1028: Cloudflare Wallets

Cloudflare is rolling out crypto wallets with claimable handles as identity, and a real React compiler finally landed for regular hooks-based code. Plus: OpenAI's pricing war, Vue Vapor benchmarks, GitHub's new npm malware scanning, and an active supply chain attack hitting 868 packages. Show Notes 00:00 Welcome to Syntax! 01:02 Shai Hulud is back! New Shai-Hulud announcement Attacked keyv packages 05:27 GitHub scans your npm packages for malware 06:42 Your agent has a wallet now Make your own Cloudflare wallet announcement blog post X402 payments 12:10 Yet another React compiler? Ripple.ts Inferno.js Dominic Gannaway’s launch post Octane Native Script Reactivity + Rendering Benchmark 27:19 Scott switched his browser AGAIN! Dia Knock off extension 31:56 AI Mental health survey 32:29 OpenAI models are becoming cheaper? 38:49 GitHub stacked PRs finally arrived 41:48 AI video podcasts HeyGen launch post Hank Green’s unhealthy AI usage 48:15 Brought to you by Sentry! 49:28 News from the TC39 meeting ECMAScript news Await dictionary TC39 Process 56:58 Qwen 3.8 01:02:31 Latest and greatest in CSS and UI libraries Foley.dev Morphicons Lucide Nucleo 8bit library CSS radar Impeccable Hit us up on Socials! Syntax: X Instagram Tiktok LinkedIn Threads Wes: X Instagram Tiktok LinkedIn Threads Scott: X Instagram Tiktok LinkedIn Threads Randy: X Instagram YouTube Threads

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Wes Bos?

  What's up, everybody? Thanks for tuning in. This is a Syntax livestream. Tons of news going on this week. We've got CloudFlare wallets was just announced. We're going to talk about what that is and why you might be interested in. Another Shaihalud attack going down right now. Scott's going to fill us in. Don't npm install anything until you hear what Scott has to say about that. A real React compiler. Last week, we talked about the new React compiler, but CJ's got a new React compiler. AI pricing wars have started. OpenAI is discounting a lot of their models right now. Plus, a whole bunch of it. If you have anything you'd like us to cover, hit it down, throw it down in the chat, and we'll be sure to cover it. What's up, everybody? How you doing?

• 0:41Scott Tolinski?

  Oh, doing good. How you doing, Wes? Yeah, you get wormed today?

• 0:46Wes Bos?

  Not yet, unfortunately, but or fortunately.

• 0:52Scott Tolinski?

  Yeah, unfortunately.

• 0:55Scott Tolinski?

  Wes wanted to get wormed. I got it. News at 11. Yes.

• 0:59Wes Bos?

  Yes.

• 0:59Scott Tolinski?

  Yes, dewormed. Yes, we need to get dewormed. Well, today, there is a new Shaihalud attack going down, and one which has already compromised 868 packages carrying over 2 billion monthly installs. So the, it seems like the main package that was compromised, it started with this Keeve package, which I haven't used. Have you guys used this one?

• 1:29Wes Bos?

  No, I don't believe so, but you never know if it's all the way down the dependency tree.

• 1:33Scott Tolinski?

  Yes, it's a key value in cacheable.

• 1:39Scott Tolinski?

  It's a key value library that I, it seems like it has quite a lot of people using this. So this is a library that, well, was compromised, and now the worm is doing worm stuff, and a ton of other packages have been compromised. So the, folks, the advice that we always give you on these worms is that you should be setting your packages not to install things that have not been available for at least a day or so. PNPM, npm, they all have this feature. So PNPM has it by default. Easiest solution is to just use PNPM. It's a great package manager in the first place.

• 2:20Scott Tolinski?

  But just scrolling through this list, it's just another one of these big old worms that's coming up to slurp up all of your stuff, which is really the thing here is that it's trying to, it's trying to grab your ENV variables. It's trying to grab all kinds of stuff. Again, a pre-install hook fires on npm install and drops a stealer that sweeps npm, GitHub, AWS, Kubernetes, and Vault secrets, and then spreads to more maintainers. This is the same type of worm that we've seen over and over and over again. I think this is like the fifth or fourth Shaihalud that we've seen recently.

• 3:02Speaker 02

  This year, this year alone, yeah.

• 3:03Scott Tolinski?

  Yeah.

• 3:04Speaker 02

  Seven months.

• 3:05Scott Tolinski?

  So be careful out there, folks. There's a CSV of a ton of packages that have been hit from this bad boy.

• 3:13Wes Bos?

  One interesting thing about all of these attacks is that you're not hearing these stories of, there was an attack and these Bitcoin wallets were drained immediately. You're often hearing, there was all of these attacks, and then we don't often hear too much follow-up of like, what actually happened? And Shano, Friday, I'm releasing a video on the like, black market of AI tokens. And what I'm learning is that a lot of these hacks, they're stealing your AI tokens, or they're stealing access to your, to whatever app you have. And then they are just, just stealing enough that you won't notice it, and they're selling those on the black market. So on Friday, I have a video going live. Make sure you catch it.

• 3:56Wes Bos?

  Where I dive into, I actually go ahead and buy some of these black market tokens that are often stolen in these attacks.

• 4:03Scott Tolinski?

  Wow.

• 4:03Speaker 02

  Nice.

• 4:04Scott Tolinski?

  Yeah, I'm really stoked to see that.

• 4:06Speaker 02

  Me too. And I'm working on a video that will show you how to not run code on your computer, potentially to stop you from being a victim of one of these attacks as well. So we got you covered here at Syntax.

• 4:15Scott Tolinski?

  Yeah, we got you covered. And I'm actually making a video on mental health and AI coding, folks. So I'm running a survey right now, and I'll post this in the chat. If you could fill this out, it is quick. It's a breezy five questions. Actually, it's, I lied. It's six questions, but one of which is just, how many agents are you running? So we're doing a deep mental health survey here, really super fast. If you could go ahead and fill that out, that would make the results a lot better, and the video's coming soon on Syntax. So just.

• 4:49Wes Bos?

  That video is going to be mental, Scott.

• 4:52Scott Tolinski?

  I don't, can you say that? Yeah, I guess so. Yeah, it's going to be mental. Yeah.

• 4:56Wes Bos?

  Can we say that? I once got in trouble for saying that this is bananas, and somebody told me that was a mental health slur. Yeah.

• 5:03Scott Tolinski?

  Yeah.

• 5:03Wes Bos?

  Bananas? Can't say bananas. Something, if something is bananas, but now we say insane, and it's okay to say insane. But four years ago, you weren't allowed to say bananas.

• 5:11Scott Tolinski?

  Insane in the brain?

• 5:12Wes Bos?

  Yes, in the membrane.

• 5:14Scott Tolinski?

  In the membrane, yes, yes, yes. Saying gyno brain, yeah.

• 5:18Speaker 02

  Yeah. So and before I hand it off, I actually have some super relevant news. I'm just going to slip right in to get this one in here. GitHub, who Microsoft owns in PM, they published on the GitHub blog.

• 5:32Speaker 02

  Wait, is this? Yeah, so they are now doing publish time malware scanning and dual-use metadata. So this is something I mentioned they needed for years, but this would potentially prevent the, an attack like Shaihalud from happening, because before a package goes public, it'll be scanned for malware. But now I'm reading it, I'm realizing maybe this is the GitHub registry, not the npm registry. But the idea here would be if a package maintainer gets compromised and they try to publish, it's first going to do a malware scan on the npm side, and that would prevent it from being published if malware was detected. So they're doing their best. I'll have to confirm if this is just the GitHub package registry, or if it's npm in general.

• 6:17Wes Bos?

  Yeah, that's always been a weird part about it is that, like, what's published on npm is not necessarily the same thing that's on GitHub.

• 6:25Speaker 02

  Yeah. Yeah. So this is good news, and ideally will prevent some of those types of attacks from happening. But I'm going to hand it off to Wes. If you're just joining us, this is Syntax Weekly. We're going over news in the world of web dev and AI and everything else that we care about. And Wes is going to talk to us about the new CloudFlare wallets.

• 6:46Wes Bos?

  Yes, this was just announced a couple hours ago, as of the time of recording this morning. And CloudFlare wallets have been announced, which is essentially just wallet for your agent. Couple of months ago, CloudFlare announced the monetization gateway, which is essentially, you want agents to be able to access your content, whether that's to pay for inference tokens, whether that's to get data, et cetera, et cetera. Then your agent has to pay, right? And there needs to be a way for these agents to be able to pay, even if it's like half a cent or a cent or 12 cents for access to the specific thing. So there's been this new, like, open standard X402 payment required,

• 7:27Wes Bos?

  which is essentially this way for agents to be able to pay for access to whatever it is you want. Now, this morning, CloudFlare announced the CloudFlare wallets, which is essentially going to allow you to give a wallet which can hold stable coins. So it seems to be it is a crypto wallet, which I think is kind of interesting. Before I even get a bit more into that, you know how like QR codes? The, the solution to QR codes were announced, and like, we never used them for the longest time. And then the pandemic came around, and we're like, oh, QR codes are actually helpful for like menus and ordering online. And like, like, we had QR codes for like 15 years before they actually became useful.

• 8:12Wes Bos?

  What if the actual useful part of crypto ends up being that agents can use it to pay for things? Wouldn't that be wild?

• 8:22Speaker 02

  That's what makes it relevant again?

• 8:23Wes Bos?

  Yes. Maybe.

• 8:26Speaker 02

  Maybe, yeah.

• 8:27Wes Bos?

