible didn't exist when Larry was trying to do this. So there's been a ton of development

and innovation in the Python core to make this all go. The big part about this is now that we're going to have a unified ABI. So basically the interfaces into using this free threading is going to be more unified, which means the same extension binaries, like if you're writing C extensions, can work both on the gill and on free threaded builds. And it looks like already greater than 50% of the Pipei binary wheels support free threading. So that's a huge, well, the top, sorry, top Pipei Bineries, not all of them. But of the major ones, we've already got half adoption.

I think that's a huge boost. I think there was a prediction made on from the stage that free-threaded may become the default version of Python. But we'll see what happens. Oh, excellent. So have you followed along with this Gill story, Michael? Yeah. Yeah.

Everyone has. I love at least having the capability to do threading. I would love to see this kind of seamlessly integrated into async io i was just reading a reddit post about somebody who was hating on async io async in a weight in particular and no it might not be the cleanest easiest programming model of the world not python in particular but just in general but it's better than managing a bunch of threads or processes or you know it's it's i think it's really quite neat and lightweight but it's big limitation is as soon as you get anything computational boom like the whole thing stops

right? Yeah. And so if you could just say, you know, put a decorator onto a function like CPU bound or something or somehow, you know, kick it off with some kind of, you know, computational thing, but clearly like just await, make that awaitable in the same event loop. That would be awesome. Yeah. Now, this is a really good article from the Linux Weekly News, really in-depth coverage by Jake

on this site. So I highly recommend folks who are interested in history and what the future is going to be. Check this out because we're finally going to have free threading in Python. Awesome. So if you were running like a Django site, I don't know, wouldn't it be cool? Wouldn't it be cool to, I don't know, search your admin site in the same way that you do? No, Michael. Like really nice, say Visual Studio code, you can hit Shift Command P for your command palette or in Vivalda. You can just hit Command K and you could say, I don't know, copy all the links on the page by just type in a thing or two.

So I, thanks to Adam Parkin, I introduce Django admin site search. And it's quite a slick thing. It's barely new. It could use some more stars. So there, it got one more. In real time. Real time. Yeah, so you just hit Command K, like I just did in Vivaldi.

And by the way, if you do Vivaldi, you might need to remap one or the other, but whatever. And then it brings up this nice search. People are probably familiar with some of the docs, like the great docs have this and lots of others. But the idea is that it's really deeply integrated into understanding what your jingo needs. So there's a little animated video here, as, by the way, everyone out there building this, if you're building something specifically to do with UI, at a minimum needs a screenshot. I've seen like an entire website design frameworks that have had like zero, zero pictures. I'm like, I have no idea what you're talking about.

Like the words sound good, but pictures people. They'll resort to interpretive dance before they'll put a screencast in. Yeah, exactly, exactly. So the idea is it just works out of the box, basically very little config. It's when you type something into it, it searches app labels, model labels, and fields, and model instances. And what's cool about it is it can search in two different ways. So for smaller sites or sort of ad hoc searches, it will run a dundra icon, and not a dundercore underscore iCentains, which is, is any part of this text I've typed contained within here without regard to? case, right? That's cool, but that's not always indexed, right? I mean, not every single field

has an index for these sorts of things. So that can be slow on large projects. So you can also tell it, use the model admins get search results method, which can be an indexed query. That's cool, right? That's super cool. And this is super nice. I've seen a couple of people doing work on the admin. This is a much-needed feature. Yeah, I think it's really neat. It has built-in off, so users can only search apps and models that they have permissions to view. Results appear on type. It's active search, I guess, is the way you would describe it. And keyboard navigation.

So obviously, Command K, like I said, you might need. It's a good, good one. Oh, you know what? I think, no. I think, yeah, I think that's the default in Chrome. I'm trying to figure out if that is, which one is it. I may have changed it. I think maybe Command E is the default on, I'm sorry, Vivaldi, for its command

palette, but, you know, it's layer. there's a command palis you might have to battle with. Anyway, supports all the way back to 3-8. They don't even care if Python supported. They're going to support it. And Django 3-2 to 6. So, yeah, that's pretty much it. It's super easy to add in.

