Issue #54: Nothing Else Matters
It was a hell of a decade, the last one — from 2016 to 2026. Not only globally but even personally, if anyone looks back at the time they had during these years, they’ll find a lot of bad and good things. Maybe more of the former but still. For web development with frameworks and metaframeworks, which is the focus of this newsletter, this decade was groundbreaking (compare it to the 70 years of Fortran and realize how different the numbers and outcomes are!). Is it the peak of it? Is it the end of the good ol’ JavaScript dance on the cornerstones of the web platform? Let’s find out by looking at some highlights of the passed four weeks and ruminating on some memories floated up by the news of them.
The Good
The realization of this decade and its significance came to me with some news about Svelte and Nuxt celebrating their 10-year anniversaries this month. But with all my love for both of them, if you start going down memory lane, October 2016 and 2016 itself were much more than that.
2016 was the year of Angular 2 wars. It was the year of Vue 2 which changed the acceptance of developer experience forever. It was the year of create-react-app and tools that could make CRA almighty — wide Redux adoption, yarn as a robust npm alternative, Electron 1.0, GraphQL, to name a few (so basically, kind of the year of metaframeworkization of the ecosystem!). Even the cross-platform .NET Core 1.0 (which I personally hate but still can’t deny the importance of). On top of that, it was the year which made ES6, service workers, and PWA state of the art in web development and laid the foundation for what we have now with really minimal changes. We even started to get things like Preact (which has just got its first major update in seven years — so stable it is!) popularized to a wide extent as a replacement for bloated ecosystem tooling, which made the shift to a performance-oriented future possible, including, tangentially but still, things like Vite.
All in all, that’s a lot to re-digest, revisit thoughtfully, and give thanks to (or feel strong PTSD about, you do you). Today’s world of JavaScript tooling is something much more boring — unfortunately or not. But in many parts, the developer experience we got throughout these ten years is not bad at all.
The Bad
Things these days are starting to move backwards slightly. From flourishing and opinionated takes to generalization and consolidation. One example is the many recent acquisitions, including the latest news about Deno becoming a part of Cloudflare (reads like “good JS runtimes do not matter this much anymore”). In the same vein, some (*coughs ignorant *coughs) people, even having discussed that with about half a dozen authoritative peers and being professional DevRels (which may explain things though, I agree), say frameworks don’t matter much anymore. One other thing that made me cry too being added to this box was the talk of Ryan Carniato at this month’s ViteConf (I highly recommend the other talks too, many of them just as nice and nostalgic as this decade’s celebration itself) which kinda buried SolidStart (and, let’s accept it, hinted at burying the idea of other metaframeworks too) alive in favor of vite.config.ts. Ryan wasn’t the first one with that, as I figured the Qwik team is getting rid of their Qwik City (their metaframework thingy) in favor of Qwik Router in the newest framework version. This idea was probably popularized even earlier by React Router with their framework mode (which got some updates in the latest new version 8.5, by the way) instead of the actual React-Router-based Remix metaframework (which is coming back though, and cooking something cool too!).
Which doesn’t stop curious people from creating new metaframeworks too, actually — like the new Svelte thing for those who don’t accept SvelteKit as something one could call “sufficient” for the job. Well, as those JS-framework haters say, that could be worse… For instance, those slopware trinkets from the very aforementioned Cloudflare, huh? (Though in fairness, maybe they buy all these projects lately to make their hackwork look better on the market, who knows…)
But with all of that, we shouldn’t forget that a metaframework is actually the robust harness (in many meanings), the well-integrated exoskeleton which leads us to a better, more solid, and safer future, contrary to what many surface opinions state. We’re not there yet though, that’s for sure, even looking at the steady flood of exposed Next.js vulnerabilities or multiple Vite threats, but the integrated tooling gives dev teams a good prop for proactive measures, as opposed to a scattered set of tooling pieces, and that’s something to not forget about.
The Noteworthy
The updates of the last month in the metaframeworks community are too abundant to go too deep into, but there are some important releases which we’d make a crime if we didn’t mention today.
The SvelteKit team released the new long-awaited major version 3.0 (with lots of “Vite-ification”-related updates, and the following bunch of important minor fixes). The team is preparing for the annual SvelteSummit and inviting everyone to celebrate the new release with them in beautiful Ljubljana.
But if custom DSLs are not something of your type, you can check out the new TSX-enabled minor version of Next.js by reading about that on the team’s newly revamped and redesigned website (from the number of blog post authors, comparable with the number of SvelteKit users probably). It brings the new Cache Components feature, gifts for your greedy agents, and lots of minor DX improvements (and experiments!), as usual.
And if one triangle there is not enough for you, you’ll definitely be satisfied with double-triangle news — the new minor version of Nuxt — and pleased with its countless technical improvements and new experimental features.
As we discussed already, SolidStart gives us a farewell (probably) with their latest minor (and quite huge) release 2.1. At the same time, another start, TanStack Start, eventually reaches a stable release, with the accompanying deep philosophical writeup which you should definitely check out if this newsletter issue has touched your nerves a bit.
Astro delivers its new minor version 7.4 with predominantly technical updates and fixes, while AnalogJS is coming full steam towards new major version 3 with the corresponding out-of-the-press migration guide.
Inferno reaches the long-awaited juvenile version 10 after 2 years since the previous major update (or even 4 if you consider that version 9 was a kind of maintenance release). The framework switched to the compiler approach (as many folks do these days) and makes a lot of adjacent shifts for that.
Last but not least, RedwoodSDK, if you look at their release notes for the recent releases and the latest 1.7.5, starts to feel a bit stagnant, while Quasar, by contrast, moves forward full steam even through the minor releases like 2.34.0.
So that’s the ever-changing life in the world of metaframeworks and tools around them in the shiny and warm October of 2026. It’s like the innovation boom that started ten years ago had never slowed down since, no matter what new industry shifts reality (or future) brings us (including the frameworks funerals, and web funerals, and development funerals, and all that silly whistle jazz). I see a lot of good stuff still, a lot of good people who bring new ideas and solutions, and a lot of ideas, in me and you (yes, you!), to make the world spin even without rotation tokens or whatever they put into the furnaces these days. And while we all are here, it’s worth keeping it spinning.
đź‘‹