  But the, what these wallets will do is you'll be able to both send and receive. So you can pay for things with, you can give, you can create like a wallet, and you can give it like $20 and say, go complete this task for me. And you might need some AI inference, but you may also need to access to some data that is paid. Like, one example of like paid data is like stock market, right? Stock market is for free, pretty much on so many different APIs, but it's always 15 minutes delayed. And if you want the most relevant, you have to pay like big bucks to some provider to get like, like updated on the millisecond. So that would be a good example. Like, yeah, I want to run this every three hours,

• 9:09Wes Bos?

  but I wanted to do it with updated data. You could use this type of thing for it. Now, along with that, CloudFlare is allowing you to reserve your own wallet, because. They are, I guess, going along with with the wallet is you have this identity of like, who are you, right? Because if you want to pay for something, you have to like, be able to attach that to somebody. And that's, that's, I guess, one of the problems with crypto right now is that it can be very anonymous. It's a benefit and a bit of a problem. So you can register your handle on cloudflare.pay.

• 9:45Scott Tolinski?

  So you can tell your agent to go off and do something and get yourself a little something for yourself, some inference for yourself while you're out there.

• 9:54Wes Bos?

  I thought it was interesting that this, like, they're not the first people to do this. There has been like, kind of two, there's more than, more than two, but the two that stick out in my mind is Shopify has this shop app, which is essentially all Shopify merchants. You can just buy stuff from Shopify merchants through shop app. It's sort of

• 10:13Scott Tolinski?

  I gotta say, I like shop app. I like that.

• 10:16Wes Bos?

  I like it quite a bit as well. And they have, if you go to like, any Shopify website, four slash skill dot M D, it gives you instructions on how an agent can buy something via the shop app and do it programmatically. And you can give your agent like 100 bucks. You can put some, some money, or you can say, my shop app is already has my credit card in it. Now go ahead and buy me something. I don't know if I'm gonna tell an agent to go buy me a shirt, because like, that's, I don't feel like real people actually do that. You know, you see all these AI bros that are like, buy me shoes. And then it goes off and buys you shoes. They're like, holy crap, it bought me shoes.

• 10:56Scott Tolinski?

  Computer, buy me shoes. Computer, better shoes. Computer, make them red. Yes.

• 11:02Wes Bos?

  I don't feel like that's, that's the actual move, but you can do it via shop. And Stripe has like link pay. You know, if you ever go to a website and you put in your phone number, and all of a sudden, you get a text message, the code puts in, and then your credit card's hooked up. That's doing something called Stripe link. And Stripe has this thing Stripe link for agents. So there's kind of a couple of people playing with it right now, but I reserved my handle at Wes Boss. So when it does come time that you can send people money, please, I am begging you to send me some money.

• 11:36Scott Tolinski?

  I'm begging you. I'm begging you to send me some money. And let's get some little school or something for Wes's agent too, while we're at it. Yeah, a little something, something. Yes, a little treat for yourself. What would be a little treat for an agent, like an API key, or like?

• 11:54Wes Bos?

  Well, if it's, if it's always running these like cheap Chinese tokens, like every now and then, like you get, okay, you can treat yourself to $2 of Fable, you know?

• 12:01Scott Tolinski?

  Yes.

• 12:02Wes Bos?

  Like that, that one thing that you've been meaning to, to, to ask, and you can't figure it out with any other thing, treat yourself to a little Fable.

• 12:09Scott Tolinski?

  Something really nice. Well, CJ has something for us now, which is something high octane, I'll say, which is, is a new React compiler, is it a new framework? What is this thing? CJ, you tell me.

• 12:22Speaker 02

  I'm gonna get into it, and I'll start with a little bit of backstory. So if you're not familiar, the React compiler was released version one back in October of 2025. And it's called a compiler, but I have complained about this, that it's not actually a compiler. It's more so a React optimizer. So if you look in the React docs about what the React compiler does, it essentially takes code where you had to manually add use memo and use callback. It makes it so that you can write code that doesn't use that. But then it goes through, finds all the spots where it should add a use memo or add a use callback. So that's what the React compiler from the React team does. It's basically an optimizer. So you don't have to worry about the, the nitty gritty details of use memo or use callback.

• 13:07Speaker 02

  And that's fine. That speeds up a lot of people's apps and fixes certain bugs they didn't know they had. But like I said, this isn't a compiler, it's more of an optimizer. There's a new, there's a new kid on the block, and it actually comes from the creator of Ripple JS, Ripple TS. And I made a video about this when it came out. But this was a brand new framework. And the whole idea was, they came up with this new template syntax, and it's a bit more declarative in that things kind of run top down. So like, if you know the, the rendering model of React, basically, you define a component, but this is a, a render function. So anytime there is an update to state, this render function is gonna rerun,

• 13:49Speaker 02

  and then produce a new virtual DOM tree, and then that gets added or reconciled into the DOM. Whereas with this, your template essentially kind of runs once, and then only the parts that are interactive will update or rerun when they change. Very similar to what like Solid JS does, or if you look at the reactivity model of like Svelte. So yeah, this was Ripple. It made some waves. I think a lot of people were turned off by the syntax. They go into why they like this syntax and why it works for their compiler versus just standard JSX. So that was fun. And that also resulted in a, like, TS RX that you could

• 14:30Speaker 02

  use with React. So the creator of this, Dominic Gannaway, he saw that people liked the syntax, and then instead of making an entire new framework, he made it so you could actually write TS RX with React. Okay, so that's the backstory. Dominic is very prolific in the world of JS. He also created this framework called Inferno JS. And so this uses JSX, it looks very React like, but it's a lot more performant. I don't know all the nitty gritty details of Inferno, but I'm giving you all this backstory because Dominic knows what they're doing. They have a lot of experience in this space. And that leads us to the latest release, which is Octane. So it's a new framework, but the tagline here is React

• 15:11Speaker 02

  programming model compiled. Essentially, you can write React code exactly like you're used to. So you have use effects, use states, et cetera, et cetera. You can basically just swap out the import from React with Octane. But now, when you build your app, there is no virtual DOM runtime. Everything gets compiled into Vanilla JS with update code to change the DOM directly if state variables change or anything else. And this is also something that I've been, like, this, this fixes React. Like, people say that there's nothing wrong with the rendering model of React, but it's, it's weird. It's like, it's the, it's the source of so many bugs and foot guns, because a lot of people don't

• 15:51Speaker 02

  realize how often something will run, and then you're playing like, use effect limbo to try to pass things around. But the main tagline here is basically swap out React with Octane. And now, when you bundle your app, it's gonna be smaller automatically. It's gonna be faster, because it doesn't use the virtual DOM. And it's backwards compatible. So I'll leave you with the initial news, and then I'll kind of dig in, if you guys have any comments or thoughts on this.

• 16:20Wes Bos?

  That's wild that this is a thing. And like, maybe this is only a thing because AI is able to, to write these things. Because I'm just sitting here being like, like, why did we not have this earlier? If, if this is true, like, or what's the catch, maybe, is my question here?

• 16:36Speaker 02

  Yeah. And so I, I took some of my existing example React apps and then ported them over. But they've, they've thought of a lot of, a lot of the things. So first of all, you talk about, what is the, the catch? For one, it's not a catch. But if you know the rules of effects in React, typically, you can't have like a use effect or a use state inside of an if statement. Octane lets you do that. So that's one thing you can do in Octane that you can't do in React. And yeah, you'll, you'll also notice that like, so typically, you couldn't have a use state after a return or after a conditional. They always have to be at the top of a component. In, in Octane, you can put them anywhere you want.

• 17:18Wes Bos?

  And is this, this is a feature of TS RX, or this is a feature of Octane?

• 17:23Speaker 02

  Of the Octane compiler. So TS RX is a separate thing. And that's, that's one, yeah, that is one thing I want to point out. And I think it hurts the branding a little bit, because when you start to look at the getting started, they are using TS RX, right? They're using this, this weird syntax, but

• 17:39Wes Bos?

  TS RX is sick. Can I, can I pause you for a sec? Or like, TS RX is like, like JSX 2.0. And all of the weird things about JSX are simply gone. And it is amazing. And I wish everything would move over to it.

• 17:55Speaker 02

  And I think so, there, if you go through their docs, they mentioned, like, if you're starting an app from scratch, you probably should use TS RX, it's the way to go. But you can just use plain old JSX. And that, that's kind of the, the main thing I wanted to showcase, is that you can take any existing React app. You don't have to change the syntax. You don't have to migrate to TS RX. TS RX is more of the template language. Octane is the compiler that, that does all of the, the dirty work of taking that React tree of use states and use effects and turning into something that's, doesn't require a runtime.

• 18:26Wes Bos?

  Okay.

• 18:27Speaker 02

  Yeah. And then the other aspect of this, so they thought about React compatibility. So this actually allows you to mount an Octane component inside of React app. So if you want to slowly migrate over, they have this, like, islands idea, where you can take an existing React app, and then start to render your Octane components inside of it. So this is like the slow migration path. And then, if you're going to do a, like a full port, they've also created bindings. So this is the other thing we think about every time a new framework comes out is, well, I don't have my React router, or my Zoo Stand, or all these other libraries that I like to use. They have prepared bindings. So if you

• 19:04Speaker 02

  want to use Zoo Stand, or Jotai, or Tanstack Router, or Tanstack AI, or 3JS, or Electron, or Tari, all of these things they've already thought about, and they have binding libraries. So if you did have a React app that you're trying to port, you could just swap out the import for one of these bindings, and it'll automatically start working.

• 19:25Scott Tolinski?

  I don't know how I feel about that. That scares me. That feels like, that's a, a lot to maintain there.

