TMiR 2025-11: Cloudflare outage, ongoing npm hacks, React Router is getting RSCs: täielik transkriptsioon

I ER's performance. 00:03:34.813 --> 00:03:39.493 And I spent well over a hundred hours collectively across September 00:03:39.493 --> 00:03:43.963 and October, deep diving into Immers code base, comparing it to other 00:03:43.963 --> 00:03:46.523 libraries understanding how it worked. 00:03:47.060 --> 00:03:51.620 And trained to do a bunch of performance optimizations and so ended up filing 00:03:51.620 --> 00:03:57.040 a few different prs Imer 10.2, shipped last month with some small tweaks, and 00:03:57.460 --> 00:04:02.470 Michelle West Rate just merged a big architectural change and released it as 00:04:02.500 --> 00:04:04.870 Imer version 11 just a couple days ago. 00:04:05.410 --> 00:04:09.760 And then I was able to put out RTK two point 11, which picks that up. 00:04:09.850 --> 00:04:13.240 So free performance wins for Redux toolkit and nier. 00:04:13.635 --> 00:04:14.270 2-vcarl: Heck yeah. 00:04:14.330 --> 00:04:16.880 That was what was the average speed up? 00:04:16.966 --> 00:04:21.736 1-acemarke: the current builds average about 20 to 25% across all the different 00:04:21.736 --> 00:04:24.186 benchmarks varying by scenario. 00:04:24.546 --> 00:04:29.526 And then there's another outstanding PR, which adds an optional array 00:04:29.526 --> 00:04:34.656 methods override plugin, which drastically speeds up array methods. 00:04:35.251 --> 00:04:35.521 2-vcarl: Heck yeah. 00:04:35.671 --> 00:04:36.241 That's awesome. 00:04:36.301 --> 00:04:41.701 We love a core performance improvement and I, I wanna shout out the ecosystem 00:04:41.701 --> 00:04:43.501 performance, you know, I don't know, 00:04:43.561 --> 00:04:45.451 1-acemarke: E 18 E folks are awesome. 00:04:45.481 --> 00:04:51.091 They have been doing a ton of work to clean up large dependency trees and slim 00:04:51.091 --> 00:04:55.846 down packages and remove dependencies across a wide variety of tools 00:04:56.686 --> 00:04:56.986 2-vcarl: Yep. 00:04:57.196 --> 00:05:01.636 a stat that I believe we'll talk about later, I saw in general 00:05:01.636 --> 00:05:03.206 ecosystem commentary yeah. 00:05:03.206 --> 00:05:06.536 I guess in the NPM supply chain attack that we'll talk about later, a a a 00:05:06.536 --> 00:05:09.716 stat that came out of a blog post from that was that the average JavaScript 00:05:09.716 --> 00:05:14.576 project pulls in 683, I believe in was it transitive dependencies? 00:05:14.636 --> 00:05:15.686 Just like, yeah. 00:05:15.716 --> 00:05:16.226 It's a lot. 00:05:16.286 --> 00:05:16.796 So 00:05:17.036 --> 00:05:17.696 when you've got that, 00:05:17.696 --> 00:05:21.056 many dependencies, performance of the ecosystem is challenging. 00:05:21.086 --> 00:05:21.296 Yeah. 00:05:21.593 --> 00:05:22.643 1-acemarke: that doesn't surprise me. 00:05:22.673 --> 00:05:28.223 I know that React app the last I saw it defaulted to about 1500 dependencies, 00:05:28.223 --> 00:05:30.353 largely because of Webpac, et cetera. 00:05:30.733

--> 00:05:35.053 Vet's obviously a lot fewer, but still that's, that's not great, 00:05:35.053 --> 00:05:36.943 just for a whole variety of reasons. 00:05:37.491 --> 00:05:39.141 2-vcarl: I guess teaser for y'all. 00:05:39.191 --> 00:05:43.091 Later towards the end of the episode we're gonna talk about web ecosystem 00:05:43.091 --> 00:05:47.231 and there's been some blog posts talking about like AI and how it defaults to 00:05:47.231 --> 00:05:51.521 building react instead of using the web platform and, how do we promote 00:05:51.521 --> 00:05:53.231 the web overreact, things like that. 00:05:53.271 --> 00:05:55.941 Which I don't know, it just feels that we just started bleeding into that 00:05:55.941 --> 00:06:00.171 topic and so we will get much more deeply into that later in the episode. 00:06:00.728 --> 00:06:01.538 1-acemarke: Next release. 00:06:01.598 --> 00:06:03.458 So, storybook 10 came out. 00:06:03.808 --> 00:06:05.608 Few different big points there. 00:06:05.698 --> 00:06:08.848 One of the biggest is that it's now ESM only, they've removed 00:06:08.848 --> 00:06:10.768 all the, the common JS artifacts. 00:06:11.078 --> 00:06:14.918 So they've shrunk the install size and they've also drastically shrunk 00:06:14.978 --> 00:06:17.138 the dependency chains as well. 00:06:17.508 --> 00:06:20.968 As well as some improvements to things like the mocking and 00:06:20.968 --> 00:06:22.408 some of the storybook formats. 00:06:22.847 --> 00:06:26.387 And then one other release of note, the remix folks are continuing to 00:06:26.387 --> 00:06:30.377 work on , remix V three after the initial demo and announcements. 00:06:30.377 --> 00:06:32.927 And they're trying to build it, as I understand it, with a lot 00:06:32.927 --> 00:06:34.427 of smaller reusable packages. 00:06:34.794 --> 00:06:41.004 And they just put out the first 0.1 alpha of their event interaction package, 00:06:41.034 --> 00:06:46.374 which is providing an abstraction for being able to compose events and 00:06:46.374 --> 00:06:48.294 listeners and behaviors together. 00:06:48.784 --> 00:06:49.594 Looks interesting. 00:06:49.654 --> 00:06:54.244 I'm not gonna claim I understand like exactly the benefits this comes, 00:06:54.694 --> 00:06:59.524 but nice to see the progress there and I suspect that a lot of folks 00:06:59.524 --> 00:07:00.664 are gonna get some use outta this. 00:07:01.004 --> 00:07:04.244 2-vcarl: Yeah, I'm reading the syntax here and it looks like 00:07:04.304 --> 00:07:09.284 a different way of adding event listeners to dom nodes, which, sure. 00:07:09.314 --> 00:07:09.644 I don't know. 00:07:10.064 --> 00:07:11.174 There's weird quirks. 00:07:11.604 --> 00:07:15.444 And certainly the remix team and, you know, Michael Jackson and Ryan Florence 00:07:15.444 --> 00:07:20.029 specifically have shown, have demonstrated a great willingness to jump into. 00:07:20.779 --> 00:07:26.269 What many people consider solved problems to reexamine from first principles. 00:07:26.329 --> 00:07:29.929 So I dunno, they, they've done that, they've done that repeatedly with their 00:07:29.929 --> 00:07:34.889 own solved problems of data loading in routing the interaction between 00:07:34.889 --> 00:07:37.899 data loading and routing I feel like that's one of the bigger complaints 00:07:37.899 --> 00:07:40.959 about React router is that they keep reexamining that same question. 00:07:40.959 --> 00:07:42.039 So, I don't know. 00:07:42.069 --> 00:07:45.436 It's interesting that they do have interesting things to say 00:07:45.436 --> 00:07:48.622 when I've seen them reexamine, quote unquote solved problems. 00:07:48.622 --> 00:07:50.123 So, I'll, I'm curious. 00:07:50.123 --> 00:07:51.503 This looks a little strange. 00:07:51.503 --> 00:07:55.283 It looks pretty divorced from the DOM structure, so if you just like on 00:07:55.283 --> 00:08:01.935 input element and then add a bunch of listeners for it, that's curious anyway. 00:08:02.115 --> 00:08:02.475 Yeah. 00:08:02.625 --> 00:08:04.605 O one release, we'll see what they come out with. 00:08:05.440 --> 00:08:05.740 Okay. 00:08:05.740 --> 00:08:07.750 That's all the new releases we got so far. 00:08:07.800 --> 00:08:09.030 Let's go into some main content. 00:08:09.430 --> 00:08:13.893 First off we want to talk a bit about, yeah, mark, you are on a, an ecosystem 00:08:13.893 --> 00:08:15.633 panel at React Summit this month. 00:08:16.120 --> 00:08:16.786 1-acemarke: So, Let's see. 00:08:16.786 --> 00:08:18.526 Addie has money, was moderating. 00:08:18.656 --> 00:08:23.076 We had Seth, Webster from the React team, react Foundation shun person, 00:08:23.076 --> 00:08:26.306 Amy Ner Nicholas Gallagher, and myself. 00:08:26.816 --> 00:08:28.496 A lot of the discussion was. 00:08:29.015 --> 00:08:34.775 Us asking Seth Webster questions about, so this React Foundation thing, 00:08:34.835 --> 00:08:37.014 like how is this actually gonna work? 00:08:37.044 --> 00:08:41.634 What does this actually mean for the React team and the project and the ecosystem? 00:08:41.988 --> 00:08:44.028 there is a video for the panel. 00:08:44.028 --> 00:08:45.528 It's not unlocked yet. 00:08:45.578 --> 00:08:49.028 I have access to it because I speak at GI Nation events. 00:08:49.348 --> 00:08:52.498 But the video should unlock within a couple weeks, I think. 00:08:52.798 --> 00:08:57.118 a few different points that Seth made during the panel itself, and then 00:08:57.118 --> 00:09:01.168 I had a chance to have a follow-up discussion with him afterwards. 00:09:01.708 --> 00:09:06.135 And there's a few different very interesting pieces of news about 00:09:06.165 --> 00:09:09.405 how the foundation's going to work and what their plans are. 00:09:10.277 --> 00:09:13.412 he talked a bit about some of the, the funding process. 00:09:13.502 --> 00:09:17.432 So, you know, large companies pay, you know, a good chunk of money to be