Not a lot. Not a lot to do there. It's customizable. Yeah. No, I like that. As a Django developer, that's definitely a missing piece. Because I feel like it's like the old days of the internet where there was no search and you had to always know where to browse to you to find

the things in the admin console. Yeah. This is going to save a lot of people a lot of time, especially if we've got a larger Jango app. And you've got hundreds of models in there. Like this is a little going to be a lifesaver. And some people use the admin section for like all the admin staff of the Jingo site. I don't recommend that.

I don't know. I don't necessarily either. But people do. And so having this to navigate it, I think would be good. True. In the last minute crunch when you're trying to triage something that admin is indispensable. I guess it depends, though. You know, is this an internal company app that 20 people use and three people admin?

Yeah, yeah. How serious you got to take that versus Instagram. Right. But it's not, I bet it's not admin through the admin. We should get some, we should find out. We should get some inside news into whether they use the admin or not. What if it is? Right, no, you should see ours.

Arts is amazing. Well, I can tell you right now, someone's going to mention it in the comments. They're going to call us out. Let's do it. Let's do it. We could ask an AI about it probably. A local AI. And local AI, actually, this is really exciting.

I have been certainly on the bleeding edge of, of local AIs for quite some time running them on my framework desktop here. And then I've always managed to try and buy a MacBook Pro with the Apple Silicon and about as much memories I can afford. And so being able to run these local models has been mostly an experiment up to this date. Jim Afforge gotten really good. Some of them have, you know, Mistrel, Dev and Devstrel have been really interesting for live coding agents.

But this article that came across my feed today, multiple people in the company actually posted this. which is Quinn 3.627B is the sweet spot for local development. So there's a couple models that just got released in the Quinn family of open weights models. There's also a quantized version of it and a multi-tokin prediction version of this model out there, which gets some pretty blazing speed. On his M5 MacBook Max, he was getting 32 tokens per second with the 8-bit quantized version of it. But if you run the 4-bit quantization, you can actually run it on a 32 gigabyte or less MacBook machine,

and you get pretty blazing speeds. I think they puts a chart in here on the bottom. But the nice bit about this article is he compares the various models, what they're capable of inside the same family, and it actually gives you a whole how-to on how to hook this up using Lama CPP and OpenCode. And so you can actually rebuild the demos that the author, wrote here. So using open code, hooked it up, and then wrote up a minesweeper app, and then again, went into some depth on the performance metrics here. So you can see the 3635B-A-3B model on the Apple

MLX, 85 tokens per second running with that model. That is incredibly usable. But the real sweet spot for the development was using Lama CPP and getting 32 tokens per second using about 42 giga-ram. So if you've got a 64 gig machine, this is absolutely doable for developers that are out there today and felt like it was, I mean, this is not going to be like mind blowing for anyone who's already using frontier level models like Opus and Sonnet and GPT55 Codex to do code, but this is local. Everything stays on your machine. The data never leaves. It's tool using, so there's still risk. You're using models that have been trained by foreign nations. and who knows what's going on inside of these things. But this one seems to be gaining a lot of traction and actually producing, you know, I don't think it's mythos level yet, but we're only, what, maybe six, nine months behind

on open weights models that I can run myself. So this looks really good. The interesting thing he also pointed out in the article was, and I wasn't aware of it, the, I've been an O-Lama user for quite some time. And I think I'm going to be switching to the Lama CPP version. There's been some authorship and IP, concerns between the two groups that I wasn't aware of and I think I feel like to be responsible,

I should switch over to using that. Or I'm considering VLM on my framework desktop so that I can run multiple models simultaneously and have multiple instances of the same model running so that, for example, my home assistant can have multiple people in the house make requests simultaneously. Okay. This is pretty exciting. I can't, I've not tried three, six. I have three five currently running on my desktop at the moment. And that has been impressive from a home assistant standpoint. I've not used it for much of my coding agents, but I'm going to give 3627B a real shot using, I've been using, again, Pi, still my daily driver. There was a new release of superpowers since our last episode that has made the Pi experience a lot more user-friendly and smooth. So check

that out with Quinn-36. It's really neat. One thing is, can you pull up that speed? Yeah. I love how they have like a heat. It's definitely going to heat up your laptop if you're using these tools. What surprised me on MacOS is that MLX is the slowest. I know. I was surprised at that too. I think he mentions it down here in the text. RTX, consumer RTX, if you've got enough V-Ram on that card, that is the way to go.