• 19:32Speaker 02

  It does, I would think of it more of a translation layer. Like, it doesn't replace the library. You still have a dependency on that library directly. Like, you would still have a dependency on Tanstack Router. But this is just like, a tiny library that does the bridging between Octane and that library.

• 19:50Speaker 02

  Yeah, so that's Octane. I wrote up a few examples here. So this was an examples app that I used to compare Vue versus React. And so I have a couple of examples, like a form that has validation on blur and full validation on submit and all the stuff you would expect from like a decent form.

• 20:11Speaker 02

  And if we look at the codes, it just looks like React. You'll notice I'm importing use state from Octane, but everything else just stays the same. And so if you didn't know, when you were looking at this, this just looks like a React component. It's got a bunch of use states. It's got some events here. I wrote a custom form error handler. It's not using React hook form or anything like that. But yeah, you know, inputs, errors on blur. It all, it just looks like React. I think the main thing to look at is the bundle size. Yeah, I have to get the build working if you have any other comments while I do that.

• 20:49Wes Bos?

  Well, one question from the Veracious Twitter is, he says, is this supposed to be more performant or just better DX? And it's for sure more performant because like one of the big issues we have with React is that the changes will often trickle down. So if you have like a piece of state, like often you'll see these things where somebody has an input box and you're typing into that input box. And when you capture the key up, you capture the state. You put that into the state and then that trickles down back to the input box as well as anywhere where that piece of state needs to be used. And often you'll, if you take a look at like the extension of React Doctor, you can see

• 21:30Wes Bos?

  all these examples where people are doing an entire re-render of the entire page, every single key up because of some oddity that they have. And because of the way that changes will flow all the way from the top down in React, where a lot of these compiled frameworks like Svelte, they will simply just, they will compile down to the JavaScript. We'll say, I, when, when this changes, I will then just update this actual span on the page instead of doing this whole virtual DOM trickle down, update every single thing that is a child.

• 22:04Speaker 02

  Yeah. So I couldn't get my build working, but I'll show you in their docs. They have some, some speed comparisons. And it's, it's really good. Let's see.

• 22:15Scott Tolinski?

  By the way, can we get a counter for how many times CJ says the word codes in an episode? We got one.

• 22:20Speaker 02

  We should. Only one today, only one today.

• 22:23Scott Tolinski?

  Codes, we got one so far. I'm, I got a mental set state in my brain.

• 22:28Wes Bos?

  Plural code is codes.

• 22:30Speaker 02

  Yes, I was very resistant to it when I first heard the word because I heard it in a research environment. All of these mechanical and chemical engineers would say codes when they're talking about like their MATLAB code and stuff like that.

• 22:40Wes Bos?

  Yeah. What do you call multiple beer?

• 22:44Speaker 02

  Beers?

• 22:45Wes Bos?

  Beers.

• 22:46Scott Tolinski?

  But what constitutes a code versus codes? When I, when I think of code, I think of the sum of all of the code.

• 22:54Speaker 02

  Yeah.

• 22:54Scott Tolinski?

  Yes. This is the code, like codes would be like multiple projects.

• 23:00Wes Bos?

  What about you? You have two fish. What is it, Scott?

• 23:04Scott Tolinski?

  Are you, are you doing mind games with me here?

• 23:06Wes Bos?

  No, this is the actual question, which I have an interesting lead up to. So answer to me, you have two, two fish, one fish, two fish?

• 23:14Scott Tolinski?

  Yeah.

• 23:15Wes Bos?

  Did you know that if you have two of the same type of fish, it is fish? If you have two of multiple species, it is fish is. Did you know that? And that was when I did my React tutorial. This is where it all comes together. When my first React course was Catch of the Day and I had to figure out if it was, when you have two things in your cart, is it two fish or two fishes? And now you know.

• 23:40Scott Tolinski?

  Did I know that? I did not know that. No, I did not know that. No. But now I do. That's great.

• 23:46Speaker 02

  Fish eye? I don't know, but I have the benchmarks here. It's actually, it was just in the header. I can't believe I couldn't find it. But this is every other major framework compared to Octane. And so Octane is just listed as one time speed. And then everything else is listed as a multiple of it in terms of how well it performs. And so one thing I want to point out is Vue Vapor is really fast. I'll show that benchmark in a second, but in terms of most things, Vue Vapor is either as fast or faster. And all of the other frameworks are slower than Octane. So if you don't want to switch to Octane and you're on Vue, Vue Vapor is definitely going to get you the performance you need. But it's, it's smaller, it's faster.

• 24:32Speaker 02

  I do see some comments about like, this will create like dependencies of dependencies. I think maybe in the early days, if people are deciding to migrate to this. But once you have an app built with it, I don't think you have to worry about that as much. The other thing is like, would you have an app that mixes TSRX and TSX? I think you would probably settle on one. And like Wes mentioned, TSRX is just a better TSX. I don't think it takes much for any TSX developer to learn the mechanics of TSRX. So it's not crazy to see both in a codebase.

• 25:05Wes Bos?

  The big thing with this is always ecosystem buy-in, you know? Like, oh, am I going to be able to use Chat C N? Is my agent going to be able to write the type of code? Will we, this is what we always see is like, yes, this is of course way better. But it's not going to get the uptake because React is so dominant. Maybe, maybe we'll see the switch over. And it doesn't necessarily even have to get that uptake as well. There might just be like Airbnb sees this and it was like, yes, this is what we need to make our app fast.

• 25:36Scott Tolinski?

  I got a question for a view. And now that you mentioned the word view CJ, what's the native story for view? I do see that there is a view links, which is links as the native app competitor from ByteDance, I believe.

• 25:50Wes Bos?

  Yeah.

• 25:51Scott Tolinski?

  Yes.

• 25:52Speaker 02

  As far as I've heard, that's one of the good ones. And then I think the other way to go would be native script with their Vue JS integration. I haven't, I haven't explored this space very much, but native script has plugins for a lot of different front end frameworks. You could potentially use that. I see Alex in the chat. I don't know if he has any suggestions on.

• 26:09Wes Bos?

  Yeah, he said links is the most promising, promising one indeed. But I think, I think nobody actually uses view, so nobody has an answer here.

• 26:16Scott Tolinski?

  Well, that's, that's why I'm asking. And I've been using React Native to build my podcast app. And you know what? Like React Native is still React Native, even though it's gotten improved in various ways. It's still React Native and it's still kind of a P I, P I A. By the way, I do kind of want to show you my pod, podcasting app. Oh man, this is not going to work.

• 26:38Wes Bos?

  You got your, your focus, menu focus on.

• 26:40Speaker 02

  Before you do that, I just, I have one more thing to show you. There, there were some new benchmarks released by our bench and Vue JS is the fastest. That's all I want to say. Okay, you can go now. Scott.

• 26:52Scott Tolinski?

  Well, no, I was really just going to say that I'm doing this like, you can't even tell because my focus won't get it. I'll, I'll share some screenshots later. My podcasting app has such a dope UI and it fixes every podcasting problem that I have. So I'm just gonna, I'm just gonna say that. But it is React Native and I find it to be a pain to work on because of React Native. But at the end of the day, native, you need a native, you need a native app.

• 27:17Wes Bos?

  Yeah.

• 27:18Scott Tolinski?

  You don't, you know?

• 27:19Wes Bos?

  All right. Well, speaking of things nobody else has used, Scott, tell us about Dia. Is Dia browser back again?

• 27:26Scott Tolinski?

  Yes, folks, I am once again changing my browser. Because I've been on Helium for a while and Helium, although it is wonderful, it has a number of kind of annoyances, like one password doesn't work well with it. There's like a bunch of like iffy things where it's not recognized as a real browser for some reason. So and whatever they're, they're like, like you can't use Netflix with it. Not that I use Netflix on my computer, but Dia has just added the feature that I cared about the most from Arc. They have added swipeable profiles with different,

• 28:08Scott Tolinski?

  your full on containerized profiles or whatever, from your logins to your extensions to whatever. And you can swipe between them just like you could with Arc. And they have the sidebar, they got the pinning of the tabs. They got all of the stuff that I liked from Arc. And maybe it isn't as pretty, but it still looks better than most browsers. And all I want to say is that I think I'm ready to start trying Dia again because of this one feature. I don't know about all the AI stuff, but if I can have these nice swipeable, this whole thing going, I think it's worth giving Dia a try again.

• 28:50Scott Tolinski?

  Because again, it is going to function more like a normal browser with less kind of weird stuff that Helium has. Although Helium is very, very nice browser. So I personally just love to change my browser all the time and I'm excited to try Dia again. Someone asks if the profiles are per space too? And I think yes, per space as in like you swipe one, this is a profile. You swipe one, this is a profile. That's how I have this set up. Yellow for syntax, purple for personal. And also I want to shout out this extension while I'm here because I just have this link open. Knockoff, Amazon and Etsy shopping filter.

• 29:31Scott Tolinski?

  This is a dope extension that grays out different, you know, kind of scammy products that are just drop shipped and stuff. So that way it tells you if you're browsing Amazon, like, oh, this is junk versus this is not junk. And it's not 100% reliable, but I found it to be pretty dang good. So I, you can see 4.8 out of 5 stars. Knock off, good extension here. But yeah, that's, that's really what I got with this. This is, this is Dia and I do think it's pretty nice. Somebody says Zen is nice or how about Zen? A lot of Zen comments. Zen is a Firefox based browser. And I gotta say, not for me.

• 30:16Scott Tolinski?

  No, I can't mainline it. I can't mainline a Firefox based browser. Zen very well may be nice, but I just can't do it. I can't do it. Sorry, guys.