members. 00:09:17.432 --> 00:09:21.165 The, the existing board member companies have already paid up 00:09:21.165 --> 00:09:22.455 for like three to five years. 00:09:22.755 --> 00:09:26.337 And it sounds like he'd actually been working on putting the foundation 00:09:26.337 --> 00:09:28.347 together for over four years. 00:09:28.347 --> 00:09:31.536 Like this was a very long term kind of a process. 00:09:32.016 --> 00:09:35.286 And they're hoping to do a number of different things with the money. 00:09:35.686 --> 00:09:38.741 They want to, hire some additional engineers. 00:09:38.771 --> 00:09:41.981 And he even said some of that would be spent on having 00:09:41.981 --> 00:09:45.011 engineers triage the issues. 00:09:45.131 --> 00:09:48.521 And if you've ever looked at the React repo, you know, that's a thing 00:09:48.521 --> 00:09:51.071 that they do not do very often. 00:09:51.371 --> 00:09:53.441 So even something like that sounds good. 00:09:53.881 --> 00:09:57.991 They want to hire some more people to work on the actual docs and they're hoping to 00:09:57.991 --> 00:10:01.141 give money out to communities in some way. 00:10:01.351 --> 00:10:02.641 Don't know details on that, 00:10:03.201 --> 00:10:03.931 2-vcarl: Tell me more. 00:10:04.155 --> 00:10:04.725 1-acemarke: exactly. 00:10:05.225 --> 00:10:06.125 Another data point. 00:10:06.215 --> 00:10:11.605 So a lot of the plans have to do with the development process 00:10:11.665 --> 00:10:14.095 of re the React project itself. 00:10:14.770 --> 00:10:17.080 And transparency around that. 00:10:17.450 --> 00:10:20.430 Apparently they were even hoping to launch the foundation like a year and 00:10:20.430 --> 00:10:26.078 a half ago, and then they realized that like our own in day-to-day development 00:10:26.078 --> 00:10:28.538 processes aren't ready for that. 00:10:28.838 --> 00:10:35.078 We need to dog food the processes ourselves first so that we're more 00:10:35.078 --> 00:10:37.118 ready when the foundation goes live. 00:10:37.505 --> 00:10:41.655 And so some of that was you know, actually having docs and plans as 00:10:41.655 --> 00:10:46.545 new features are being developed or treating things and kind of like a, you 00:10:46.545 --> 00:10:50.625 know, TC 39 staged process approach. 00:10:51.105 --> 00:10:53.950 But one of the biggest things that Seth told me directly was that. 00:10:54.515 --> 00:10:58.055 Up until now, the React team has had, you know, their, like their weekly status 00:10:58.055 --> 00:11:00.815 meetings, and those are purely internal. 00:11:01.165 --> 00:11:05.575 First off it was just all the React team working at Meta and then it 00:11:05.575 --> 00:11:07.945 was, you know, meta plus Versal. 00:11:08.245 --> 00:11:12.835 But those meetings have always been internal, private, and non-visible. 00:11:13.072 --> 00:11:19.282 And Seth said that they eventually want to literally open up the weekly React team 00:11:19.282 --> 00:11:24.532 planning meetings to the public even as like a, like a Zoom call with no password. 00:11:25.072 --> 00:11:29.032 And that eventually, like designated community reps would even be able to 00:11:29.032 --> 00:11:31.612 actively participate in those discussions. 00:11:32.142 --> 00:11:37.932 So that's a radical change in the development of React itself. 00:11:38.392 --> 00:11:42.747 He also said that they're planning to, resurrect or like completely 00:11:42.747 --> 00:11:47.307 redo the RFC process, which frankly was kind of dead from the beginning. 00:11:47.307 --> 00:11:51.297 It was where random people from the community posted ideas that got ignored 00:11:51.627 --> 00:11:53.637 and the React team posted, finished. 00:11:53.907 --> 00:11:57.237 things that they planned to ship and then people argued over the naming 00:11:57.237 --> 00:11:58.467 and the React team ignored that. 00:11:58.644 --> 00:12:02.724 So like historically, the RFC process was pretty much irrelevant 00:12:02.814 --> 00:12:04.704 to actual React development. 00:12:05.197 --> 00:12:10.687 Now they're talking about making it like a real meaningful discussion 00:12:10.687 --> 00:12:12.787 and part of the planning process. 00:12:13.357 --> 00:12:16.357 Now, again, like do I know how this is gonna work out in practice? 00:12:16.417 --> 00:12:16.957 No. 00:12:17.226 --> 00:12:22.926 But I can see the amount of effort and the intent behind the foundation and 00:12:22.926 --> 00:12:24.966 the plans that they're talking about. 00:12:25.446 --> 00:12:30.911 And even if it doesn't end up going exactly as they or we would hope, I 00:12:30.911 --> 00:12:35.591 see good faith and good intentions behind all the work that they're 00:12:35.591 --> 00:12:37.391 trying to do around the process here. 00:12:37.841 --> 00:12:40.361 So I'm actually very, very excited about this. 00:12:40.808 --> 00:12:41.828 2-vcarl: Yeah, that's interesting. 00:12:41.888 --> 00:12:44.127 Oh got a lot of thoughts rattling around my head. 00:12:44.157 --> 00:12:47.937 'cause that's a lot of stuff that I've looked at trying to contribute towards 00:12:47.937 --> 00:12:50.157 the solution of but yeah, I don't know. 00:12:50.157 --> 00:12:53.089 It's, I fundamentally, I feel like the problem they're dealing with 00:12:53.089 --> 00:12:55.879 is that a lot of people care about. 00:12:56.197 --> 00:12:59.977 The outcome, the output, what they make and what React is. 00:13:00.157 --> 00:13:04.117 And not a lot of people actually have like the context and the background 00:13:04.147 --> 00:13:09.647 to like meaningfully contribute to the advancement of that goal., you know, 00:13:09.647 --> 00:13:13.350 to respond to what you just said about like their motivations and their, 00:13:13.681 --> 00:13:17.998 what they want to do, I don't think that's ever really been the problem. 00:13:17.998 --> 00:13:23.937 They've always had good intent and I even think a pretty clear eye

picture of 00:13:23.937 --> 00:13:27.837 what the problems are, but I think the problems are really thorny and complex 00:13:27.897 --> 00:13:30.447 and the solutions to them are non-obvious. 00:13:30.447 --> 00:13:36.135 And even when you have a good solution idea, executing it in 00:13:36.135 --> 00:13:38.475 a way that is effective and. 00:13:38.888 --> 00:13:42.698 That operates on the, kind of like timescales that React is around four. 00:13:42.788 --> 00:13:43.688 It's just really challenging. 00:13:43.688 --> 00:13:48.278 Like it's, you know, anyone can spin up a process and run it twice and you 00:13:48.278 --> 00:13:52.567 know, maybe that, like you do, you call for comments and you run a community 00:13:52.567 --> 00:13:54.487 discussion and round tables and whatever. 00:13:54.877 --> 00:14:01.067 But like that plays out over the timescale of like weeks and months and React 00:14:01.637 --> 00:14:06.542 seems to plan on the order of years and is now, has existed for a decade. 00:14:06.542 --> 00:14:10.922 So, you know, with like there's possibility for its scope to expand from 00:14:10.922 --> 00:14:14.882 thinking in terms of years to thinking in terms of, I don't wanna say decades, 00:14:14.882 --> 00:14:16.532 but that's the next unit of time. 00:14:16.532 --> 00:14:17.672 That's the next order of magnitude. 00:14:18.104 --> 00:14:19.514 I hope they can achieve their goals. 00:14:19.514 --> 00:14:23.294 I am looking forward to seeing some more roadmap, transparency. 00:14:23.758 --> 00:14:24.988 Of some sort actually. 00:14:24.988 --> 00:14:28.198 That'll be really great for us, for you and I. That'll would be great. 00:14:28.618 --> 00:14:31.918 That'll be a great source of signal that we can then pick apart and 00:14:31.918 --> 00:14:38.608 analyze to surely, to much to their chagrin by way of drawing attention 00:14:38.788 --> 00:14:41.395 to things that are high context text. 00:14:41.789 --> 00:14:44.234 1-acemarke: As opposed to like stalking the open PRS list. 00:14:44.589 --> 00:14:44.799 2-vcarl: Yeah. 00:14:44.799 --> 00:14:45.249 Right. 00:14:45.453 --> 00:14:45.873 Cool. 00:14:45.903 --> 00:14:46.383 Nice. 00:14:46.563 --> 00:14:46.983 Love that. 00:14:47.245 --> 00:14:49.893 1-acemarke: So on that note Another item that I've personally been 00:14:49.893 --> 00:14:55.593 involved in is the work on the upcoming React Concurrent Stores, API. 00:14:56.143 --> 00:15:02.073 So when React 18 came out the React team included the Uyc external store hook 00:15:02.283 --> 00:15:07.173 as the official built-in way for third party libraries like Redux, Zein, Jo Tai, 00:15:07.173 --> 00:15:09.933 whatever else to integrate into React. 00:15:10.145 --> 00:15:14.138 take an external state update, turn it into a React re-render, and it worked. 00:15:14.198 --> 00:15:18.818 But it also had a very intentional design limitation that if you do an 00:15:18.818 --> 00:15:23.078 update in the middle of a paused React render, you know, something transition 00:15:23.078 --> 00:15:27.908 like, then it bails out of that and just does a complete top to bottom render. 00:15:27.968 --> 00:15:31.328 So you're kind of throwing away some of the benefits of React, 00:15:31.328 --> 00:15:32.738 being able to pause itself. 00:15:33.248 --> 00:15:36.158 And that was because they, they really didn't have a better idea 00:15:36.308 --> 00:15:38.258 for how to do that kind of interop. 00:15:39.023 --> 00:15:44.453 So this concurrent store's API is supposed to be essentially a new 00:15:44.663 --> 00:15:49.913 equivalent to use sync external store, but concurrent transition compatible. 00:15:50.272 --> 00:15:55.162 And Jordan Eldridge, who's on the relay team, has been working on 00:15:55.192 --> 00:15:58.312 the design and initial prototype IMP implementation for this. 00:15:58.744 --> 00:16:02.344 And so he put out a polyfil package that. 00:16:03.154 --> 00:16:06.454 Roughly implements the prototype as a standalone library. 00:16:07.084 --> 00:16:11.404 And then I've been talking with him about, you know, here's what Redux 00:16:11.404 --> 00:16:12.904 would probably need to make this work. 00:16:12.904 --> 00:16:14.824 Here's our, our requirements for this. 00:16:15.304 --> 00:16:19.114 And so I was able to then take his proof of concept package and put 00:16:19.114 --> 00:16:23.644 up a draft PR for React Redux that just swapped out our use selector 00:16:23.644 --> 00:16:26.254 implementation to use that instead. 00:16:26.584 --> 00:16:30.604 And it basically worked, I mean, just the one test file found a couple edge 00:16:30.604 --> 00:16:34.924 cases, but like most of our tests passed, we're not doing anything 00:16:34.924 --> 00:16:36.994 transition related in our tests. 00:16:37.504 --> 00:16:40.984 But it's good to know that like your baseline standard behavior works. 00:16:41.374 --> 00:16:44.734 And this also turned up several more things we'll have to figure out, like. 00:16:45.649 --> 00:16:50.319 Quality checks and how do you handle selectors throwing errors? 00:16:50.319 --> 00:16:55.389 And we're talking about when can react safely, call selectors, can it delay it 00:16:55.389 --> 00:16:57.189 due to batching and things like that. 00:16:57.491 --> 00:17:01.241 So also very excited to see the progress being made on this. 00:17:01.241 --> 00:17:05.321 A PIA lot of people have asked for external state transition 00:17:05.321 --> 00:17:09.191 compatibility, and so this is going to be a big deal for the ecosystem. 00:17:09.808 --> 00:17:10.348 2-vcarl: Interesting. 00:17:10.408 --> 00:17:10.918 Cool. 00:17:11.227 --> 00:17:13.477 I'm just trying to put this in a little bit of context for myself. 00:17:13.477 --> 00:17:19.747 So this is a, an exploration of an API for a concurrent store that they've 00:17:19.747 --> 00:17:22.417 indicated intent to ship eventually. 00:17:23.227 --> 00:17:23.407 1-acemarke: Mm-hmm.