If you're serious about running these models, I believe. But just the standard Lama CPP plus MTP with that quantum. version, 105 tokens per second. That's plenty fast. But 32 tokens per second is plenty fast for doing code dev where you're in it, you're reviewing, you're using it as a true assistant, and you're not trying to

just vibe code away an app. Another takeaway that when I see anything like this, definitely a cool topic, is I don't know what percentage of people it is out there. Maybe I'd love to hear your thoughts on this, but I think that there's certainly a strong contingent of people out there who want

AI to go away or think the bubble's going to burst and it's just not worth getting into or paying attention to like i i know open-a-i or anthropic or whatever the whole thing is just going to go up like in a smoking dumpster fire in six months so why would i change the way i work and melt my brain and whatever else you the people say about it this stuff is not going away this is on this is here right right this is on hugging face the models are published on hugging face this is open open source and published i say open source i shouldn't say open source they're open weights uh some of the models are fully open source but this is this is sustainable maintainable a level of usefulness that is undeniable

there's there's hypesters on both sides of that equation the answers lies in the middle and the answer really is that these tools are great amplifiers of your existing skills so if you're already good at the software development life cycle these things make you better if you're not good at it these things can make you worse exactly you have no idea that's called vibe coding right Let's all be realistic about what this is all about. Yeah, and I'm also jealous of your framework. So, side note, Apple raised their prices this week on MacBook Pros. I went and double-checked and framework had raised their prices too.

So the same 120, the same 128 gigabyte machine is like about $6 or $700 more than it was a year ago. Wow. Yeah. I don't know. Maybe I'll have a, I was going to start this episode with a good morning. Good morning, everyone. At Apple horse, we've been so honored and proud to have fleeced you on RAM and on VVM that today we're taking it to an entirely new level. Like, oh my goodness. Yeah, so the price is like, you're going to be excited, Michael.

Oh, boy. Let me tell you what we're going to bring on Calvin to show you what we've built. And boy, is it expensive. No, like seriously, just really quick, the follow-up to the sidebar. Like, my, I think my MacBook Pro 16-inch, which I just bought, just received, I bought it a little bit before because it took a while, but just got two weeks ago, something like that.

I think it went up $800. Yeah. Luckily, I bought two in preparation for this prior to the risism price. So I just saved myself $1,600. That's awesome. There's some guy math we could do it. There's totally some guy math going on here. All right, well, we're going to spend it on.

Let's talk about a we're spinning on motorcycles, Calvin. You know what we're going to spend it on. We're absolutely going to spend our motorcycles. So I would I would like to tell people about what might be the largest single pep except finalization in one batch ever. Go on that sounds exciting. It sounds exciting. It's more interesting like that's weird. Interesting.

So all in one. go, I believe these are right peps. I might be off by one. I think I might be off by one. Anyway, it's basically 668, 687, 691. There's like 10 peps or something that got, with the final, just all at one shot. Then when you look at them carefully, so I link to the 11, yeah, it's 11 false change. I link to the commit that did all this. And so you'll see things like, if we expanded here, the diff is not always perfect. PEP 668, making Python based environment is externally managed. But if you can, you can, go over and you check out Pep Pep, Pep,

PEP, 668, which is here. It was created. Oh, wow. May 20, uh, 20, 21. House clean. Others are like 2023 and so on. So what had happened here. Oh gosh, I forgot who, who did the switch. I guess I could probably look at the commit. Zero intensity, which is Peter. I didn't really need to click on that. In fact, that was useless. Um, had Peter Birmah, uh, core developer did all this. And,

What it really is is there have been all these peps that were sitting in accepted but not final. And accepted means, well, accepted final means it's shipping. Well, clearly, the one that went was out and accepted and final in 2021 is out. So there's just a bunch of peps, 11 in particular, that were just hanging around and never, ever got moved to final for some reason, you know, reasons. I don't know. Someone was not a completionist. Exactly. Yes.

Yeah, exactly. someone's like oh we're gonna get all the little badges baby all the badges need to be put on here yeah so for example this this one is uh 20 22 to over 22 and so on so i just think this is kind of interesting and people have been looking at these these older peps and going like why is this not final why is it just accepted well you called out not completionist yeah i'm glad well it's nice to see those all cleaned up so that at least there's a canonical source of the truth here yeah absolutely i totally agree yeah um but Bit of a comment out on the audience here. We've got a pixel says, like the sentence. It makes you worth, it makes you worth if you don't know the software cycle, because I saw people during vibe coding

and they're happy, but they didn't know what it actually makes, what it actually makes them worth. Yeah, it's interesting. Very true statement. Yeah, and also, yes, also being a subject matter expert combined with some coding skills is absolutely. 100%.