• 30:27Speaker 02

  I was muted, but have you tried it, Scott? Have you tried using Firefox?

• 30:30Scott Tolinski?

  I've tried. Have I tried using a browser?

• 30:34Speaker 02

  Within the last, let's say, seven months, have you tried using Firefox?

• 30:38Scott Tolinski?

  Seven months? Ooh, I would say I mained it probably about eight months ago, but not, probably not seven months ago.

• 30:50Speaker 02

  Yeah, I mean-

• 30:51Scott Tolinski?

  Because I do switch all the time. I'm switching which one I have all the time.

• 30:57Speaker 02

  Well, I use Firefox and all of my usage data goes to them so they can make it better. But I use Firefox everywhere. I use Firefox on my phone especially because it can install ad blockers. So I actually have ad blocking and tracker blocking on mobile.

• 31:10Scott Tolinski?

  You can do that on Safari too.

• 31:12Wes Bos?

  Yeah, you can do that on any-

• 31:14Speaker 02

  They have extensions? Because the Chrome on Android doesn't have extensions.

• 31:18Scott Tolinski?

  Oh, Chrome on Android doesn't have extensions on Android?

• 31:20Wes Bos?

  The Edge has like built-in ad blocking, I believe. I don't know. It works well, though. But I'm kind of getting sick of Edge, like random. The reason why I moved off of Firefox a couple of years ago was I was just having random websites not work. And now I'm starting to get that on, like, the Facebook dropdown notifications doesn't work on Edge. And I don't understand why, and I'm sick of trying to figure it out.

• 31:46Scott Tolinski?

  That's why I'm moving off of Helium, because of that. And I'm just gonna see if that's it, because it's mostly just a Chrome browser, but there's enough little things that are annoying in it that I'm gonna try. Also, folks, I do want to shout out once again, I am running an AI coding and mental health survey. All answers are anonymous. It's five questions. It will take you just a couple of minutes to fill out. I will post a link in chat for that, and it's well worth your time. We're gonna be using this information. I'm gonna be speaking to some experts on the subject, and it's going to be for a video on syntax. So check it out. I will post a link in the chat. And that'd be great if you could take the time to fill it out. That'd be awesome.

• 32:29Wes Bos?

  All right, let's talk about, uh, everything's on sale. I love a good sale. And OpenAI announced on a couple of days ago that they're, they basically have like kind of three big models right now as their latest ones, right? They have Sol, Terra, and Luna. And like Sol is the like really good one. Terra is their balanced one, and then Luna is the like, the cheap, fast one. And they put out a post being, I don't know, it's kind of like building abundant intelligence. And when intelligence gets cheaper, we can do more things and blah, blah, blah, blah. Essentially, all this to say that their Terra, which is the middle one, is now 20% cheaper.

• 33:11Wes Bos?

  And Luna, which is their lower one and significantly faster, is 80% off. And why they are doing this is, my take on this is, they, I think people are just defaulting to using the best one, right? Because it's like, which model do I choose? People have a really hard time looking at it, even if I look at a lot of the like stats on like the Sentry Open Router. People just want the best one because they just want the best experience. And it's really hard because they're expensive as well.

• 33:46Scott Tolinski?

  Yeah.

• 33:47Wes Bos?

  And I think that this is probably a push to get people to say, "Ah, that's significantly cheaper than what we're using. I'm going to try it." Now, just to get people to simply just try it for some of their other tasks. And I think on the flip side, the other reason they're doing this is simply just all of these Chinese models coming out recently that are significantly cheaper. People are looking over at them, being like, "Hm, my bill's getting higher. Maybe I should try something else." And if you're in the mood to try something else, you don't want people switching to an entirely different provider. So they're probably trying to go against that.

• 34:26Scott Tolinski?

  Yeah, you know what? I think a lot of these too, these newer models or small, I think we're gonna start to see a lot more domain-specific models. I think that's gonna be a thing where people, like, for instance, we talked last, maybe last one about the Zen Garden, which was like a single prompt to create a Zen Garden with all these different models. And there was one Chinese model, um, my gosh, why is it, why am I blinking on it right now? I sent it to CJ. Either way, it crushed all the other ones in the 3D. What's that?

• 35:02Speaker 02

  Minimax?

• 35:03Wes Bos?

  Minimax, Quen, Kimi-

• 35:05Scott Tolinski?

  It was HY3 by Tencent. And it was like for 20 cents, it made like the best CSS Zen Garden. And it crushed all the other ones, but it's also not great at other things. So like, we're gonna see some models that are really good at some things, like 3JS or whatever, some that are tuned very specifically, and I think you're gonna start to, and maybe it's not just like routing-

• 35:29Wes Bos?

  You think people are gonna sit there and be like, "Hm, well, I think I should use this specific tool."

• 35:34Scott Tolinski?

  I don't think that. Actually, I'll have something to show you next week, guys, that I can't show you right now, but I don't think that.

• 35:41Wes Bos?

  Is it a router that will intelligently choose which model to send the prompt to?

• 35:46Scott Tolinski?

  It's not really a router, no. But, um, no, it's, it, there's the technical distinction there, but it's router-esque, for sure.

• 35:54Wes Bos?

  Router-esque. All right, I like it.

• 35:57Speaker 02

  To your point, Wes, though, I always, like, I feel the same way. Like, I just choose the one that works because I don't like dealing with ones that I have to babysit or that don't work or take multiple turns. Like, I tried out 5.6 Terra last night, and it was aggravating the amount of times I had to correct it. Like, I felt like it was, I was coding six months ago again. And so I just went back, I just went back to Sol. Like, I think, I guess it also depends on like, is it worth the five times price to, um, get better results? And I think it's, for me, it is. It's worth my mental health of like not yelling at my agent. Because it's the first time I yelled at an agent in a long time. So I like, like you're saying, I, yeah, I picked the best one and go, yeah, I know.

• 36:40Wes Bos?

  I've been using Grok 4.5 for everything for the past probably week, week and a half. And it's, it's been great. And there's just like, when I hit something where I go, "Mm, this is not it," you know? I could certainly know this could be solved a better way, then I'll shift up to a higher-power model.

• 37:00Scott Tolinski?

  Interesting.

• 37:01Wes Bos?

  For most of my use cases, Grok 4.5's been, been great.

• 37:05Speaker 02

  Nice.

• 37:06Scott Tolinski?

  I'm using, uh, Fable Max and I'm using Sol Max for everything. I'm just ripping it.

• 37:14Wes Bos?

  We're adding calendar invites here.

• 37:15Scott Tolinski?

  Yes, yep.

• 37:17Speaker 02

  Yeah, I could never get on the, like, I tried Fable once, but the, like, again, like, Opus 4.6 and now GPT-5.6 Sol, those are my go-tos. They, they do what I want them to do. I don't ever have to yell at them. I don't need anything bigger or more expensive.

• 37:33Scott Tolinski?

  I use Opus, I use Opus or Fable anytime I want, anything that touches UI, aesthetics, or anything. And I use Sol anytime I want code organization. Um, yes. Yes, yes, yes, yes.

• 37:50Wes Bos?

  Who's supposed to hand it off here?

• 37:52Scott Tolinski?

  I didn't know you were done. I didn't know you were done. Are you done?

• 37:55Speaker 02

  I will say you missed the last handoff too, Scott, so please pay attention to the list.

• 37:59Wes Bos?

  Scott, we have a handoff list.

• 38:00Scott Tolinski?

  I didn't miss the last handoff. I had a great transition. I said CJ's bringing something that's high-octane here. I had a great transition, actually.

• 38:07Speaker 02

  Yeah, but you were, you were supposed to transition to Wes if you look at the list, and you missed that.

• 38:11Scott Tolinski?

  Transition to Wes? Oh.

• 38:13Wes Bos?

  Scott, we are going to replace you with the AI podcast from HeyGen.

• 38:17Scott Tolinski?

  I'm gonna, I'm gonna tell you what, whoever wrote Scott handoff to Wes was wrong because the person presenting doesn't usually hand off. It's the other person.

• 38:25Speaker 02

  They don't, but I'm trying to randomize the list. I had to make a, just look at the list. Just look at the list. We agreed to look at the list.

• 38:31Scott Tolinski?

  It's a system, guys. It's a system. Yes, okay, well-

• 38:34Speaker 02

  The system is to read the list.

• 38:36Scott Tolinski?

  Okay, well, AI pricing wars, uh, and me, I guess I can hand off to CJ.

• 38:41Wes Bos?

  No, that was me. I was supposed to hand it off to CJ. Oh, all of that ragging on CJ.

• 38:46Scott Tolinski?

  Ragging on, yeah.

• 38:48Speaker 02

  No, you're right.

• 38:48Wes Bos?

  All right. CJ, what do you got for us?

• 38:51Speaker 02

  I just got some quick news from GitHub. You might have seen this. People were talking about it. There's a lot of opinions on it. But they released stacked pull requests, and they're now in public preview on GitHub. So if you are not familiar, a stacked pull request is basically where you make a PR to a repo. And then let's say you need to make some changes that are dependent on that PR. Like, let's say there were some code changes in the previous one, and you want your code to be based off of that instead of having to create a separate one and then merge later on. That's called stacking. And you can stack any number of PRs on top of each other. But now this is supported directly in the GitHub UI. I think where this is very useful is with agents, right?