00:17:23.492 --> 00:17:23.782 Yeah. 00:17:24.011 --> 00:17:29.399 so basically a, a new equivalent of use syn external store, but 00:17:29.549 --> 00:17:31.889 minus the sync restriction. 00:17:32.465 --> 00:17:32.945 2-vcarl: Got it. 00:17:33.010 --> 00:17:33.725 Okay, cool. 00:17:33.725 --> 00:17:34.265 Interesting. 00:17:34.602 --> 00:17:35.802 you called it a polyfill. 00:17:35.802 --> 00:17:40.542 And I see in the blog post it's labeled a pony fill, which I had to 00:17:40.542 --> 00:17:42.132 refresh my memory on the distinction. 00:17:42.132 --> 00:17:47.622 A polyfill fills in a gap in the platform transparently, and a pony fill 00:17:48.192 --> 00:17:51.942 offers the same functionality, but as a standalone module that you have to import. 00:17:52.364 --> 00:17:52.964 There you go. 00:17:53.024 --> 00:17:53.474 Now you know. 00:17:53.775 --> 00:17:55.700 1-acemarke: So one more item kind of related to that. 00:17:55.850 --> 00:18:00.170 At React Con Freaky Hanin announced a new Async React working group to 00:18:00.170 --> 00:18:04.190 provide support to the, you know, the community and the ecosystem on, you 00:18:04.190 --> 00:18:07.610 know, improving the React docs around async behavior, helping libraries 00:18:07.610 --> 00:18:09.170 get set up to use these methods. 00:18:09.620 --> 00:18:14.300 And so Matt Brophy from the React Router team had a good discussion with Ricky 00:18:14.660 --> 00:18:18.760 about how they're trying to handle, using transitions in React router. 00:18:19.180 --> 00:18:23.710 So there was a, a good technical discussion about the nuances of behavior. 00:18:24.160 --> 00:18:29.440 And then the React router team also put out a pre-release version or 00:18:29.440 --> 00:18:34.870 unstable option to enable some of the transition behavior as well. 00:18:35.158 --> 00:18:39.298 So starting to see some adoption and this ties into the React team 00:18:39.298 --> 00:18:44.428 wanting to see libraries pick up these behaviors like transitions and 00:18:44.428 --> 00:18:46.618 action props and, and build them in. 00:18:46.942 --> 00:18:51.742 I will say that I had a couple discussions at JS Nation React Summit were some 00:18:51.742 --> 00:18:57.502 people were questioning whether making the entire ecosystem change semantics 00:18:57.502 --> 00:19:02.692 and behaviors and add props to make the React team happy was gonna be a 00:19:02.692 --> 00:19:04.782 good idea or feasible, but we'll see. 00:19:05.083 --> 00:19:05.373 2-vcarl: Yeah. 00:19:05.528 --> 00:19:06.098 Interesting. 00:19:06.435 --> 00:19:10.545 I see a bit of why they, why the core team desires that. 00:19:10.935 --> 00:19:14.037 Transitions are definitely like, my understanding of the core 00:19:14.037 --> 00:19:17.907 team's incentives and what they care about is ability to use the 00:19:17.907 --> 00:19:20.067 platform and things like that. 00:19:20.494 --> 00:19:24.874 So given that like transitions are a part of the platform and they are 00:19:24.964 --> 00:19:29.465 interested in allowing people to take advantage of platform improvements, I 00:19:29.465 --> 00:19:35.642 understand why they would message, in order to use this, the entire ecosystem 00:19:35.642 --> 00:19:39.932 must make this change to the patterns they use for compatibility reasons. 00:19:40.369 --> 00:19:43.185 But that is very, I don't know, that's a big challenge. 00:19:43.625 --> 00:19:46.835 it's hurting a lot of cats who don't necessarily. 00:19:47.020 --> 00:19:50.446 Aren't even necessarily aware that there's an attempt to herd being made. 00:19:50.853 --> 00:19:53.433 1-acemarke: if you watch Ricky's talk from React con, you know, he, 00:19:53.433 --> 00:19:57.423 he demoed switching from like a bunch of manual use effects and loading 00:19:57.423 --> 00:19:59.613 states to suspense and transitions. 00:19:59.833 --> 00:20:02.893 And then he pointed out, and so now you have a much better experience, 00:20:02.923 --> 00:20:05.533 but you also had to do, you know, call start transition and all 00:20:05.533 --> 00:20:07.183 your click handlers and so on. 00:20:07.463 --> 00:20:09.443 , there's even some painful nuances there. 00:20:09.743 --> 00:20:13.943 Because start transition is based on setting a flag 00:20:13.943 --> 00:20:16.163 internally for the current event. 00:20:16.163 --> 00:20:17.003 Loop tick. 00:20:17.753 --> 00:20:20.483 If you have a callback that has an await. 00:20:20.885 --> 00:20:25.119 You leave the event loop tick, and if you then want to do another state 00:20:25.119 --> 00:20:30.369 update after the await, you have to call start transition a second nested time. 00:20:30.899 --> 00:20:35.369 And Ryan Cardo actually pointed out to me that like Ricky has repeatedly 00:20:35.369 --> 00:20:37.229 messed this up in his own demos. 00:20:37.499 --> 00:20:41.219 It's a very foot gun prone behavior pattern. 00:20:41.814 --> 00:20:47.369 And so the combination of like you currently have to call, start transition 00:20:47.369 --> 00:20:52.709 yourself, plus trying to get the patterns right is a good argument for let's 00:20:52.709 --> 00:20:54.599 just build this into all the libraries. 00:20:54.839 --> 00:20:57.479 But that takes time and effort and coordination. 00:20:58.000 --> 00:20:58.330 2-vcarl: Yeah. 00:20:58.360 --> 00:20:58.660 Right. 00:20:58.660 --> 00:21:04.290 Like that was one of my big takeaways from Ricky's talk was, oh, this looks hard. 00:21:04.640 --> 00:21:07.848 Which is not the best takeaway if you're trying to convince the whole 00:21:07.848 --> 00:21:09.828 ecosystem to move on to something. 00:21:10.361 --> 00:21:12.371 1-acemarke: there is a lot of complexity in. 00:21:12.677 --> 00:21:17.951 Mental model of React and the kinds of apps we're trying to build, and both 00:21:17.951 --> 00:21:21.731 the fact that you now need to keep track of additional API methods that 00:21:21.731 --> 00:21:25.781 you have to

call and the semantics and when you're supposed to use them, 00:21:25.781 --> 00:21:30.641 and the fact that your, your state is now kind of in multiple Schrodinger 00:21:30.821 --> 00:21:33.764 box, multiple versions at once. 00:21:34.094 --> 00:21:36.824 It's, it's definitely a lot of additional mental overhead. 00:21:37.529 --> 00:21:38.909 2-vcarl: Yeah, right. 00:21:38.909 --> 00:21:42.110 I'm just looking at in this GitHub thread that you linked, Ricky 00:21:42.110 --> 00:21:44.330 says, you know, oh, I think there may have been some confusion. 00:21:44.330 --> 00:21:47.990 Here's how you do it for navigations, and it's, you know, function, navigate 00:21:47.990 --> 00:21:52.880 to A URL that has start transition, which, you know, wraps set router 00:21:52.880 --> 00:21:57.530 state, which then wrap, you know, it's argument is a function that. 00:21:57.775 --> 00:22:00.015 I guess returns, A-A-U-R-L. 00:22:00.045 --> 00:22:03.945 It's a thunk for set router state in start transition. 00:22:03.945 --> 00:22:04.845 In navigate. 00:22:04.845 --> 00:22:06.135 I don't know, they're like, that's just complicated. 00:22:06.165 --> 00:22:08.415 Like what, why are we wrapping this with so many different 00:22:08.415 --> 00:22:09.645 functions just to navigate? 00:22:10.042 --> 00:22:11.441 That's uncomfortable. 00:22:11.531 --> 00:22:15.191 It feels like the wrong API in some way, you know, like it an API should 00:22:15.671 --> 00:22:20.721 obscure complexity and this does not obscure the complexity so Interesting. 00:22:20.781 --> 00:22:24.171 I, I want to be able to take advantage of transitions in apps. 00:22:24.201 --> 00:22:28.256 'cause that's super great if you can just like, easily convey that 00:22:28.406 --> 00:22:32.756 you are here and you're about to go here instead of just going there. 00:22:32.936 --> 00:22:36.266 Like, you get a lot of really powerful things that you can do for user 00:22:36.266 --> 00:22:40.614 experience and, you know, whatever, all sorts of stuff if you communicate 00:22:40.884 --> 00:22:44.873 the transition instead of just updating the current state of your app. 00:22:45.453 --> 00:22:47.553 But yeah, I don't, I, I'm not sold on this. 00:22:47.913 --> 00:22:49.833 I'm not sold on what I'm seeing so far. 00:22:50.495 --> 00:22:50.785 1-acemarke: Okay. 00:22:51.460 --> 00:22:52.600 Moving on to the. 00:22:53.275 --> 00:22:54.715 Next related chunk of topic. 00:22:55.085 --> 00:22:59.735 The Tans stack folks have a bunch of updates that they've put out 00:22:59.735 --> 00:23:02.875 both blog posts and versions. 00:23:03.505 --> 00:23:10.675 So Tanner put out an article on the State of Tans stack as an organization in 00:23:10.675 --> 00:23:14.815 his own work as a full-time open source maintainer for the last couple years. 00:23:15.295 --> 00:23:17.665 Also, again heard, you know, heard some of this from, from him 00:23:17.665 --> 00:23:20.718 directly at react Summit loosely.

00:23:20.718 --> 00:23:25.960 He himself has been full-time supported open source for the last couple years 00:23:25.990 --> 00:23:27.820 sponsorships to the TN stack org. 00:23:28.340 --> 00:23:32.240 Hand stack libraries are, you know, as we all know, very, very widely 00:23:32.240 --> 00:23:36.080 adopted and Certainly query is the standard fritz thing at this point. 00:23:36.430 --> 00:23:41.470 Router and is picking up a lot, lot of popularity start is almost 1.0. 00:23:42.070 --> 00:23:46.660 And he's hoping to be able to start to bring in some part-time or even 00:23:46.660 --> 00:23:52.030 full-time people to hand stack the company in the near future. 00:23:52.370 --> 00:23:58.820 Essentially other current maintainers for various hand stack libraries who 00:23:58.820 --> 00:24:02.060 would then be able to potentially switch to doing that as their actual job. 00:24:02.415 --> 00:24:06.886 So essentially he's been fortunate enough to have this workout for him and 00:24:06.886 --> 00:24:10.936 he's hoping to share that ability with other people working on the projects. 00:24:11.722 --> 00:24:16.806 2-vcarl: I see says monthly sponsorships for 12 core contributors 00:24:17.122 --> 00:24:19.822 and short term contracts for another three to five people. 00:24:19.822 --> 00:24:22.222 I guess it's not clear what monthly sponsorships means. 00:24:22.342 --> 00:24:26.148 I, my assu, my assumption there would be like their, maybe not full 00:24:26.148 --> 00:24:29.538 time, but getting like sign, you know, significant compensation, but 00:24:29.538 --> 00:24:31.518 that may not necessarily be true. 00:24:31.929 --> 00:24:34.483 1-acemarke: I don't know numbers, but I think at the moment the others are 00:24:34.483 --> 00:24:39.103 doing open source and they're full-time and getting some, some stipends, and 00:24:39.239 --> 00:24:43.255 in the next couple years he's looking at actually hiring some of them. 00:24:43.780 --> 00:24:44.140 2-vcarl: Cool. 00:24:44.286 --> 00:24:44.796 Love that. 00:24:44.856 --> 00:24:45.906 That's super hard. 00:24:46.076 --> 00:24:49.586 Building revenue so that you can actually support other humans 00:24:49.586 --> 00:24:50.786 is actually really challenging. 00:24:51.569 --> 00:24:54.599 I really like this blog post about, you know, tan Stack for two years. 00:24:54.844 --> 00:24:55.594 like this line. 00:24:55.594 --> 00:24:57.244 You know, building a full stack framework is hard. 00:24:57.244 --> 00:24:58.534 I knew that from watching other teams do it. 00:24:58.564 --> 00:24:59.824 Most of them had something I didn't. 00:25:00.094 --> 00:25:04.671 Capital Next Gatsby Redwood Remix all had funding companies or acquisition paths. 00:25:05.171 --> 00:25:08.471 And that, that definitely is something that feels very different about Tan Stack. 00:25:08.471 --> 00:25:12.725 It feels very much like he couldn't not do this. 00:25:12.755 --> 00:25:15.818 Like he just, has spent several years