It does make up being a generalist a little more useful. Yeah, actually. exclusively, but also. I finally come into my own, being the journalist that I am. Yes, exactly. Me too. My time is here. It's incredible. I can do all the things.

Well, I mean, I've just found myself building so much more stuff now. Like, I know the patterns I wanted to get done and it was just a lot of busy work that didn't feel like doing, but now I can make those tools super quick. I love it. Big fan. Yeah, big fan. And it's like, well, if it took me two hours, this is worth it. If it took me two weeks, it's not worth it.

Totally. No. Now it takes you two hours. That's absolutely the math. All right. And you can power it out of your framework in the basement. Okay. That's true.

So let's go and jump into extras. Yes. And I already got my screen up. So I'll just grab the first one here. Go for it. So I wrote a blog post inspired by a guy named Pito, who had reached out to us because you and I both were a fan talking about how ELS instead of LS is quite cool. It is quite cool. It is quite cool.

You can go and when you LS on something, if you alias it, I'll do that. has some drawbacks. It uses color for like hidden file. Like GitHub. Dot GitHub is shown a few LS, even though it's technically hidden, but it uses color to show that it's like Git ignored. It shows Sim Links, which is really cool. And I've learned how to create a SimLink in Git. So when you check it out, it's Sim Links on a new machine, which is always fun. It uses icon. All this stuff is super cool. But Pito pointed out something is weird with installing this package. If you click on the Pi Pi link, it says, version 6. Some of the links 404, which is kind of weird, but then if I actually go and click on,

say, the repository, it takes me to the Rust-based ones. Like, oh, this must be the Rust one, because when I go to the PiPi and I click on the repo, it goes to the Rust one. He points out, no, that's the Python one. And I'm like, wait a minute. Is there something suss going on with this thing? And he inspired me to dig into it. So what is going on is GitHub is redirecting to the newer version when you click on it, but Pi Pi actually installs a frozen 2023 version. So I wrote this really short article. I didn't know that. Yes, I know.

I didn't either because everything about it like redirects to the one that you think you're going to want, but it's not. So anyway, it talks about like, okay, did you install this with UV or PIPX or somehow PIP install? If you did, you're getting the old frozen in place slower one that you think is not the what you're getting, but that's what you're getting. And here's how to actually install it. Well, I'm going to go do that. Yes, exactly.

So I wrote a really short write up about this. What the PLS. What the PLS, PLS. Come on now. All right, Kyle. Well, if you think that makes your command line glamorous, I've got something for you. You're going to crush it? We're going to remember your first crush.

This, because the command line was probably my first crush, if I'm going to be honest out there. And if you want your command line to look absolutely beautiful and glamorous, there's a, library out here called Charm that can do all kinds of fun extension. So forms and it's got interfaces to build inside there with like SSH animations, physics based animations, you know, get the bubble T component for your toolkit. So you should definitely check out crush or no, sorry, it's charm. dot land to add some fun bling to your terminal. So if you think that if PLS was just the start, oh my goodness, charm is going to, you're going to be busy. Michael going down the rabbit hole.

Dang it. I was just starting to work on some projects and now here I am. Yeah. Rebuilding my terminal prompt again. Yep. It looks pretty cool. So I've not checked it out. I think the open code folks may be using some of this.

But yeah, it looks a lot like a lot of fun. So if you want to go bling your terminal, this is a way to do it. And then it's not very popular, is it? No, not at all. 213,000 stars over on GitHub. Not bad. Pretty cool. Add some fun to your.

your terminal. The other bit, and I kind of already alluded to this, was building a fully local voice assistant in 2026 using Python. So this actually is a Raspberry Pi-friendly version. As for those you were falling along at home, I've been doing the more ridiculous version with like RTX 3070 running my voice models on a giant graphics card. This runs on a Raspberry Pi. A couple different models. He mentions here in the articles with the full architecture and how to set it all up and how to get it going and you can even run it on old Raspberry Pies. So there's a 40 meg model the small 015 version. Meg.