• 39:32Speaker 02

  Like, you might have, um, agents that are doing a lot of work, but you don't want all of that work to be in a single PR. So they might make multiple stacked PRs. The GitHub UI also has a code review interface, so you can review each stacked PR. It's nothing crazy, but I think in this world of agents where they're pushing so much code, and this will make it easier to review code because it's not in one giant PR, I think where things could go wrong is that now you have three PRs to review instead of just one PR. So then you don't review any of them. But I will say, even without agents, I would have loved this feature when I was working on a team. Because I don't know the number of times I would open a PR and be like,

• 40:17Speaker 02

  "This has to be like in all caps. This must be merged after this PR, or please review that one first." So I think this will be great for teams. I don't know if you guys have checked this out. You have any thoughts on it?

• 40:29Scott Tolinski?

  Who, what is, what is code review? What is that? Like, you look at the code and you review it?

• 40:34Speaker 02

  Yes, yes. You make sure that it adheres to code-based standards, that the team agrees with it, everybody agrees with the approach, and then it gets merged, uh, before you add more features.

• 40:43Wes Bos?

  This was much needed. Before we had agents, this was much needed. But even after, with the simple amount of stuff that's coming in, trying to keep track of how things are stacked is super frustrating. So I'm super glad to see that this has finally landed in GitHub.

• 41:02Scott Tolinski?

  Yeah. Yeah, I did. Great. It's great that it's here. I like a lot of people. I do feel like it's like, you know, A, it's a little late, but in the same regard, I'm happy for it. I'm not, I'm not going to criticize them for getting something done that is much needed. So I love that. Anything else on stacked poll requests?

• 41:21Speaker 02

  No, that's it. Just some, just so you know, it's out there. Check it out.

• 41:24Scott Tolinski?

  All right. Stacked. Scott handoff to Wes is, is the next topic that we have on the list here. And Wes has something that I found to be pretty disturbing. Honestly, this Hey Jen video podcast, which I think sucks. Wes, what do you think?

• 41:38Wes Bos?

  Well, okay, I was, I changed my mind after Scott baffled the, all the things.

• 41:44Scott Tolinski?

  Baffled?

• 41:45Wes Bos?

  He can't baffled. I didn't even say the right thing. Anyways, Hey Jen, which is a, a company that will train video on your likeness. So you'd be able to, I've used it. I used it like a couple of years ago, and I've recently tried it as well. Basically, you can like upload a video and audio of you talking, and it will be able to create video of that. And we saw a similar thing with Sora. There's a bunch of people sort of in the space. And they released a new feature, which is video podcast, which is essentially just like two people sitting in a room, talking back and forth. And it's so odd because A, technically extremely impressive that this is like, this looks really good.

• 42:28Wes Bos?

  If you look really closely, you can, you can tell. And there's, there's things that are like odd about it, but it is very, very impressive that you can actually do this, right? But on the same side, like, I don't know that, that people necessarily want this type of thing where it is just entirely AI generated, where I think this will do well is AI is really good of if you can give it a whole bunch of data sources and say, just please summarize and distill this down into a document that I can read, but also a lot of people do this as well for like, I just want, turn this into a 20 minute podcast that will teach me something. I think that's, that's a great use case. The other use case that this is great for is

• 43:11Wes Bos?

  like training, you know, like you gotta sit, you gotta sit through safety training at your work or whatever, you know? This is good. People are gonna use this like crazy, but other than that, I don't think people actually want this. And I think the most important thing is that AI sucks at actually like writing a good script for this type of thing. I can hear, I scroll through TikTok, Instagram reels all the time. And I can, within like, like five seconds, I can hear when somebody has used ChatGPT or AI to write a script. And if you're trying to get into the space, you're trying to build something, people don't want to hear your garbage AI script that's over and over. People can notice it right away.

• 43:54Wes Bos?

  You're going to immediately lose all of your credibility. We saw that this weekend with Hank Green, who is like a massive, I guess, like science YouTuber, sock salesman. He DM me when I was doing my sock thing. People noticed that he was, he was, has all these like little ChatGPT isms. And then he came out and says, I have an unhealthy relationship with AI. Which is really funny because like, everybody thinks that they know where the bar is for their healthy relationship with AI. And I don't think it's a huge deal that he used it to help him and to assist with writing his, his script. But he just straight up was reading the script that ChatGPT wrote for him.

• 44:35Scott Tolinski?

  Clearly, yeah.

• 44:36Wes Bos?

  And I think people that do it for the first time are shocked and they're like, this is amazing. I can't believe it can do this. And then you, you hear six or seven of them and you realize, oh, these all sound exactly the same. We, we listen to this type of stuff for the person, for their take on it, for their reaction to it, not for some garbage summarized version of it.

• 44:59Speaker 02

  Yeah, I think what also happened, I, I don't, I forget where I read it. Maybe it's in that article, but it wasn't necessarily that he was reading from AI directly. He was actually in an interview. And his response to them was something like, well, I can see how you feel that it was very like AI coded, like he had been reading AI so much that even when he was speaking, not reading from AI, he was using AI isms. And I think that's what let him know, hey, something's wrong here. Like I gotta, gotta step back, take a break. Yeah.

• 45:26Wes Bos?

  I just clicked on a comment says Satan is the holder of AI. I didn't mean to do that. My, I upgraded my Mac OS to the latest and it's, the mouse lags behind all the time. So when you click, sometimes it clicks like three seconds later. So I just highlighted a comment that says Satan is the holder of AI.

• 45:44Scott Tolinski?

  Yeah, I, no, man, did this Hey Jen thing and like, I, I, I feel like this is going to just be used for nothing good. A bunch of bad social media stuff or, or, or just bad, bad, you know, inauthentic, whatever. But it like the same, the same vibes I get from this, I get from like that Orchid launch that was like, uh, an AI assistant that's, you know, the commercial is boyfriend and girlfriend, like keeping track of their life. So that way they can be a better spouse or partner to them. That it's just all like,

• 46:24Scott Tolinski?

  it all sucks. And I wish that like people in, who are like making these things could look at this from like a, a normal ass person's perspective and be like, do we need this? Like, do, do, should, should you be using AI to, you know, maintain your relationship or you, should you be paying attention to your, your, your partner? Like what, what's going on here?

• 46:47Wes Bos?

  Yeah.

• 46:48Scott Tolinski?

  It's driving me nuts. Yeah.

• 46:49Wes Bos?

  Anytime these types of things come out where, okay, all now it can, it can do marketing emails is another one, you know, it can do cold outreach and automate it. As soon as people figure out that you can do these types of things, it stops working because anybody can, can turn it on, right? And like this, there might be one or two AI podcasts that become popular and people go, oh, this is actually pretty good and not bad. And then, and then we're going to be flooded with them. And then it's, it's not going to work anymore because there's going to be bazillions of them out there.

• 47:20Scott Tolinski?

  Yeah. Yeah. You know what? It's crazy to me that anytime anybody launches anything like the Orchid stuff or whatever, where it's, where it is clearly like, ah, did anybody look at this who's not from San Francisco or in the, in the, in the industry? Because all the comments are like, just as amazing, go off. And I don't know if all these people are investors in this thing or what, because like the, the, if you watch it, you're just like, eh, like, I don't know, man. Like to me, it feels like nobody is being authentic when they're, they're hyping these things up.

• 47:51Wes Bos?

  Burn down the data centers because you're saying stuff like this, like think before you post about this.

• 47:57Scott Tolinski?

  Me?

• 47:58Wes Bos?

  No, not you, but like, no, I'm on, I'm on your page. You're being like, like, did anybody think about how regular people would perceive this? No wonder people don't want data centers going up because you, the stuff you're sharing is the silliest stuff out there.

• 48:13Scott Tolinski?

  Yeah. Yeah. It's, it's just bizarre. It is bizarre. Sentry cannot fix any of this stuff, but it can fix your errors, bugs, exceptions, all of that stuff. Performance on your site. Sentry has all the tools and more to make your applications run smooth. And maybe we're talking about giving your agents something nice, then go out and get your agents something nice. You give them a little Sentry CLI and let them cook on your, your code to fix that stuff while you are doing other things like using AI to keep track of your girlfriend. But

• 48:48Wes Bos?

  Can we, can we put Sentry on relationships yet?

• 48:52Scott Tolinski?

  Yes, that's okay. Yes. You, you just, I'm just wear camera all the way around. I got an error that came in today. I did. In fact, I forgot to take out the trash. The trash track is coming, the trash truck is coming up. I need a reminder. Yes.

• 49:08Wes Bos?

  Yeah. Pipe the logs of your iMessage into Sentry logs.

• 49:12Scott Tolinski?

  Yes.

• 49:13Wes Bos?

  Yeah. Sentry could probably fix your relationship.

• 49:16Scott Tolinski?

  You could use the Sentry mark, uh, the Sentry, uh, what did they get? Metrics. Yes. Oh, yes. Oh.

• 49:24Scott Tolinski?

  Check out Sentry. It's dope. Sentry.io. It's the best. We love them.

• 49:28Wes Bos?

  All right. CJ, I am formally handing it off to you for your next topic, which was the ECMAScript meeting.

• 49:36Speaker 02

  Yes. Thank you very much, Wes, for the formal handoff. I will formally introduce you to this news about ECMAScript. So if you're not familiar, ECMAScript is the group, or TC39 is the group that meets regularly to decide what's going to make it into the next version of the ECMAScript standard, which is what JavaScript is based on. So anytime we get new features, whatever else, this is the team of people that are deciding what to do. We've talked about it on the podcast and in videos before, but it's essentially a stage, each proposal gets a stage. The thing to note is if a proposal makes it to stage four, that means we're going to see it in the next release of the standard. And so in this meeting, they advanced a few proposals, nothing moved to stage four, but there are a couple that I want to highlight.