working in this problem space and 00:25:15.818 --> 00:25:20.288 solving his own problems with the freedom to publish them so that other 00:25:20.288 --> 00:25:22.268 people can take advantage of his work. 00:25:22.418 --> 00:25:28.088 And then got to a point where it just made sense to invest more effort and take those 00:25:28.088 --> 00:25:29.708 a little bit further than they could go. 00:25:29.967 --> 00:25:33.567 Absent that level of dedication I, which I, I love. 00:25:33.567 --> 00:25:38.067 That's just like, that's, I don't know, finding that kind of work to do for 00:25:38.067 --> 00:25:40.857 yourself is what I've been looking for. 00:25:40.947 --> 00:25:43.516 And it's really hard to find, and I love that Tanner Wesley 00:25:43.516 --> 00:25:44.716 has apparently found it. 00:25:45.072 --> 00:25:48.732 And also having 16 partners doing sponsorships is also great. 00:25:48.732 --> 00:25:52.272 Like that, that, that may not be capital, but that's, that's great. 00:25:52.452 --> 00:25:52.962 That's nice. 00:25:53.346 --> 00:25:56.796 having tried to find sponsors myself, like getting 16, it's like, oh shit. 00:25:56.796 --> 00:25:57.276 Good job. 00:25:57.546 --> 00:25:58.026 That's awesome. 00:25:58.814 --> 00:26:03.114 1-acemarke: So it tens Stack DB is a new work in progress package largely being 00:26:03.114 --> 00:26:06.834 worked on by Kyle Matthews, formerly of Gatsby, currently of electric SQL. 00:26:07.174 --> 00:26:11.584 That is meant to be kind of like a half a sync engine layer 00:26:11.914 --> 00:26:13.954 on top of another data source. 00:26:14.284 --> 00:26:18.726 And so like one of the common usage patterns for that would be you 00:26:18.726 --> 00:26:24.005 wrap it on top of your 10 stack query, you know, API definitions. 00:26:24.425 --> 00:26:26.825 And so maybe you've got some endpoints for fetching, 00:26:27.235 --> 00:26:28.915 post comments users, et cetera. 00:26:29.215 --> 00:26:34.285 But then it presents like a sync engine like interface on top of that, and 00:26:34.285 --> 00:26:38.035 it's got a bunch of data flow analysis baked in so that it can do things like 00:26:38.095 --> 00:26:42.655 partial fetches or only re fetching, you know, certain pieces of data. 00:26:42.935 --> 00:26:43.832 joining. 00:26:44.308 --> 00:26:48.598 Essentially presenting like a normalized facade layer over the individual 00:26:48.598 --> 00:26:53.008 endpoints so that they act like they're unified almost as a, like a 00:26:53.008 --> 00:26:56.278 replacement for a GraphQL style API. 00:26:56.696 --> 00:26:59.426 And so they've been working on this and it has integrations with 00:26:59.426 --> 00:27:02.696 Tan Stack Query and then Electric SQL and a few other data packages. 00:27:03.086 --> 00:27:06.566 And Tanner pointed out to me at React Limit that. 00:27:06.869 --> 00:27:11.999 It really just needs an adapter layer to be able to connect to another library and 00:27:11.999 --> 00:27:14.339 I could totally build one for RTK query. 00:27:14.549 --> 00:27:17.219 And so I was actually briefly playing with that on the flight home. 00:27:17.269 --> 00:27:18.619 Haven't had a chance to push it further, 00:27:18.979 --> 00:27:22.639 but I'm actually, I'm actually very excited to play with that more myself 00:27:22.879 --> 00:27:29.329 because like 10 stack query, we made the decision to make RTK query non normalized. 00:27:29.329 --> 00:27:31.339 It just caches the response from the server. 00:27:31.969 --> 00:27:37.759 And so if this essentially let us make a normalized layer on top of 00:27:37.759 --> 00:27:41.719 our own library without us having to make any further API changes. 00:27:42.199 --> 00:27:43.274 That's very interesting to me. 00:27:43.984 --> 00:27:46.744 2-vcarl: I'm also thinking about, you know, this is from Kyle Matthews 00:27:46.744 --> 00:27:52.984 who did Gatsby, which, you know, notoriously exposed a GraphQL, 00:27:53.074 --> 00:27:56.374 you know, layer on top of whatever data sources you wanted to give it. 00:27:56.704 --> 00:28:00.304 So just, I'm thinking about how how much of Kyle Matthew's work 00:28:00.304 --> 00:28:02.785 could be described as providing 00:28:03.295 --> 00:28:07.174 query abstractions over arbitrary data layers with data 00:28:07.489 --> 00:28:07.789 1-acemarke: Yeah. 00:28:08.309 --> 00:28:08.849 that's a good point. 00:28:08.849 --> 00:28:11.729 I, I had, I hadn't even drawn that connection that this is actually 00:28:11.999 --> 00:28:13.319 very related to his prior work. 00:28:14.028 --> 00:28:16.658 2-vcarl: if you describe this at a certain level of abstraction, this 00:28:16.658 --> 00:28:22.058 sounds just like GraphQL, relay Gatsby, a bunch of other stuff, but I do see 00:28:22.058 --> 00:28:25.208 the niche that it occupies and how it's distinct from those projects. 00:28:25.508 --> 00:28:29.438 it's kind of like GraphQL except instead of the network boundary, 00:28:29.438 --> 00:28:34.598 it's like the component to data store boundary on your client app. 00:28:35.078 --> 00:28:38.791 Which that's useful if you can, like what you just said about how you 00:28:38.791 --> 00:28:43.752 don't normalize in RTK query I've landed on that as being generally a 00:28:43.752 --> 00:28:49.402 good architectural decision to make in applications because like if you 00:28:49.402 --> 00:28:51.202 start doing data transformations. 00:28:51.369 --> 00:28:55.442 You know, bespoke data transformations per endpoint, then it becomes really 00:28:55.442 --> 00:28:56.762 hard to change those endpoints. 00:28:56.827 --> 00:29:00.782 If, if you're changing the underlying data, then you've gotta change 00:29:00.782 --> 00:29:02.882 all the code at the same time. 00:29:03.305 --> 00:29:08.225 Whereas if you just take whatever the API gives you and then write like 00:29:08.225 --> 00:29:13.985 a transformer on top of that, then it's just you, you add a layer of 00:29:13.985 -->

00:29:17.915 abstraction for yourself that gives you a little bit more flexibility 00:29:17.915 --> 00:29:20.405 around timing of some types of changes. 00:29:20.745 --> 00:29:22.185 So yeah, I dunno that's, it's interesting. 00:29:22.185 --> 00:29:23.415 This is an interesting project. 00:29:23.737 --> 00:29:27.158 I dunno, something about it feels like I don't quite know where I would take 00:29:27.158 --> 00:29:31.898 advantage of it for myself, but I could see the right app with the right kind 00:29:31.898 --> 00:29:36.188 of team operating it, finding this and being like, this is exactly what we need. 00:29:36.188 --> 00:29:41.421 This solves multiple challenging, thorny problems that we have 00:29:41.421 --> 00:29:42.651 or, so, yeah, I don't know. 00:29:42.651 --> 00:29:43.131 It's interesting. 00:29:43.641 --> 00:29:47.061 1-acemarke: my own personal to-do list is massive at this point. 00:29:47.061 --> 00:29:49.821 Both like redux and even just like personal life stuff. 00:29:49.881 --> 00:29:53.871 But I, I absolutely want to go back and, and play with my own RTK query integration 00:29:53.871 --> 00:29:55.761 prototype for that in the near future. 00:29:56.095 --> 00:29:56.335 2-vcarl: Yeah. 00:29:56.335 --> 00:29:59.005 I generally love the idea of sync engines. 00:29:59.303 --> 00:30:04.444 one of my core beliefs about writing code is that many more applications 00:30:04.444 --> 00:30:09.591 that we use have no technical necessity to be online only. 00:30:09.651 --> 00:30:12.801 Like there are a lot of things like, for example, my favorite canonical 00:30:12.801 --> 00:30:14.451 example of this is like Gmail. 00:30:14.451 --> 00:30:18.441 If you're using the Gmail app, it doesn't actually care if you're online 00:30:18.441 --> 00:30:21.831 because it is doing synchronization of your emails in the background. 00:30:22.113 --> 00:30:25.203 And so you can just interact with it normally while you're fully offline 00:30:25.203 --> 00:30:28.893 and later it will do all the things like synchronize the read states or 00:30:28.893 --> 00:30:31.863 like, you know, which emails you've archived and deleted and yada, yada 00:30:31.863 --> 00:30:33.693 yada, completely transparently. 00:30:33.693 --> 00:30:37.501 Like, I have never thought about, oh, I should wait to do this 00:30:37.501 --> 00:30:40.381 later when I'm online, so I don't cause weird problems for myself. 00:30:40.651 --> 00:30:44.761 And I do think about that all the fucking time in various other apps that I use. 00:30:44.761 --> 00:30:47.797 So like there, that's something that we used to do that we no 00:30:47.797 --> 00:30:49.237 longer do because it's hard. 00:30:49.539 --> 00:30:54.009 I love seeing more effort and energy being expended into making 00:30:54.009 --> 00:30:57.166 it easier to do offline things. 00:30:57.256 --> 00:30:58.696 And sync engines are a big part of that. 00:30:58.964 --> 00:31:03.764 And so abstractions over sync engines are also a part of that. 00:31:03.794 --> 00:31:04.124 Great. 00:31:04.592 --> 00:31:06.997 1-acemarke: the Tans stack router folks also put out a really good 00:31:06.997 --> 00:31:10.477 blog post talking about how they did some massive optimization on route 00:31:10.477 --> 00:31:13.987 matching, including using a somewhat lesser known data structure called 00:31:13.987 --> 00:31:16.477 I think it's pronounced a try, TRIE. 00:31:16.791 --> 00:31:19.616 I love, I, I love seeing optimization breakdowns. 00:31:19.766 --> 00:31:21.056 So good article there. 00:31:21.298 --> 00:31:25.405 And then I. Since we've done plenty of, you know, stuff being teased on 00:31:25.405 --> 00:31:29.275 Twitter discussion in the past with Remix and everything else Tanner has 00:31:29.275 --> 00:31:35.035 been teasing that they are working on RSC support for Tant Stack Start. 00:31:35.755 --> 00:31:41.434 And I can say that he showed me a demo and it's unconventional, it is not 00:31:41.434 --> 00:31:46.924 what you would expect after having used next, and yet it totally makes 00:31:46.924 --> 00:31:49.804 sense in its own way once you see it. 00:31:50.204 --> 00:31:53.534 So as I understand it, , they've made real progress on implementing it. 00:31:53.584 --> 00:31:57.304 , they figured out the design and now it's push it forward 00:31:57.394 --> 00:31:59.464 and make it actually a thing. 00:31:59.654 --> 00:32:04.701 I have no idea how soon this will be revealed, but I think once this comes out, 00:32:04.791 --> 00:32:11.661 it will be a very interesting, challenging take on our understanding of server 00:32:11.661 --> 00:32:13.161 components and how they, how to use them. 00:32:13.601 --> 00:32:16.481 2-vcarl: I would certainly say that the way next has done it does 00:32:16.481 --> 00:32:19.931 not seem to be like the end all, be all of great implementation. 00:32:19.931 --> 00:32:23.291 So if they have found a new way to approach it, hell yeah. 00:32:23.547 --> 00:32:24.027 Here for it. 00:32:24.504 --> 00:32:24.984 That's awesome. 00:32:25.267 --> 00:32:27.367 1-acemarke: So there, there's your, vague tweet for the month. 00:32:27.551 --> 00:32:27.971 2-vcarl: Yeah. 00:32:28.061 --> 00:32:28.601 Right. 00:32:28.867 --> 00:32:30.397 let's talk a little more broadly. 00:32:30.457 --> 00:32:33.097 Let's talk about the state of the web ecosystem. 00:32:33.607 --> 00:32:38.307 I guess like to intro our subtopics under state of the web ecosystem. 00:32:38.624 --> 00:32:41.174 I wanna talk a bit about like the CloudFlare outage, 00:32:41.400 --> 00:32:43.636 the NPM ongoing attacks. 00:32:43.636 --> 00:32:47.176 Like this is not a new attack, this is the, a continuation of the previous 00:32:47.322 --> 00:32:49.212 attacks plural, that we've discussed. 00:32:49.722 --> 00:32:54.882 As well as, Alex Russell had a decent blog post, I say decent because I 00:32:54.882 --> 00:32:56.862 don't know, it's, it qualified, decent.