Yeah, megabyte. Megabyte. Wow. Okay. Inaccurate, but we'll run on one of those. So if you've got a more- So it's like Siri basically. Yeah.

Wow. Shots fired. This is its default failure mode. Is this what I found on the internet? Yeah. But you can see even the largest one, 1.8 gig. can even run that on probably the largest of Raspberry Pi's that are out there or just on a CPU and not have to use a full-blown GPU to run your voice pipeline. This is part of the Plattie Push

platform which is not necessarily a competitive home assistant but is like Home Assistant in that you can glue together ZigB and Wi-Fi IoT devices and all those kinds of things inside your house. So I just thought that was really interesting because it's on my list of things to keep up to date on, which is I want to replace all the ladies in the box in my house, if you know what I mean. Yeah. And that's what I got. What about something, you're a Docker person. I am.

You know, and you've done a lot of local doctor. I'm more of a server Docker. One of the things I think would be really cool. It would be so easy to set that something to an always auto starting and less stopped sort of Docker thing. But at least on MacOS, can you give it access to like the GPUs or MPUs? You know what I mean? because it's basically Linux within MacOS, which is not exactly the same.

I have not tried the GPU on Docker with MacOS. I am using this on. I have installed ProxMox onto the AMD box downstairs that has the Kuta card in it, the 3070 card. And in that one, you absolutely can pass with ProxMox, the GPU drivers. And you just have to make sure you get the Kuda drivers installed. ProxMox doesn't come with them out of the box. but you can install them on to ProxMox.

And once you do, you can pass through either shared access or dedicated access to the containers. I'm using containers to run all my voice assistant stuff on the Prox. Sweet. Okay. So that is absolutely possible on Linux. And ProxMox makes it just stupid easy to do. Nice. Yeah, I have this nice Mac Mini M4 Pro sitting around.

I would love to just fire the things up. But all the stuff that requires sort of some acceleration is like sometimes slower. Maybe I don't really want it that much. Well, since I had that RTCX that was donated by my child to me, I was like, I'll use that. Because it's going to be the fastest tokens per second, but it's only an 8 gigabyte V-Ram. So whatever I can fit into that 8-gig goes very fast. Yeah, perfect. All right.

Let's finish this up with a joke. So this one comes to us from Jeff Triplett. This one's good. Have you seen this? I have seen this. The team's been talking about it. Yeah, it's, well, I think you all should adopt this for your project. We totally should.

You let me look. A lot of times people create this kind of a little not so easy to understand thing as a way to like get, create a repeat business or to not, you can't fire me. I'm the only one that knows how the what it does. And CSS was already crystal clear with how, you know, to author it. Indeed. So what we get here is a legit CSS framework called Bimoji. It's the CSS framework that your team will actually argue about.

And so here's what's going on with this is why. waste all the characters right when you could have something much more you know discoverable or understandable with emojis right I mean this totally makes sense I'm here for it so if just don't even know so for example you might want to talk about like a hero image and you want to stylize it so you would say joker emoji underscore underscore picture frame emoji and like a lot of the utility classes, you would need to like specialize that. So you would say Joker emoji underscore underscore picture emoji dash dash star because this is your hero. The hero. And then there's the

editable part which is Joker emoji underscore underscore underscore notebook editor if you want to have a little card, like a written card. And then Joker underscore underscore ABCD for the title and Joker underscore unquote quote emoji for a description bit. So on. The footer's the best. Okay, the footer. Where's the footer? Oh, right there. The class is Joker underscore underscore foot emoji. Oh, there it is. Yes, it's literally the foot emoji. What is the, I guess the disabled is a ghost?

Yeah, that's perfect. I mean, it's obvious. You read that, you know exactly what's going on, Michael. Yes, I guess you do. Once you learn to speak the emoji, about the block, the element. Oh, it's so good. You want to link, you just put a link emoji on it. Anyway, I'm not sure that I'm going to be adopting this.

I've already done a couple major framework remodels. I put it on our tech radar. Yeah, it's going to be important. I mean, you got to stay on top of what's coming in the industry. And sometimes, sometimes it's a ghost emoji. I think that's a good place to leave it, Calvin. That is a good place. I love it.

Thank you all. That was a lot of fun. Yeah, thanks. See you later.