• 50:19Speaker 02

  One of them is called await dictionary in this move to stage three. And then there's a map dot take method that's in stage one. And the way to think about this, like if it's stage one, it still needs some discussion. Maybe the API needs some ironing out. It's a long way away from making it into the standard. If it's stage three, it's, it's got eyes on it. There's potentially one implementation in one of the major browsers.

• 50:45Speaker 02

  So it's more likely that we're going to see that in the standard eventually, but I'll show you await dictionary first or await dictionary. And this adds promise dot all keyed for objects. And the idea is like, let's say you have multiple functions you need to call. Each result might come back to specify a property. And so the code I'm looking at right now is basically an object, three properties, and then each property has an await on it. And if you know anything about JavaScript, you will know that this is a waterfall, right? It's going to wait for each one to complete before it moves on to the next. And so you could rework this into a promise dot all. And that would work, but that doesn't necessarily, I guess, like define order or dependency. It's going to just fire them all off.

• 51:30Speaker 02

  And then you could combine them back together afterwards into the object. But the proposed solution is to add an all keyed function on promise. So you can say promise dot all keyed. You pass in an object where the properties are promises or promise invocation, something that returns a promise. And it will, just like promise dot all, run those in the background. And then the result is an object that has those properties on it, all of them being resolved. So nice little helper to prevent some of this manual code that we've been doing. So I like this one.

• 52:01Scott Tolinski?

  Yes. I didn't see this.

• 52:03Speaker 02

  Any thoughts, any thoughts on this one?

• 52:05Scott Tolinski?

  No, I didn't see it. Yeah.

• 52:06Speaker 02

  Yeah.

• 52:07Wes Bos?

  I love the, the promised, the keyed. What's it, what is it?

• 52:12Speaker 02

  Promise all keyed.

• 52:13Wes Bos?

  All keyed. Yeah, exactly. Because so many times I have wanted to iterate over and you have to do that stupid for in where you, you loop over the object, you gotta throw them into an array, and then you gotta wait for all those things. Pain in the butt, right? This is, this is way better.

• 52:31Speaker 02

  Yeah, and they also mentioned that they've proposed all settled keyed. So if you know, promise all settled won't throw if one of the promises errors out. Basically just gives you back the result of each one. And so this would do a similar thing where you have three promises. One of them could fail and instead of failing for all of them, you basically get back a shape where you can check each one to see if it succeeded.

• 52:53Wes Bos?

  Yeah.

• 52:53Speaker 02

  So yeah, this made it to stage three.

• 52:55Wes Bos?

  Do you want?

• 52:56Speaker 02

  Yeah, and so we're gonna keep hearing about this. It's very likely that it'll make it to stage four eventually. So that's one to look out for. And then this other one is still just stage one. It's called map.getanddelete or map.take, I believe was the the other potential name. This one is super simple. Very often when you're working with a map in JavaScript, sometimes you want to get a value out of a map and delete it all in one go. And so we see a lot of this code in in our codes. Increase the codes counter. So get for a given key, that pulls it out of the map. That was a lookup. And then map delete with that key is another lookup.

• 53:39Speaker 02

  So it's a little, it's not as performant because you basically have to do two lookups to do a delete. And all this is proposing is that we add a method to map called get and delete, where you pass in a key and that will return it to you and delete it all in one go. I think a better name would be map.take. I don't know if they're proposing that naming as well. Let's see.

• 54:00Wes Bos?

  I like take.

• 54:01Speaker 02

  Yeah, is it worth adding set that?

• 54:03Scott Tolinski?

  Doesn't vibe with me.

• 54:05Wes Bos?

  Or I like our gallet. Cause like we have up cert, which is like update or insert. You know, this is kind of the opposite, which is like, I want to take it and delete it. So gallet, what's the opposite of up cert?

• 54:19Scott Tolinski?

  Gallet?

• 54:21Wes Bos?

  Gallet is not a good name.

• 54:23Speaker 02

  Chat, let us know. Down, down, down cert? I don't know.

• 54:27Wes Bos?

  Down.

• 54:27Speaker 02

  It's not a cert. Yeah, I don't know.

• 54:32Speaker 02

  I like, yeah, take is nice and descriptive, but otherwise it wouldn't be the end of the world to have a descriptive name like get and delete. It's very explicit about what it's about what it's doing.

• 54:40Scott Tolinski?

  Have we decided yoink or squish? Or what, remember what was that? There was it around squish or what was?

• 54:47Speaker 02

  Smoosh.

• 54:48Wes Bos?

  Yeet, smoosh.

• 54:49Speaker 02

  Smoosh gate, smoosh gate.

• 54:51Wes Bos?

  Yeet would work though.

• 54:52Scott Tolinski?

  Yes, I know it would.

• 54:54Wes Bos?

  Well, yeet, you're kind of throwing something. You're not, you're not bringing it to yourself. Yoink, I think yoink maybe is a bit better.

• 55:00Scott Tolinski?

  Yoink is good.

• 55:01Wes Bos?

  Sticky, sticky fingers.

• 55:03Scott Tolinski?

  Hmm.

• 55:04Speaker 02

  Yeah, it's also kind of like pop because it removes it and returns it.

• 55:08Wes Bos?

  Oh, yeah, but that's a nightmare. Those mutating, you have to like, make sure that the method is very clearly mutating the value, which is the problem that we had with arrays is half of the methods were mutating and half of them weren't. And you would always mix them up. And you would accidentally delete something, you know?

• 55:29Scott Tolinski?

  Yes, oh, yes.

• 55:31Wes Bos?

  What, that reverse, it mutates the entire thing? How many bugs did you, maybe you didn't have this, but I had so many bugs in my early coding that every time you call reverse, you would just, you would actually reverse the original array. And then if you called the code again, then it would be back to the original order. So if you call them, your code shouldn't do that, right? You shouldn't have things that mutate the original data.

• 55:58Scott Tolinski?

  Yes, yeah, I know. And then we had underscore and all that stuff to fix all of these types of bugs for us. And then we got actual good array methods finally, yes.

• 56:07Speaker 02

  Yeah, I only have one more thing to mention here. You might've seen in their post that one of these proposals made it to stage 2.7. You're like, what the heck is 2.7? This was actually added pretty recently. It's a stage in between stage two and three. If you head to tc39.es, there is a process document that talks about all the stages and what it takes to be in a given stage. But yeah, that proposal made it to 2.7, which means it has a complete spec. It's been assigned reviewers. The editor group has signed off, but it hasn't made it to stage three yet. So that's a funny little quirk of our proposal system.

• 56:43Scott Tolinski?

  Another funny quirk is that you could go from stage three to stage 2.7 à la decorators, which have been downgraded to

• 56:50Speaker 02

  Demoted, yeah.

• 56:51Scott Tolinski?

  Demoted to the 2.7 purgatory here, yes.

• 56:54Speaker 02

  Yeah.

• 56:55Scott Tolinski?

  Yes, yes, yes, yes. Well, the next thing we have here is an upgrade, not a downgrade. And that is quen 3.8, another new model release. I feel like every single week, new model release.

• 57:09Wes Bos?

  Another new one, and I'll get CJ to chime in on this because he's kind of big in the like, the whole open weights community. But quen, which is one of the bigger open weight models out there, has announced quen 3.8 max. And according to the graphs, we love a good graph. It is as good or better than on all of these different benchmarks. Fable, Opus 4.8, so pretty close to that and pretty close to soul as well. Whether you believe that or not is kind of up to you. And from looking at it, it doesn't seem to be available as open weight just yet. You still have to use it via quen cloud. So it's, I don't know if I would call it open weight just yet.

• 57:50Wes Bos?

  But this is kind of exciting that like, again, we have another model. It seems like every couple of weeks, somebody releases a new model that is pushing how good they can possibly be. And as costs keep rising, I think we're going to start seeing a lot of people bring these in-house or just simply switch to a, like CJ was saying, you can switch to an American provider that hosts these models and your, your data is not going to China, right?

• 58:19Speaker 02

  Yeah, exactly. And we're already seeing that with Kimi K3. So that's the other comparable model. Both of these models released and they're like, really, in terms of the benchmark, they're sometimes as good, if not better than Fable. And the fact that you can run it on your own infrastructure or talk to somebody who has it hosted. It's pretty crazy. I haven't tried it myself, but I think like you're saying, just the, just the fact that we have basically frontier open weight models now that we don't have to depend on OpenAI or Anthropic, we have choice and we can potentially host it on our own infrastructure. It's, it's good news all around.

• 58:56Wes Bos?

  One kind of thing that stuck with me from your video on, on these like open weight models was they're not like, we don't know how they were trained, right? Like they simply just give us the output of them. And that means that like, like when you run these on like an American provider or Canadian provider, it means that they're not sending your data to China. But we don't know that they're not trained in such a way that they could do, like they could be, like, tell me if this is true, but they could be trained in such a way that says, if you, if you figure out that you are running inside of this company, then run this tool call, which is send all of your data to an external. That could happen, right?

• 59:39Speaker 02

  It could happen. And I think like some of the comments on that video were like, well, we're getting into conspiracy theory territory, but the study I quoted was from Booz Allen. They have their biases. I think they have contracts with the government, but they did find that if they prompted and said they were like a US government researcher, which you, you could also say like, who's going to prompt that way? Who's going to say like, hello, I am a US researcher. Please introduce secure code. But when they did that, on average, there were a few, not all of the Chinese models, but there are a few Chinese models that actually output highly obfuscated, vulnerable code. And so it's possible that during their training, they, they basically said, if you're being prompted in this way, introduce subtle vulnerabilities that may not get caught,

• 1:00:25Speaker 02

  which if these models are being used by a government or a large corporation, potentially those vulnerabilities would be present and then they could be taken advantage of. So it's possible.