00:32:56.922 --> 00:32:59.712 It, it's well written, it's well argued, but I don't necessarily 00:32:59.712 --> 00:33:00.852 agree with his conclusions. 00:33:01.122 --> 00:33:06.240 And also there, there's some conversation about reacts place in the 00:33:06.270 --> 00:33:09.600 ecosystem and AI and things like that. 00:33:09.600 --> 00:33:14.014 So that's the shape of this state of the web ecosystem chat that I want to 00:33:14.044 --> 00:33:16.504 start or that I want to have, I guess. 00:33:16.814 --> 00:33:18.164 Yeah, let's start at CloudFlare. 00:33:18.254 --> 00:33:24.144 They had an outage on the, earlier this month, about a week ago 18th of November. 00:33:24.194 --> 00:33:27.284 They say in their outage postmortem that it was triggered by a change to 00:33:27.284 --> 00:33:31.364 the permission system of one of the databases, which caused it to output 00:33:31.394 --> 00:33:35.954 entries ba Basically, it produced two large of a file that then got distributed 00:33:35.954 --> 00:33:41.494 out and the size of that larger file getting distributed, caused performance 00:33:41.494 --> 00:33:43.804 load that took down the network, 00:33:44.227 --> 00:33:46.477 1-acemarke: check me on this, another factor on it was they, 00:33:46.657 --> 00:33:51.697 one of the services was written in Rust and it had an expectation 00:33:51.697 --> 00:33:53.947 that the file would only be so big. 00:33:54.337 --> 00:33:57.460 And there was, I think there was a rust line of code that basically 00:33:57.460 --> 00:33:59.590 said if it doesn't fit, panic. 00:34:00.149 --> 00:34:06.839 Yeah, that when a bad file with like , they had 60 features, a limit of 200 00:34:06.839 --> 00:34:10.299 and a line of code that basically said if it gets over 200, throw 00:34:10.349 --> 00:34:11.759 throw an error, kill the service. 00:34:12.399 --> 00:34:13.149 2-vcarl: That'll do it. 00:34:13.359 --> 00:34:13.989 Yeah. 00:34:14.259 --> 00:34:17.319 Each module ran on the proxy service has a number of limits to 00:34:17.499 --> 00:34:19.419 avoid unbounded memory consumption. 00:34:19.855 --> 00:34:23.875 somewhat questioned the wisdom of that because, the reason you would have 00:34:23.995 --> 00:34:28.555 limits on memory consumption in place would be to avoid outage and downtime. 00:34:28.555 --> 00:34:34.245 And so if your resolution to, the outcome of your limit, if memory usage gets too 00:34:34.245 --> 00:34:37.365 high, is to crash, that is functionally. 00:34:37.755 --> 00:34:39.975 Very similar to an out of memory error. 00:34:40.472 --> 00:34:43.528 Maybe it's, I guess it's easier to debug, I guess you can, out 00:34:43.528 --> 00:34:46.558 of memory errors cause Crazy, unpredictable failures, I guess. 00:34:46.558 --> 00:34:50.588 So at least this is a predictable failure that strongly indicates where the , the 00:34:50.588 --> 00:34:52.148 excessive memory use is coming from. 00:34:52.608 --> 00:34:54.618 But yeah, it it was pretty serious. 00:34:55.038 --> 00:34:57.408 I actually didn't notice, which I'm glad about. 00:34:57.438 --> 00:35:01.158 It's nice to not be so online that you don't immediately notice 00:35:01.158 --> 00:35:03.255 every, I don't know every outage. 00:35:03.638 --> 00:35:08.165 But yeah, you know, as people are discussing how React takes over, 00:35:08.315 --> 00:35:12.245 you know, is now the default for web development, this kind of, I think this 00:35:12.245 --> 00:35:15.665 was a, a a good demonstration of the place in the ecosystem that CloudFlare 00:35:15.665 --> 00:35:20.195 occupies and how many different services truly rely on its stable operations. 00:35:20.555 --> 00:35:20.645 Yeah. 00:35:21.307 --> 00:35:25.283 1-acemarke: Trying to make the trade-offs between, the web is decentralized. 00:35:25.283 --> 00:35:28.193 Anyone can run their own thing versus, Hey, look, , there's 00:35:28.193 --> 00:35:30.383 benefits to having centralization. 00:35:30.713 --> 00:35:34.973 Let's all put our GitHub, or let's, let's all put our git repositories on the same 00:35:34.973 --> 00:35:39.053 service so we can, you know, log in and share them and comment back and forth. 00:35:39.343 --> 00:35:40.363 Let's put all our 00:35:40.537 --> 00:35:42.907 social media comments on the same service. 00:35:42.907 --> 00:35:44.887 We can have commentary back and forth. 00:35:45.117 --> 00:35:50.637 Let's all use this widely spread CDN and DDoS protection service that also has a 00:35:50.637 --> 00:35:52.527 bunch of really cool backend services. 00:35:53.097 --> 00:35:57.788 And then when one of those big services goes down, it ends up having, or everybody 00:35:57.788 --> 00:36:00.308 put everything in A-W-A-W-S East one. 00:36:00.657 --> 00:36:00.957 2-vcarl: Yeah. 00:36:01.017 --> 00:36:01.407 right. 00:36:01.752 --> 00:36:05.637 1-acemarke: Or, or, or even like we, we put it in AWS East two, 00:36:05.667 --> 00:36:10.437 but AWS themselves relied on AWS East one and then it goes down. 00:36:10.824 --> 00:36:14.694 we keep running into these big dependencies that end up having massive 00:36:14.694 --> 00:36:15.844 effects on the internet when they go down. 00:36:16.445 --> 00:36:18.785 2-vcarl: This feels like a good transition point to a link that 00:36:18.785 --> 00:36:20.165 we had slightly later down. 00:36:20.485 --> 00:36:24.115 Tom McRight, who is not a name I recognize, put on a blog 00:36:24.115 --> 00:36:27.845 post called What If People Don't Want To Create Things and. 00:36:28.191 --> 00:36:29.391 that feels like what you're talking about. 00:36:29.391 --> 00:36:34.102 Like the ideals of the web is decentralized, you know, 00:36:34.207 --> 00:36:35.287 1-acemarke: their own domain. 00:36:35.412 --> 00:36:35.832 2-vcarl: right? 00:36:35.892 --> 00:36:40.637 Sharing knowledge, sharing skills, sharing tools, and I love that. 00:36:40.637 --> 00:36:41.447 I agree with that. 00:36:41.447 --> 00:36:43.367 I

aspire to those ideals. 00:36:43.637 --> 00:36:47.259 But it's also true that like, we kind of just want stuff to work. 00:36:47.553 --> 00:36:50.853 we want it to just kind of be stable and, you know, it's not stable, 00:36:51.123 --> 00:36:56.013 decentralized things that have a smattering of volunteers maintaining it. 00:36:56.383 --> 00:36:57.703 That's kind of the distinction. 00:36:57.733 --> 00:37:02.214 Like that's kind of the trade off is like over time decentralized things 00:37:02.214 --> 00:37:09.184 with enthusiastic maintainers and, open source over time that stabilizes 00:37:09.184 --> 00:37:14.404 and grows and becomes great in a way that proprietary code can't, because, 00:37:14.404 --> 00:37:15.484 you know, you can't inspect it. 00:37:15.484 --> 00:37:17.224 You can't say like, I'm encountering this bug. 00:37:17.224 --> 00:37:18.304 Let me go look at the code. 00:37:18.544 --> 00:37:20.164 Oh, here's where it's coming from. 00:37:20.164 --> 00:37:24.994 Let me tell them about it and give them a reproduction and contribute to a fix. 00:37:25.144 --> 00:37:30.583 Like, if I encounter a problem like that in, any of these many proprietary 00:37:30.793 --> 00:37:34.464 tools that I use, then there's just no path for me to do that. 00:37:34.464 --> 00:37:39.672 So, but on the other hand, they have the capital and the employees 00:37:39.672 --> 00:37:43.212 to have people proactively looking for that, where it's their job. 00:37:43.663 --> 00:37:44.413 that's the tension. 00:37:44.563 --> 00:37:49.273 I tend to believe over a long enough time scale that distributed, 00:37:49.273 --> 00:37:55.793 decentralized things like the web will be better than anything that's 00:37:55.793 --> 00:37:57.803 like proprietary developed, but. 00:37:57.955 --> 00:38:00.025 It has to work well enough to get the attention. 00:38:00.425 --> 00:38:04.235 And you know, in this attention economy we're in, it's very hard to get people who 00:38:04.235 --> 00:38:08.945 have enough bandwidth of their attention to say, I have found this problem. 00:38:08.945 --> 00:38:12.095 Let me chase it down to its logical conclusion and contribute 00:38:12.095 --> 00:38:15.365 to a fix being actually shipped for me and people like me. 00:38:15.983 --> 00:38:18.473 1-acemarke: aren't gonna go off on this, but I could even see some 00:38:18.473 --> 00:38:22.583 tangential connections with things like open source as a concept, 00:38:22.613 --> 00:38:24.413 open source maintainer, burnout. 00:38:24.783 --> 00:38:29.103 Like most people, like using an app or a library or even even developers using 00:38:29.103 --> 00:38:33.063 a library, like in the end, do they actually care that it's open source? 00:38:33.453 --> 00:38:37.533 No, they just want to download the thing and have it work, whether it's Libre 00:38:37.533 --> 00:38:43.653 office or you know, a paint, you know, paint program or better off or whatever. 00:38:44.173 --> 00:38:47.173 And

the fact that the code's open doesn't matter. 00:38:47.173 --> 00:38:51.692 It's like I, as the consumer of this thing at this level just want it to work. 00:38:51.932 --> 00:38:55.952 And if it doesn't, I ask the person who builds it, just make it work. 00:38:56.447 --> 00:38:59.451 2-vcarl: Right, I'll call out a couple of quotes from here. 00:38:59.581 --> 00:39:03.471 Beautiful data, visualizations are free to make, but the supply of people 00:39:03.471 --> 00:39:06.531 who really love and know D three is a lot lower than I expected it would be. 00:39:07.051 --> 00:39:09.661 I love home-cooked apps and malleable software, but I have 00:39:09.661 --> 00:39:12.121 a gnawing feeling that I'm in a bubble when I think about them. 00:39:12.151 --> 00:39:15.361 Most people's lives are split into the things that they affect and create 00:39:15.361 --> 00:39:17.581 and the things that already exist and wanna tune out and automate. 00:39:17.911 --> 00:39:19.685 I think that's a, I think that's insightful. 00:39:19.685 --> 00:39:23.650 I think that speaks to a deep, I don't know, tension in this. 00:39:23.650 --> 00:39:28.958 Like, yes, AI lets you do a much larger range in a much shorter 00:39:28.958 --> 00:39:31.358 time span than used to be feasible. 00:39:31.838 --> 00:39:34.076 But do people want to do that? 00:39:34.136 --> 00:39:34.466 I don't know. 00:39:34.466 --> 00:39:37.046 Given the, given a range of infinite possibilities, how do you pick 00:39:37.046 --> 00:39:41.154 which one that you are going to do and then give it the attention so 00:39:41.154 --> 00:39:45.054 that it actually produces something that other people can benefit from? 00:39:45.348 --> 00:39:48.318 How do you find the things that you want to build on top of, you know. 00:39:49.016 --> 00:39:52.556 Well, I guess I'll transition from that, from that comment to this other post. 00:39:52.826 --> 00:39:56.366 When everyone's a developer, how do we promote the web platform overreact? 00:39:56.466 --> 00:40:00.153 It's in reference to this Alex Russell Post that I'll, I'll, I'll share after. 00:40:00.424 --> 00:40:03.676 But I guess it's talking about, it's talking about that like when people 00:40:03.676 --> 00:40:06.406 can make their own software, when they can just use plain English to 00:40:06.406 --> 00:40:13.316 ask AI to generate code that in one form or another, solves the problem. 00:40:13.316 --> 00:40:15.416 They stated, you know, maybe not the way they intended, maybe not 00:40:15.416 --> 00:40:18.597 the way they wanted, but it, it's doing its best just like all of us. 00:40:18.755 --> 00:40:20.975 It's talking about how if you just ask it to make something, it's 00:40:20.975 --> 00:40:25.234 probably gonna use React, because that's what, by volume of text, by 00:40:25.234 --> 00:40:26.674 volume of code that it's trained on. 00:40:26.674 --> 00:40:27.394 That's what. 00:40:27.556 --> 00:40:30.376 That's the mode, that's the most common example. 00:40:30.801 --> 00:40:34.191 I guess like this is framed as a, an AI problem. 00:40:34.191 --> 00:40:35.901 Like why does AI do this? 00:40:36.231 --> 00:40:39.942 But this is actually the same problem that I've observed well before AI 00:40:39.942 --> 00:40:44.096 came out, you know, 2022 or when everyone chat GPT really started. 00:40:44.380 --> 00:40:47.290 I have observed all of these problems that it's talking about 00:40:47.290 --> 00:40:48.580 over the life of my career. 00:40:48.580 --> 00:40:52.180 Like, you know, especially I feel like, you know, there's a 00:40:52.180 --> 00:40:57.010 certain golden age of tech between 2018 and, I dunno, 2021 or so. 00:40:57.790 --> 00:40:59.706 And that was absolutely the case. 00:40:59.706 --> 00:41:03.246 Like I would, you know, join a new workplace because I did that a lot. 00:41:03.246 --> 00:41:05.556 I worked at a lot of different companies and I'd look at the 00:41:05.556 --> 00:41:06.846 practices people were using. 00:41:06.846 --> 00:41:08.826 It's like, why did you use React for this? 00:41:08.826 --> 00:41:13.056 Like there's, here's a platform feature that solves this more easily. 00:41:13.671 --> 00:41:14.751 And with less code. 00:41:14.751 --> 00:41:17.061 And a lot of the time people just didn't know it existed. 00:41:17.061 --> 00:41:18.711 Like it's that attention problem. 00:41:18.711 --> 00:41:20.781 How do you know what you should build on top of? 00:41:21.361 --> 00:41:21.661 don't know. 00:41:21.661 --> 00:41:26.266 It's, the solutions here that people are talking about is like, what can we, the 00:41:26.266 --> 00:41:30.856 web community do to promote web platform features overreact code and it proposes 00:41:30.856 --> 00:41:36.096 teach vibe coders to explicitly prompt for web native solutions get the LLMs to 00:41:36.096 --> 00:41:41.486 ingest more web platform code, spotlight teams and projects that ship modern web 00:41:41.486 --> 00:41:43.466 experiences without heavy frameworks. 00:41:43.856 --> 00:41:48.002 I mean, like, that's, to me all of those, essentially just like how do 00:41:48.002 --> 00:41:50.012 you make the web platform more popular? 00:41:50.318 --> 00:41:53.357 to me, AI fundamentally spits out an average. 00:41:53.387 --> 00:41:57.917 It spits out like you ask it to do something and it will give you a deeply 00:41:57.917 --> 00:42:03.497 profoundly average result for, you know, if you ask a thousand people, 00:42:03.497 --> 00:42:08.616 that question here is approximately what the mass of those people would say. 00:42:08.825 --> 00:42:15.182 And it's true that if you asked a thousand developers to do something 00:42:15.212 --> 00:42:20.042 on the web, the most common output would probably be a kind of shitty 00:42:20.042 --> 00:42:24.182 implementation with React that overused things like use state news effect because 00:42:24.182 --> 00:42:26.192 that's what developers did and do. 00:42:26.582 -->

00:42:29.342 So, I don't know, it's like a lot of these questions that we ask 00:42:29.342 --> 00:42:34.446 ourselves about AI and code, I feel like distill down to the same kinds 00:42:34.446 --> 00:42:36.546 of questions we were asking before. 00:42:36.861 --> 00:42:39.260 But like now, there's a boogeyman of ai. 00:42:39.260 --> 00:42:43.820 Like this is ai, this is this, this is code, this is a product people have built. 00:42:44.025 --> 00:42:48.675 And so it is now a very convenient boogeyman, but it, it's a distillation of 00:42:48.945 --> 00:42:53.775 the thoughts of a vast number of people as codified by what they published and 00:42:53.775 --> 00:42:55.575 what was ingested for the training set. 00:42:56.032 --> 00:43:00.189 1-acemarke: I just tossed in a link to the obligatory XKCD from like a decade 00:43:00.189 --> 00:43:06.129 ago, which is the, the one of the points out that sometimes your, your map or 00:43:06.129 --> 00:43:11.199 your heat map of users is actually just a heat map of the population. 00:43:11.199 --> 00:43:13.119 It's a one-to-one correspondence. 00:43:13.633 --> 00:43:13.933 2-vcarl: Yeah. 00:43:13.948 --> 00:43:17.488 1-acemarke: So I think that that is partly what we see with like react usage. 00:43:17.518 --> 00:43:20.218 You know, there, there's a lot of people who use React. 00:43:20.278 --> 00:43:25.522 There's a lot of people who use React badly and therefore like, the spread of 00:43:25.642 --> 00:43:32.482 types of usage of React, the quality of usage of React kind of maps to the quality 00:43:32.482 --> 00:43:34.732 of programmer skills and backgrounds. 00:43:35.128 --> 00:43:37.178 Like, not necessarily an excuse. 00:43:37.448 --> 00:43:41.138 And that's not to say that like everything about using React is, is 00:43:41.138 --> 00:43:46.028 good by itself, but just that when you have a full cross spectrum of the 00:43:46.028 --> 00:43:50.948 population using a thing, you're going to have the full cross spectrum of the 00:43:50.948 --> 00:43:54.428 population using it in good and bad ways. 00:43:54.998 --> 00:43:56.648 And so in the same way when it like. 00:43:57.348 --> 00:44:01.518 , when it comes to, you know, people building apps, re React got popular. 00:44:01.968 --> 00:44:05.298 Where LLMs came out and saw that was re react was popular. 00:44:05.328 --> 00:44:08.718 So LLMs repeat what they were trained to do. 00:44:09.168 --> 00:44:12.318 And plus it's also, you know, like not just popular, but it's especially 00:44:12.318 --> 00:44:15.318 popular within Silicon Valley, within all the companies making 00:44:15.318 --> 00:44:16.998 the AI builder tools, et cetera. 00:44:17.298 --> 00:44:20.298 So in a lot of cases, they're prompting them to just use React 00:44:20.298 --> 00:44:23.208 because they assume it's the default that you would use anyway. 00:44:23.730 --> 00:44:28.622 2-vcarl: I wanna shout out a term that I love path dependence, which is like, where 00:44:28.622 --> 00:44:33.242 you're at right now is the outcome of a series of choices that were made in the 00:44:33.242 --> 00:44:40.647 past, and those past choices constrain what paths are available moving forward. 00:44:40.947 --> 00:44:45.582 That's path dependence, where you came from, effects where you can go and. 00:44:46.090 --> 00:44:49.240 So I guess like to me that this is a complaint about 00:44:49.240 --> 00:44:51.400 the path dependence of React. 00:44:51.430 --> 00:44:56.960 Like, because React got popular now React is, , we've worn that rut and so 00:44:56.960 --> 00:45:00.980 now it's hard to get out of that rut, even if there is a better alternative. 00:45:01.640 --> 00:45:05.567 And so I would say like the solution there isn't like, I don't know, 00:45:05.867 --> 00:45:08.177 tell people they're bad at things. 00:45:08.267 --> 00:45:13.127 The solution is to like make better resources so that a different path 00:45:13.457 --> 00:45:15.707 becomes the easiest one to take. 00:45:15.797 --> 00:45:21.290 And mass, you know, writ large when you know, it's, one thing to convince 15 00:45:21.290 --> 00:45:27.230 people that this is the best solution and it's correct and it's another to make 00:45:27.230 --> 00:45:31.466 the argument clearly enough, strongly enough, and consistently enough over a 00:45:31.466 --> 00:45:36.506 long enough span of time that it affects the behavior of millions of people. 00:45:37.496 --> 00:45:38.216 So. 00:45:38.357 --> 00:45:42.707 You know, how do we promote the web platform overreact, like do a better 00:45:42.707 --> 00:45:44.387 job promoting the web platform? 00:45:44.447 --> 00:45:51.148 'cause there are, thousands of people who are writing resources 00:45:51.208 --> 00:45:55.138 for React because that's where money and attention and time is. 00:45:55.288 --> 00:46:00.688 So if you want to do better than that, you have to make a resource that is better. 00:46:01.138 --> 00:46:01.618 I don't know that. 00:46:02.008 --> 00:46:02.428 There you go. 00:46:03.153 --> 00:46:03.373 1-acemarke: Yeah. 00:46:03.654 --> 00:46:03.804 Yeah. 00:46:03.804 --> 00:46:06.984 So the, the, the other, there was a related article from last month 00:46:06.984 --> 00:46:10.974 entitled Dead Framework Theory, which was specifically addressing like 00:46:10.979 --> 00:46:15.474 the, the LLM training data aspect of this and how it, how it becomes a, 00:46:15.474 --> 00:46:17.184 sort of a self perpetuating cycle. 00:46:17.544 --> 00:46:21.924 And also pointing out that like if you have a, a new library or a new tool or 00:46:21.924 --> 00:46:26.688 a new web platform feature that is just now baseline, newly available, there's 00:46:26.688 --> 00:46:31.248 a lag in common LLMs knowing about that. 00:46:31.444 --> 00:46:35.164 So it's sort of like that, defaults win kind of an issue. 00:46:35.408 --> 00:46:39.592 2-vcarl: Yeah, and I'll say, yes, defaults win, but That doesn't mean that it's

00:46:39.592 --> 00:46:44.505 impossible for an, you know, to, to approach the problem in a way to mitigate 00:46:44.505 --> 00:46:48.495 that kind of, you know, home field advantage that tools like React have. 00:46:48.679 --> 00:46:51.979 I've shouted out a number of times in a bunch of different contexts because 00:46:51.979 --> 00:46:55.429 they do a good job, but I love effect, one of the things they do really 00:46:55.429 --> 00:46:59.119 well is documentation, and one of the things they do really well in their 00:46:59.119 --> 00:47:01.789 documentation is making it legible for ai. 00:47:02.099 --> 00:47:06.059 I recently over the summer built a Greenfield project with Effect, 00:47:06.059 --> 00:47:08.519 and I did it extensively with ai. 00:47:08.924 --> 00:47:12.254 , a mixture of having it right and then having it educate me and then rewriting 00:47:12.254 --> 00:47:16.934 it and then having AI rewrite what I did, you know, just get there over time. 00:47:17.252 --> 00:47:21.632 But I was able to do that despite it being a new library that's even at odds with the 00:47:21.632 --> 00:47:23.582 kind of patterns that AI wants to write. 00:47:23.582 --> 00:47:28.168 It's a functional type, it's a functional programming typed standard 00:47:28.168 --> 00:47:31.468 library essentially, which is not at all what AI wants to write. 00:47:31.468 --> 00:47:33.718 AI wants to write object oriented classes. 00:47:33.959 --> 00:47:36.689 And despite that, I was able to make it, figure it out and learn it 00:47:36.689 --> 00:47:39.719 and do a good job, good enough job that I got something I was proud of. 00:47:39.929 --> 00:47:41.219 You know, not just vibe coded. 00:47:41.219 --> 00:47:42.749 I was proud of it. 00:47:42.749 --> 00:47:46.349 I told people in my life, this is the best code base I've ever worked on. 00:47:46.645 --> 00:47:50.425 And it's a large part of that was because of how they affect. 00:47:50.716 --> 00:47:53.416 As a, you know, organization and library approach, their style 00:47:53.416 --> 00:47:57.526 of documentation, they made it legible for these automated tools. 00:47:57.526 --> 00:48:02.296 So I guess like if you wanna promote the web platform over incumbents, 00:48:02.296 --> 00:48:07.066 like react, then like figure out how people are writing code, figure out 00:48:07.066 --> 00:48:13.366 how people are consuming documentation, consuming new knowledge do a good job. 00:48:13.880 --> 00:48:16.940 1-acemarke: I'll note though, that you, you are the one who decided you were 00:48:16.940 --> 00:48:21.230 going to use effect and told the LOM this is what I'm going to build with. 00:48:21.606 --> 00:48:22.056 2-vcarl: It's true. 00:48:22.146 --> 00:48:22.596 Yeah. 00:48:22.852 --> 00:48:26.632 1-acemarke: the last wink on this particular topic Alex Russell who used 00:48:26.632 --> 00:48:32.092 to be at Google now has been, Microsoft has been very vocally beating the drum 00:48:32.272 --> 00:48:38.962 that most client side js is bad, that JS bundles keep growing, that most people 00:48:38.962 --> 00:48:44.062 are using, you know, older Android devices on mobile with very weak CPUs, and 00:48:44.062 --> 00:48:49.162 therefore Js load times are worse versus, you know, developers using an iPhone or, 00:48:49.167 --> 00:48:50.932 or a Power or a Fast Mac or something. 00:48:51.412 --> 00:48:56.992 And he just put up the latest article in his ongoing series you know, documenting, 00:48:57.213 --> 00:49:03.333 sta like average bundle sizes, average network speeds, average CPU uses. 00:49:03.760 --> 00:49:07.030 I have a lot of respect for him Technically his stats are correct. 00:49:07.060 --> 00:49:08.650 His points are right. 00:49:09.250 --> 00:49:15.340 I have questions about his delivery and tone in how he accuses people of doing 00:49:15.340 --> 00:49:19.393 the wrong thing, but it's also very true that a lot of people should be paying 00:49:19.393 --> 00:49:24.043 more attention to him and actually trying to shrink down bundles instead. 00:49:24.717 --> 00:49:25.377 2-vcarl: That's interesting. 00:49:25.437 --> 00:49:29.577 I a thought I've had, I haven't validated this thought against, you know, 00:49:29.577 --> 00:49:32.176 ground truth, but been hearing about. 00:49:32.776 --> 00:49:35.086 Bundle size as a problem. 00:49:35.356 --> 00:49:39.496 Essentially what he's talking about here, like we ship too much JavaScript 00:49:39.526 --> 00:49:45.856 for the compute capabilities of commonly used devices like that as a 00:49:45.856 --> 00:49:49.691 thesis has been something I've been hearing about for a very long time. 00:49:50.191 --> 00:49:55.445 Much of my career even, and I'm not sure that the bundle size problem 00:49:55.445 --> 00:50:00.515 has gotten that much worse over the last, I'll say eight years, 00:50:00.575 --> 00:50:02.135 you know, call it, since 2017. 00:50:02.815 --> 00:50:07.288 And he shows a chart of compute speeds and basically shows like, yes, top of the line 00:50:07.288 --> 00:50:12.778 devices have gotten five times faster, low end devices that 75% of the world 00:50:12.778 --> 00:50:15.088 uses have not gotten measurably faster. 00:50:15.462 --> 00:50:17.629 And that's true. 00:50:17.945 --> 00:50:18.485 Can't argue it. 00:50:18.485 --> 00:50:19.715 His stats are correct, like he said. 00:50:20.235 --> 00:50:24.118 But also like, I'm not sure the problem is getting measurably worse. 00:50:24.178 --> 00:50:29.878 The optimizations of how we can split code and, you know, what the 00:50:29.928 --> 00:50:34.488 JIT and the execution environment is doing to make it work more quickly. 00:50:34.848 --> 00:50:36.738 Those have gotten better over time. 00:50:37.288 --> 00:50:41.428 And it's also true that like, so if he's looking at the, basically like 00:50:41.428 --> 00:50:48.854 every processor, anyone with a device is using, and that is the correct stat 00:50:48.854 -->