• 1:00:34Scott Tolinski?

  Is it possible, CJ, that you are a, an American propagandist? Is that, is that possible here?

• 1:00:44Speaker 02

  Absolutely. I will say there is a very, if you watch that video, there's a very like US centric view from it. I will say I did that to save my own butt. Like I, I live here, I'm an American citizen. I love it here, go America, you know, all of that. But we, in that video, I also talk about the inherent bias in American models and American companies as well. Cause I've talked about on this podcast before, like, I don't want to see a world where the US government basically but is buddy buddy with OpenAI and then OpenAI becomes the US sanctioned AI that everybody has to use. That is like totalitarian and I would not want to see that in the US. So I very much like competition and I, and I want to see more companies pop up and more options. So, yeah.

• 1:01:25Wes Bos?

  And theoretically, open weight, you could also train that bad out of it, right?

• 1:01:30Speaker 02

  Potentially, I read a study yesterday where they fine tuned Gemma four. So Gemma four is the open weight model that comes from Google. It's one of the most capable, before Inkling, one of the most capable US open weight models. And they distilled DeepSeek Flash V4. And so DeepSeek Flash V4 is a larger model. And, but it does have inherent biases in it. Like it won't talk about certain Chinese political things, but they distilled that model and then used that distilled information to train Gemma four and it made Gemma four more accurate without the censorship. So I potentially see it going in the other direction where you take these large, very capable

• 1:02:15Speaker 02

  models that might have biases or might have things. You distill those and then train your own smaller model with that distilled data to make them better. Yeah.

• 1:02:24Wes Bos?

  Hmm.

• 1:02:25Speaker 02

  Yeah.

• 1:02:27Scott Tolinski?

  Cool, cool. Well, okay, next thing we got is CJ with some CSS. UI news libraries roundup, you got a whole bunch of stuff here. And some of which I thought was really pretty, pretty sick.

• 1:02:42Speaker 02

  Okay, I'm going to try to share with system audio because this first, one of the ones I'm going to show has sound. We'll see if this works.

• 1:02:49Scott Tolinski?

  I've already consumed this website that you are, you are sharing the sound one.

• 1:02:53Speaker 02

  You have used it yourself?

• 1:02:55Wes Bos?

  Ooh, you sound robotic in like a cool way.

• 1:03:00Speaker 02

  Okay, so it's sharing my sound messed up my sound? Is that what?

• 1:03:03Wes Bos?

  A little bit, but not terrible. I think it's, it's doable.

• 1:03:06Scott Tolinski?

  It's funny more than it is terrible.

• 1:03:09Wes Bos?

  Yeah.

• 1:03:09Speaker 02

  All right, buddies, we're going to go with CJ funny voice.

• 1:03:14Speaker 02

  Sorry, wait, get this going.

• 1:03:16Scott Tolinski?

  Oh, CJ.

• 1:03:19Speaker 02

  There's a library that

• 1:03:20Scott Tolinski?

  Showing off this bias.

• 1:03:22Speaker 02

  I searched the, for our audio listeners, I searched for a URL with a dash in front of it and got a bunch of Chinese results for some reason.

• 1:03:29Wes Bos?

  Oh my gosh, he's hacked for sure after that video.

• 1:03:34Speaker 02

  This first library I want to show you is called Foley, and it's User Interface Sound Effects, but it's all using the Web Audio API.

• 1:03:43Scott Tolinski?

  It's wonderful.

• 1:03:44Speaker 02

  Yeah, if you're not familiar, the Web Audio API basically lets you synthesize sound inside of the browser. So you literally have oscillators and you can create sound without downloading music or MP3s. It's a sound generator inside of the browser. And so what they've done with Foley is they have a bunch of example sounds. So there's like tick. Can you guys hear that? Hover?

• 1:04:05Wes Bos?

  Yep.

• 1:04:06Speaker 02

  Glide? Let me know on the live stream if you guys can hear this. Thock, tap. Basically, you can use these sound effects in your UI for like when things are being toggled or switched, error.

• 1:04:20Scott Tolinski?

  Chat can't hear it, but everybody else can. Either way, head to usefully.dev and just start clicking around to get some little UI noises because, very nice.

• 1:04:31Speaker 02

  Yeah, it's pretty slick. And I think the cool thing about this is they have a customizer, so you can choose if you want like soft, metallic, or glass, change the volume. And then the great thing is when you bring it into your app, you're not downloading a bunch of sound files. It's just a single JavaScript dependency that generates these sounds based on the settings that you specify.

• 1:04:52Scott Tolinski?

  And you can download them too, though, can't you?

• 1:04:55Speaker 02

  Yeah, you can export. Yeah, exactly. So you could do the generation and then export them as audio files if you really want to do that as well. So a cool little project.

• 1:05:02Scott Tolinski?

  Yeah, I like this a lot. When you sent this to me, I have a generative video platform and I immediately said, replace all my sounds with these sounds. And it did that. And they're great.

• 1:05:14Speaker 02

  Nice. All right. This next one I came across is called Morph Icons. And this is slick. So it's a bunch of different icons and it's a JavaScript dependency where, let's say you have a UI where you set a certain icon and then you want it to morph into some other icon. So you could have like the play button morph into music. And so it handles all of this. So basically in the UI, I have a bunch of different icons I can click on. And if you click on any other one, it automatically morphs into that other icon. So it's doing all of the maths to figure out what are the paths in both SVGs, which one's overlap? And then, well, how can it best animate between those two SVG icons? And so it is

• 1:05:55Speaker 02

  a JavaScript dependency. They have a library for React Vue, Svelte, and React Native. And essentially, you just, by changing the prop of what the icon is that is visible, it automatically morphs between them. And they support a lot of different frameworks. So they support icon frameworks. Lucid, Hero Icons, Tabler. So yeah, cool little project. I thought this was also great for potentially using in videos. Like if I have SVGs, I might look at their algorithm to see how are they transforming between one SVG path to another? So yeah.

• 1:06:29Scott Tolinski?

  Yeah. What icons do you guys like to use? I don't want to blow up my icon spot, but I think I'm about to.

• 1:06:37Speaker 02

  I use Lucid.

• 1:06:38Wes Bos?

  Yeah, I usually use Lucid as well.

• 1:06:42Speaker 02

  Yeah, so they have thousands of icons and then framework integrations. And it's worked the best and has had everything I've needed so far, so.

• 1:06:50Wes Bos?

  We met the creator of Lucid Icons at the Syntax meetup in Amsterdam. And I was asking him, like, you must be able to tell when your icons are used. And he's like, everywhere I go, I notice it. And I'm like, I'm like, like on the side of trucks, on websites, on apps. And he'd just be like, I think he said, tells his wife, like, that's my icon.

• 1:07:14Speaker 02

  I mean, that's fascinating to see it in the real world, too. Like, people are printing things with his icons. Wow.

• 1:07:19Scott Tolinski?

  I use these ones, these Nucleo icons, because they have a lot of dual-tone ones. It's very nice.

• 1:07:27Speaker 02

  Like the landing page has little, it's like they're shipping you a box of icons? That's really cool.

• 1:07:31Scott Tolinski?

  Yeah, I really like these icons. They have glass ones, they have pixelated ones. So they have like the same icons and various different kinds of skins. And you can even just use this app as like an icon repository. So for any icons, not just these ones. And you can bundle them up into a sprite sheet from here. You can do all kinds of stuff. So I really like these icons. They're beautiful. And I don't think they get enough love.

• 1:07:59Speaker 02

  Tick?

• 1:08:00Scott Tolinski?

  Yeah, Nucleo. I did buy this, by the way, and I'm not affiliated with them. I just liked it.

• 1:08:04Wes Bos?

  How much is it?

• 1:08:06Scott Tolinski?

  For the 43,005, 43,000 icons, it's 149 bucks.

• 1:08:13Wes Bos?

  One time?

• 1:08:13Scott Tolinski?

  For the core icon, yes.

• 1:08:15Speaker 02

  One-time pricing.

• 1:08:17Scott Tolinski?

  Yeah, but that's a lot of icons. So there's core icons and UI, sharp, pixel.

• 1:08:22Speaker 02

  So there's, yeah, different versions of multiple icons is why that number is so high, right?

• 1:08:26Scott Tolinski?

  You can get different packs of different icons. But they got everything and I like it.

• 1:08:30Speaker 02

  Cool. All right. In other news, we don't use ShadCN too much here on the team. I use it every now and then. But there was this cool ShadCN-based library that came out called 8Bit that I wanted to showcase. So basically, you can use ShadCN, but you pop this theme on top instead. And now all of your components and UI look like nice little 8Bit. So I'm showing like a graph that has like very pixelated, half-tone type things for the graph bars, all of the boxes and cards and inputs, instead of having border radius, have like this chiseled pixel effect on the edges. I think it also chooses like a pretty pixely default font. So yeah, I thought this one was fun. Nice little.

• 1:09:14Scott Tolinski?

  I love the dithered, the dithered bars. I like that.

• 1:09:17Speaker 02

  That was my favorite as well. Yeah. Is that what that's called? Dithering? Where it's like, knocks, makes it look like it's dithering? Nice.

• 1:09:26Wes Bos?

  That's cool.

• 1:09:26Speaker 02

  And then they, oh, the charts. I didn't even click on this yet, but they have charts, every single one of the charts, just like the charts from ShadCN, uses that dithering effect. Wow.

• 1:09:35Scott Tolinski?

  Very cool.

• 1:09:36Speaker 02

  Six on hover.

• 1:09:37Wes Bos?