00:50:54.824 to consult if you are building something that everyone in the world will access. 00:50:55.484 --> 00:50:56.954 But that's not true for most startups. 00:50:57.014 --> 00:51:01.034 Like if you're building a startup that targets like some industry, 00:51:01.034 --> 00:51:02.654 that American businesses. 00:51:02.920 --> 00:51:07.090 Is target's a problem that American businesses need solved, then like it 00:51:07.090 --> 00:51:13.680 doesn't really matter what the MO mobile performance in, you know, different PO 00:51:13.830 --> 00:51:17.670 outside of that target industry, I guess are, then it's like, like what you said 00:51:17.670 --> 00:51:22.590 about his like tone and delivery, I think a, to me when I read this, a core part 00:51:22.590 --> 00:51:27.543 of what that I think he's overlooking is that not everyone is trying to ship 00:51:27.543 --> 00:51:33.333 something that 100% of the people in the world can access smoothly and quickly. 00:51:33.573 --> 00:51:34.785 Like, and that's fine. 00:51:34.785 --> 00:51:37.515 Like, you know, not everything needs to target everybody in the world. 00:51:37.863 --> 00:51:40.113 So, I don't know, I, I guess that that's one of my takeaways. 00:51:40.143 --> 00:51:42.543 That's one thing that I think is missing from this. 00:51:42.603 --> 00:51:46.533 Otherwise, very technically complete blog post, 00:51:47.000 --> 00:51:51.500 1-acemarke: I think it's also true that an awful lot of app teams out there ought to 00:51:51.500 --> 00:51:57.800 be spending time thinking about even just basic bundle sized stuff, and they aren't 00:51:58.410 --> 00:52:02.370 2-vcarl: I do wanna talk about the NPM hack that is ongoing 00:52:02.460 --> 00:52:05.010 shy haud, the sand worm. 00:52:05.411 --> 00:52:09.729 I want to call out as far as I know, I keto.dev has been like 00:52:09.759 --> 00:52:11.619 the security researchers who have. 00:52:12.033 --> 00:52:16.413 Announced this, like uncovered it and then analyzed it and shared it. 00:52:16.813 --> 00:52:21.313 They had a really good analysis of the evolution of the code, of the 00:52:21.313 --> 00:52:23.893 attack of this worm from September. 00:52:24.323 --> 00:52:28.313 And I, noted in that when I was rereading it after this, you know, 00:52:28.313 --> 00:52:33.605 new, new attack they had shouted out that it appeared that they still 00:52:33.605 --> 00:52:35.465 had more credentials in reserve. 00:52:35.555 --> 00:52:39.028 Like they had cracked more credentials than they had actually used. 00:52:39.538 --> 00:52:43.538 And sure enough, this, you know, round two attack in their analysis 00:52:43.538 --> 00:52:47.636 of of it this time around, they note that it was timed just before MP's 00:52:47.636 --> 00:52:49.346 deadline for revoking old tokens. 00:52:49.346 --> 00:52:52.466 So like very clearly they had a lot of tokens that they had compromised 00:52:52.466 --> 00:52:55.650 previously that they were now taking advantage of before they, before 00:52:55.650 --> 00:52:57.990 they lost access to it, before those tokens were no longer valid. 00:52:58.398 --> 00:53:01.548 And I guess to reiterate, we've talked about, we talked about this attack before. 00:53:01.678 --> 00:53:05.676 I would say this is, this represents the current cutting edge of malware. 00:53:05.916 --> 00:53:11.754 It is using all sorts of fascinating root execution, like privilege escalation. 00:53:11.754 --> 00:53:16.651 I saw discussion of it, starting a docker container, mounting your file system 00:53:16.681 --> 00:53:20.687 to that privileged environment and now suddenly it has privileged access to 00:53:20.687 --> 00:53:24.466 all sorts of files that it couldn't have if it was just running in your shell. 00:53:24.801 --> 00:53:26.271 Stuff like that, that's pretty fascinating. 00:53:26.271 --> 00:53:30.406 And other things, like if you have a claw installed, it will prompt 00:53:30.466 --> 00:53:34.396 Claude to say, Hey, look around this file system for secrets. 00:53:34.756 --> 00:53:37.663 Like that is very cutting edge malware to me. 00:53:37.663 --> 00:53:39.723 It's it's not just writing its own code. 00:53:39.873 --> 00:53:42.783 It's not just a worm that is trawling your file system. 00:53:42.783 --> 00:53:47.230 It is taking advantage of other tools that you take advantage of to attack you. 00:53:47.492 --> 00:53:49.410 So that's, yeah, that's not great. 00:53:49.730 --> 00:53:51.920 1-acemarke: I mean like big picture, it's almost. 00:53:52.294 --> 00:53:57.274 It's almost surprising how many widespread pieces of malware have been very dumb 00:53:57.574 --> 00:54:02.494 and weren't as nearly destructive or smart as they could have been. 00:54:02.944 --> 00:54:07.354 And like think of how much damage we've had collectively as an industry over years 00:54:07.354 --> 00:54:10.954 with only a lot of those dumb malware. 00:54:11.226 --> 00:54:15.516 And now we're starting to see some people write more intelligent malware. 00:54:15.778 --> 00:54:16.138 2-vcarl: Yep. 00:54:16.288 --> 00:54:16.558 Yeah. 00:54:16.588 --> 00:54:19.078 'cause a month ago, two months ago, whenever it was, there was 00:54:19.078 --> 00:54:23.737 a different NPM hack that managed to get into the debug package, 00:54:24.097 --> 00:54:26.197 which is massively widely used. 00:54:26.699 --> 00:54:30.989 And I assumed that it was the same attackers as this attack because it's, 00:54:30.989 --> 00:54:32.999 it it resembled it in a couple of ways. 00:54:33.339 --> 00:54:34.749 Except that it was dumb. 00:54:34.899 --> 00:54:36.849 It was so poorly executed. 00:54:36.849 --> 00:54:43.085 Like they got close to unprecedented access to the NPM ecosystem. 00:54:43.085 --> 00:54:47.255 Like this is one of the first times that, that previous attack, the debug 00:54:47.255 --> 00:54:51.382 attack, was one of the first times that I saw, here's this compromised version. 00:54:51.442

--> 00:54:53.122 Let me check my file system. 00:54:53.422 --> 00:54:53.812 Oh, yep. 00:54:53.842 --> 00:54:54.472 There it is. 00:54:54.562 --> 00:54:55.312 I've got this. 00:54:55.952 --> 00:54:58.652 So yeah, they just straight up fumbled the bag there. 00:54:58.702 --> 00:55:03.412 my theory here, having paid attention to this, it seems like there's some. 00:55:04.717 --> 00:55:10.424 Hacker group that may have compromised set up a, a phishing tool set, you 00:55:10.424 --> 00:55:16.034 know, set up tools for other malicious actors to run phishing attacks. 00:55:16.244 --> 00:55:21.222 And it feels like two different hacker groups took advantage of those tools 00:55:21.222 --> 00:55:25.872 to target NPM maintainers, and only one of them is actually technically 00:55:25.872 --> 00:55:29.922 competent enough to take full advantage of the access that they got. 00:55:29.952 --> 00:55:30.582 So, yeah. 00:55:31.048 --> 00:55:32.938 I believe we've shared these resources before. 00:55:32.998 --> 00:55:37.361 And GitHub put out our plan for a more secure NPM supply chain. 00:55:37.491 --> 00:55:38.571 So they're still working on that. 00:55:38.571 --> 00:55:38.991 Clearly. 00:55:39.291 --> 00:55:42.711 They also just released an NPM security update earlier this 00:55:42.711 --> 00:55:44.481 month in November, November 5th. 00:55:44.799 --> 00:55:47.019 Mark, you want to just read through any of these lightning rounds 00:55:47.019 --> 00:55:48.099 that you wanna call attention to 00:55:48.099 --> 00:55:48.399 1-acemarke: okay. 00:55:48.499 --> 00:55:52.669 TS six and TS seven are hopefully gonna be coming out early next year. 00:55:52.669 --> 00:55:57.139 TS six is still the current TypeScript code base, but with some 00:55:57.139 --> 00:55:59.239 changes to defaults and deprecation. 00:55:59.539 --> 00:56:03.289 And then TS seven would be the native implementation in 00:56:03.289 --> 00:56:04.339 go that they're working on. 00:56:04.339 --> 00:56:08.179 So a couple different updates on , the release plans there. 00:56:08.269 --> 00:56:10.189 That's going to be very, very exciting. 00:56:10.189 --> 00:56:12.889 I am excited to have TypeScript compiler 10 times faster. 00:56:13.439 --> 00:56:18.979 Also there were some more the latest TC 39 proposals meeting as usual 00:56:18.979 --> 00:56:21.889 advanced a number of different proposals to different stages. 00:56:22.149 --> 00:56:23.859 A couple of things that look interesting. 00:56:23.859 --> 00:56:27.369 There's a promise dot all keyed, which is essentially promise 00:56:27.369 --> 00:56:28.929 dot all but for an object. 00:56:29.212 --> 00:56:34.552 And then I am excited, irrationally excited about object, keys length, which 00:56:34.552 --> 00:56:39.022 is literally just to get the size of the number of keys in an object instead 00:56:39.022 --> 00:56:41.032 of object keys, parentheses, length. 00:56:42.450 --> 00:56:44.040 2-vcarl: Chrome and a couple of other

browsers. 00:56:44.040 --> 00:56:46.830 Want to remove XSLT from the web platform. 00:56:47.200 --> 00:56:52.243 XSLT is a what domain specific library for transforming X and 00:56:53.262 --> 00:56:55.422 1-acemarke: transformations within a browser. 00:56:55.422 --> 00:57:00.222 So you, you send out an XML document from the server and then you have a purely 00:57:00.222 --> 00:57:04.607 client side declarative transformation that like converts it into HTML. 00:57:04.940 --> 00:57:05.630 2-vcarl: It exists. 00:57:05.660 --> 00:57:07.700 I had no idea that it was a browser feature. 00:57:07.700 --> 00:57:10.700 I've known of it as a tool that exists, but I had no idea it 00:57:10.700 --> 00:57:11.840 was baked into the browser. 00:57:12.350 --> 00:57:15.050 And they're looking to remove it and I'm okay with that. 00:57:15.380 --> 00:57:19.460 But certain people who I guess use it for one thing or 00:57:19.460 --> 00:57:22.370 another have been running a campaign to say, no, please don't. 00:57:22.590 --> 00:57:23.850 What about RSS feeds? 00:57:24.068 --> 00:57:24.338 But yeah. 00:57:24.806 --> 00:57:27.581 1-acemarke: Going along a bit with the whole use the platform thing. 00:57:27.767 --> 00:57:31.487 more people should be putting some of their state in the URL 00:57:31.637 --> 00:57:35.807 and there was a very good article on using the URL as state. 00:57:36.141 --> 00:57:39.801 And then along with that, another, video from recent React Summit that's 00:57:39.801 --> 00:57:42.381 probably paywall at the moment, but I think he's done a couple previous 00:57:42.381 --> 00:57:43.791 versions of this at different conferences. 00:57:44.071 --> 00:57:49.651 David Korshe has a talk on goodbye use state, similar to his goodbye use effect. 00:57:49.961 --> 00:57:50.951 It's hilarious. 00:57:50.951 --> 00:57:54.131 It's funny, and it also points out why you shouldn't be using 00:57:54.161 --> 00:57:56.141 actual used state all that much. 00:57:56.544 --> 00:58:00.024 2-vcarl: Aiden by who did Millions js put out an interesting 00:58:00.024 --> 00:58:01.674 new tool called React Grab. 00:58:01.944 --> 00:58:06.204 Basically letting you, you know, you, is trying to solve the problem of 00:58:06.204 --> 00:58:09.919 when you tell AI like, Hey, make this thing different, and then it goes. 00:58:10.704 --> 00:58:11.904 I don't know where that is. 00:58:12.020 --> 00:58:14.330 here's your browser, but I don't know where that is in the code base. 00:58:14.390 --> 00:58:16.850 And so this is a tool that helps map. 00:58:17.059 --> 00:58:20.639 similar to how you can use dev tools to highlight an element this is 00:58:20.639 --> 00:58:24.629 taking that kind of behavior where you highlight an element in your browser 00:58:24.989 --> 00:58:29.129 and it connects back to the actual source code that produced that element. 00:58:29.589 --> 00:58:34.499 As a means of giving AI a head start on finding what code to 00:58:34.499 --> 00:58:36.019 use which seems super cool. 00:58:36.069 --> 00:58:41.949 I question the wisdom of the recommended get started thing, which is just 00:58:41.949 --> 00:58:43.899 to add a script tag to Unpackage. 00:58:44.209 --> 00:58:44.599 I don't know. 00:58:44.689 --> 00:58:49.399 Seems, seems slightly sketch, but also for also seems really easy to to get started. 00:58:49.746 --> 00:58:52.521 1-acemarke: One thing that caught my eye in terms of like, this is awesome 00:58:52.521 --> 00:58:53.541 and you should never, ever do this. 00:58:53.841 --> 00:58:58.641 Was implementing a node module loader that imports from BitTorrent. 00:58:58.975 --> 00:58:59.905 okay, great. 00:58:59.905 --> 00:59:02.060 That's possible, apparently. 00:59:02.490 --> 00:59:03.180 2-vcarl: Love that. 00:59:03.180 --> 00:59:04.530 I do love BitTorrent. 00:59:04.530 --> 00:59:05.460 You know, as we were talking about 00:59:05.460 --> 00:59:09.433 decentralized technology a little bit earlier, like, man, what a, highly 00:59:09.433 --> 00:59:11.653 resilient, decentralized technology. 00:59:12.023 --> 00:59:15.233 But yeah, well, hey, you know, maybe if we're gonna, if we're gonna encourage 00:59:15.233 --> 00:59:19.023 people to do decentralized technology building on the shoulders of giants 00:59:19.023 --> 00:59:23.013 like it, man, if you want it to like, never stop working, ever Sure. 00:59:23.013 --> 00:59:25.803 Import your code from BitTorrent, probably, it's not gonna work very 00:59:25.803 --> 00:59:31.337 quickly, but given the right underlying cedars that would, that may never go down. 00:59:31.607 --> 00:59:32.207 Pretty interesting. 00:59:32.757 --> 00:59:37.257 1-acemarke: Earlier I mentioned Ryan Cardo streaming and talking about async signals. 00:59:37.257 --> 00:59:41.757 Here's the six hour livestream video if you want to watch that and or 00:59:41.757 --> 00:59:43.497 throw it into a summarizer somewhere. 00:59:43.843 --> 00:59:48.283 2-vcarl: I'll shout out the details of building nodes the node type 00:59:48.283 --> 00:59:49.993 script type stripping support. 00:59:50.363 --> 00:59:53.513 Just 'cause I, man, I do love a technical writeup and that is, 00:59:53.851 --> 00:59:56.071 that was a massive feature release. 00:59:56.164 --> 01:00:00.454 Just like the ability to use node to run TypeScript code without doing 01:00:00.814 --> 01:00:03.334 a compilation step was pretty cool. 01:00:03.754 --> 01:00:07.324 And so here to here to signal boost. 01:00:07.585 --> 01:00:08.815 A post from the author of it. 01:00:09.136 --> 01:00:10.186 1-acemarke: And one more. 01:00:10.186 --> 01:00:13.736 Since we spent much time talking about the, the web platform the author 01:00:13.736 --> 01:00:17.936 of the Motion Animation Framework put up a really good post comparing 01:00:17.966 --> 01:00:21.686 all the different ways you can do animations on the web and talking about 01:00:21.686 --> 01:00:22.611 some of the performance

trade-offs. 01:00:23.123 --> 01:00:23.783 2-vcarl: Super good. 01:00:23.934 --> 01:00:25.922 Cool into some conference news. 01:00:26.222 --> 01:00:27.902 We don't really have much 'cause it's the end of 01:00:27.902 --> 01:00:31.230 the year and all of the upcoming conference things don't show 01:00:31.230 --> 01:00:33.030 anything for like January or February 01:00:33.359 --> 01:00:35.729 1-acemarke: well, we don't have much of a list here, but I can say that 01:00:35.729 --> 01:00:40.079 I've actually been doing my own, trying to prepare CFPs for next year, 01:00:40.079 --> 01:00:43.632 and it ended up making a database of conferences and like my own 01:00:43.632 --> 01:00:45.617 personal list is like dozens of them. 01:00:45.617 --> 01:00:48.287 And I can tell you there several conferences have submission 01:00:48.287 --> 01:00:50.379 deadlines like by November 30th. 01:00:50.803 --> 01:00:52.693 Suffice would say there's a lot more conferences out there that 01:00:53.143 --> 01:00:56.503 knew existed until I spent several evenings researching them and trying 01:00:56.503 --> 01:00:57.553 to put this database together. 01:00:57.994 --> 01:01:00.154 2-vcarl: I've tried to do that consistently over the past three 01:01:00.154 --> 01:01:03.934 years for this podcast and holy shit, there are so many conferences. 01:01:03.964 --> 01:01:05.524 It is absolutely bananas. 01:01:05.524 --> 01:01:08.044 Some of them are even fraudulent, so it's like you, 01:01:08.044 --> 01:01:09.454 you actually gotta reschool them. 01:01:10.144 --> 01:01:11.224 So, yeah, it's a lot. 01:01:11.284 --> 01:01:11.824 But okay. 01:01:11.824 --> 01:01:14.394 All that to say I am aware of React. 01:01:14.394 --> 01:01:18.424 Paris has a CFP currently open that's gonna be in March. 01:01:18.424 --> 01:01:21.664 So if you're looking to talk, then they've got a CFP open. 01:01:21.694 --> 01:01:22.504 You can do that. 01:01:22.954 --> 01:01:28.174 JS World also likewise has a CFP that closes at the end of December. 01:01:28.605 --> 01:01:32.277 Thank you everyone who has managed to stick around for this, you know, 01:01:32.427 --> 01:01:34.437 hour and 15 minute recording session. 01:01:34.597 --> 01:01:37.782 1-acemarke: Funny, we said we didn't have a lot of main content for today and go 01:01:37.782 --> 01:01:39.522 figure, we had a lot to say about stuff. 01:01:39.663 --> 01:01:39.843 2-vcarl: yeah. 01:01:39.843 --> 01:01:40.233 Right. 01:01:40.293 --> 01:01:43.173 I give us a, give us a blank slate and we will fill it. 01:01:43.898 --> 01:01:44.228 howdy. 01:01:44.743 --> 01:01:45.011 joining us. 01:01:45.446 --> 01:01:46.976 1-acemarke: the discussion has been useful. 01:01:47.244 --> 01:01:49.524 2-vcarl: that is my hope I, if not useful, at least. 01:01:49.524 --> 01:01:50.124 Interesting. 01:01:50.514 --> 01:01:50.994 Cool. 01:01:51.324 --> 01:01:52.614 Thank you for joining us. 01:01:52.704 --> 01:01:56.574 We normally, I say we'll be back on the last Wednesday of the month, but the 01:01:56.574 --> 01:01:58.854 last Wednesday of next month I think is, 01:01:59.012 --> 01:01:59.942 1-acemarke: December 31st, 01:02:00.239 --> 01:02:02.704 2-vcarl: So I don't know, we're probably not gonna do it on the last Wednesday 01:02:02.704 --> 01:02:04.414 of next month, but we'll be back 01:02:04.851 --> 01:02:07.131 1-acemarke: somewhere about that timeframe, approximately. 01:02:07.161 --> 01:02:09.231 2-vcarl: ish before the end of this year. 01:02:09.631 --> 01:02:09.991 Yeah. 01:02:10.231 --> 01:02:13.261 And then we'll, we'll be back in your podcast feed just as soon as we can. 01:02:13.261 --> 01:02:17.415 In general generally, I say we gather sources from all sorts of news feeds 01:02:17.415 --> 01:02:22.019 and whatever, but mark, you've done a good job of being my source lately. 01:02:22.019 --> 01:02:26.309 So mostly we get news from tech news and reads, or at least that's where I get it. 01:02:26.309 --> 01:02:27.299 I don't know where Mark gets it. 01:02:27.799 --> 01:02:30.889 1-acemarke: all of the above newsletters, plus like Hacker 01:02:30.889 --> 01:02:34.647 News and Twitter and all the other social media feeds, I obsess over. 01:02:34.975 --> 01:02:37.735 2-vcarl: But yeah, we do subscribe to this week in React Bite that 01:02:37.735 --> 01:02:40.075 Dev React status next JS weekly. 01:02:40.405 --> 01:02:43.765 Mark and I are both subreddit mods of React js. 01:02:44.255 --> 01:02:48.005 But yeah, if you see anything that seems useful that you might want us to 01:02:48.005 --> 01:02:51.884 discuss drop it in the Tech News and Reads channel I'll at least read it. 01:02:51.914 --> 01:02:52.634 I may not 01:02:52.634 --> 01:02:53.594 bring it in, but yeah. 01:02:53.739 --> 01:02:56.519 if this is a show that you get value from and wanna support best way to 01:02:56.519 --> 01:03:00.301 do so is by telling people about it and submitting a review that'll 01:03:00.301 --> 01:03:02.821 help us get discovered and whatever. 01:03:03.198 --> 01:03:03.648 Cheers. 01:03:03.738 --> 01:03:04.338 Thanks so much. 01:03:04.550 --> 01:03:05.390 1-acemarke: Happy Thanksgiving. 01:03:05.704 --> 01:03:06.964 2-vcarl: Yes, happy Thanksgiving.