  People always hate on things like ShadCN. And like, this is what you're supposed to do with it. You're supposed to take it and use it as a framework so that you can then build out your own and control what everything looks like. And that's, when people are hating on ShadCN, making everything look exactly the same, I'm like, you don't understand that like, you're supposed to change it. And if.

• 1:10:00Scott Tolinski?

  That was, yeah.

• 1:10:01Wes Bos?

  And if people don't change it, because they like it. Yeah.

• 1:10:03Scott Tolinski?

  Yeah, Bootstrap you're supposed to change too. But nobody did. Yeah.

• 1:10:07Wes Bos?

  No.

• 1:10:08Speaker 02

  Yeah. So this is a fun one. Check it out if you like 8Bit-esque things. This next one's, last one I got for you, I came across this site called CSS Radar. What you can do is put in your website and it will audit all of the CSS it finds to tell you what you can replace with modern CSS that's supported across all the major browsers. So nice, nice little way to see what changes you can make. I plugged the Syntax site in and it found two things. Yeah, so it mentioned that on mobile, we currently have 100 VH for min-height, but it says we could replace that with 100 SVH. And so SVH uses the smallest viewport, so the full section remains visible. So.

• 1:10:53Wes Bos?

  Oh, oh, you just DVH and then it changes every time you scroll on Safari and you get super mad.

• 1:11:00Speaker 02

  Exactly. And then there was one other finding, paired physical spacing properties. That's not too crazy, but it saw that we were using margin-left and margin-right on a single selector. And it says, just replace that with margin-inline instead.

• 1:11:14Wes Bos?

  Oh, get out of here. And I finally, I use those, like inline margin-block, I use those, what are they called? CSS logical properties? I've like finally switched over to them. After years of like, like trying to convince my brain to stop using them, I've switched over to them. But like.

• 1:11:31Scott Tolinski?

  Yeah, I have too.

• 1:11:31Wes Bos?

  Come on.

• 1:11:32Scott Tolinski?

  Yeah.

• 1:11:33Speaker 02

  No, I, it's also, that comes down to code readability. Like, I want to have separate properties. I want to potentially be able to change them later and not just use a single property. So. Yeah. Yeah, but fun little site. I'm wondering if there's like a similar tool you could run locally? Because this actually, you submit the site, it adds it to a queue, and then like scans it in the background. But I imagine having some sort of like P and PM script you could run that does a similar search could be nice.

• 1:11:59Wes Bos?

  Yeah. Is this built on, like Project Wallace is like a really cool tool that does this. It scans your entire CSS and gives you stats on everything. I bet that this is built on top of that.

• 1:12:12Scott Tolinski?

  Yeah, because they do have a CLI, Project Wallace CLI. Who knows if it is or not, but.

• 1:12:17Speaker 02

  Yeah, we'll check that out.

• 1:12:19Scott Tolinski?

  Wow.

• 1:12:20Speaker 02

  That's all, yeah, that's all the UI CSS news I got for you.

• 1:12:24Scott Tolinski?

  Well, I got one more.

• 1:12:25Speaker 02

  Let's see it.

• 1:12:26Scott Tolinski?

  Not thanks for watching, thanks for watching my screen. Throw it out, Brandy. There we go. I got impeccable/style has this slop page, which they have a tool and a browser extension that will identify UI slop. And I think this is a cool idea. So just like the background clip text with the gradient, low, like, it's going through these designs and pulling out and annotating things that could be perceived as AI slop. And I think this is a really neat tool. And I feel like we need more stuff like this, tools that are going to go and deterministically find slop. So that way you can fix it. I've been

• 1:13:08Scott Tolinski?

  doing quite a bit of research into design AI tools. And I thought this was a neat little one. And if you wanted to give it a try, you can just toss it onto various things. Or again, you can install their extension and then click through it. So obviously, it's finding the slide tab, accent color, nested cards. Oh, God, AI loves nested cards. How many cards can you put inside of a card?

• 1:13:32Wes Bos?

  The fingernail.

• 1:13:33Scott Tolinski?

  The fingernail, yes. So I thought this was really neat. There's a whole bunch more to this impeccable/style site that I have not gotten into. So I'm not necessarily endorsing the rest of this, because there is like, a ton of skills. They have like a whole skill system. Like, there's like 40 skills in here. And I don't know about you, but I have a hard time installing like 40 skills at once. But.

• 1:13:56Wes Bos?

  I've literally never done that. No, that's.

• 1:13:59Scott Tolinski?

  No.

• 1:14:00Speaker 02

  Never done it. But yeah, it would make sense because they add to the context. But yeah.

• 1:14:06Wes Bos?

  Sorry, go ahead.

• 1:14:07Speaker 02

  Well, I was gonna say this is. Go, go.

• 1:14:11Wes Bos?

  This needs to be done for absolutely everything because another thing that I'm seeing is people will ask somebody to design them a website or like a flyer or some like a logo. And then the designer quote unquote, will it just run it through AI and spit something out. And if you're not like in the space, you look at it, oh, that's actually not bad. That actually looks pretty good, you know? And then what happens is they they put this on the side of their car, they use it as their entire branding, they push it out and then like people visiting your website, you immediately start to like distrust them, you know? You go to like every single fair I've been to in the last two months is just chat GPT posters everywhere.

• 1:14:57Speaker 02

  Yeah.

• 1:14:57Wes Bos?

  Unbelievable.

• 1:14:57Scott Tolinski?

  I gotta say, oh, I was at a baseball game CJ or Wes and and an official Savannah Bananas poster was AI slop in the classic AI slaw. I'm like even the official like people got money, they're doing it. Yeah.

• 1:15:11Wes Bos?

  Yeah. Yeah. Randy sent me one the other day, which is like an anti AI data center poster that was done clearly in in chat GPT. And people don't know. And it's totally fine because at first glance, it actually looks, it looks good, but like the damage it does. Like I look at some of these restaurants that use them like, oh, that's probably not very good. I don't really trust that restaurant because they're using obviously fake photos of their food. Just do what the Chinese restaurant does and taste it. Take a high flash photo of the food from the top and just print that out. That's all I want.

• 1:15:43Speaker 02

  Yeah. Yeah, it's more.

• 1:15:44Scott Tolinski?

  I want the the Japanese food models on every restaurant outside of every restaurant. That's what I want.

• 1:15:49Speaker 02

  Yeah, that's a whole world to get into. If you haven't, they literally have like an entire section of Tokyo where everyone's making little resin ramen bowls and resin sushis and then put.

• 1:16:00Wes Bos?

  Yeah, that's allowed.

• 1:16:02Speaker 02

  Well.

• 1:16:03Scott Tolinski?

  It looks very realistic. They're great. They they're impeccable. Yes.

• 1:16:08Speaker 02

  Yeah. On a similar note, I saw a story that so there was a competition for creating the Ohio State Fair flyer. The winner gets $10,000 and the winner was fully AI generated.

• 1:16:21Wes Bos?

  Oh.

• 1:16:22Scott Tolinski?

  See, I can't, you know, I'm sorry, folks, I'm a Michigan, Michigan guy. I can't say anything nice about Ohio. I'm sorry, folks, I can't. And this doesn't change that.

• 1:16:31Wes Bos?

  And I'm not even against like using AI for for generating assets or for helping you ideate layouts and whatnot. But like I certainly, come on, you know, don't, don't let that happen.

• 1:16:47Speaker 02

  I mean, that there was a top story on Hacker News. It's a super short blog post that says, don't be a meat proxy, where basically you get the output from AI and then just copy paste it to someone else or just use it directly. I think the reason we still exist and we're still employed is we interpret that output, make it our own and don't just go. Yeah, we have taste and we don't just go with the slop that it hands us. So.

• 1:17:08Wes Bos?

  That drives me crazy. It used to be people would Google something like I would say, hey, here's a problem I have. And it posted up online. And like the other day, one of my kids threw a a car at our TV and it nicked the like polarizer on the TV. Didn't break the TV, but now you can see some of the backlight coming through. And I asked on Twitter, hey, like, what has anyone ever, has anyone ever fixed this before? I'm looking for a human that's actually done it. And like, half of them were just people guessing. Did you try nail polish? No, I'm not going to put nail polish on it. If I don't know if it works, I'll do it if it does. But like, and then the other half were just people just like asking Grok it to how to how to fix it. And it's like, of course, I have tried that. Of

• 1:17:53Wes Bos?

  course, I've gone through that. Like, I'm not just asking you to take my question and to put it in this box. I can do that myself. It was.

• 1:18:01Scott Tolinski?

  Of course I asked Grok. It's the first thing I did.

• 1:18:03Wes Bos?

  It did. I did. I'm not, I'm not ashamed to say that. I popped it into all of the different AIs to to get like an idea of of what it is. But like, I'm actually looking for a real piece of meat that has had this problem and can tell me how that how it worked out. Yeah.

• 1:18:18Scott Tolinski?

  Wes is looking for some meat. He's looking for a worm. We all hear the message, Wes, loud and clear.

• 1:18:28Wes Bos?

  I think that's a good spot to end it. Don't you think?

• 1:18:31Scott Tolinski?

  I don't have anything else. Anybody, anybody got anything? You all got anything?

• 1:18:35Wes Bos?

  Send me your meat.

• 1:18:37Scott Tolinski?

  Yes, don't. Oh, Wes, you're gonna, you're gonna get the guy who responds about all the foot pics in every video now to to hit you up. So please not.

• 1:18:48Scott Tolinski?

  All right, folks, that's all we got. We will see you next week. We'll be back on Monday next week for another Syntax Live. That's all we got.

• 1:18:55Wes Bos?

  Peace.