Issue #22: High Voltage

🎵 …you wanna see me doing my thing, all you gotta do is plug me into… 🎵

The focus of the new issue of Metaframeworks Weekly is taken by the tool historically bringing high voltage of development dynamics and consumer curiosity, along with accompanying open-source movement energy and sparks of interest from all sides of the web development ecosystem and beyond. But enough puns and dad’s jokes, let’s get to it!

The Good

Vite 7 with the experimental Rolldown core got published the last week. It doesn’t matter this much what the update itself consists of (though the changes are quite interesting, of course), it rather that this release marks a kind of philosophical shift to the hybrid TypeScript/Rust approach to web dev tooling which is becoming quite popular (definitely check out this series Josh Goldberg, for example!).

The infrastructure vendors turn to Rust more and more (we have SWC, Rspack, Turbopack, and Farm, just to name a few) in the search for blazing speeds and maximum portability and scalability. I personally am all into this movement too so I’m in full 🍿-mode looking into the future of Vite (and other projects around) which, being the foundation of mostly all the metaframework ecosystem, became integral part of it forever. And with these changes I don’t believe Vite is going to give away its positions to alternatives. Which is cool because the team has a huge experience of not only delivering brilliant quality, but also maintaining the backward compatibility (with Rolldown matching the bulletproof Rollup API perfectly now).

Now you can continue using Vite as you normally would, but with the added performance benefits of Rolldown. […] Alongside this, we plan to “rustify” more of Vite’s internals to reduce communication overhead and unlock even greater performance gains.

The Bad

A new tasty kind of metaframework appeared on web ecosystem radars recently in the face of Hashbrown.

…a framework for building joyful, AI-powered user interfaces that predict users’ next steps, speeds them through forms, and intelligently reorganizes views based on context.

It’s a bold [React || Angular] powered bet on AI-powered apps in the vein of Vercel’s AI SDK providing a nice opinionated set of shovels for AI fever victims. It’s built by the team of Brian Love and Mike Ryan, the gang of GDEs famous and trusted in the Angular community (but went weird commercialization way lately). It’s probably too early to evaluate the tool (which is open-source and MIT-licensed BTW) for now (the docs are minimal, or even empty for React) but the idea is clear and I believe it will find its audience in the dev community. It’s just, it’s probably too opportunistic, but it’s hard to blame the guys for (quite common unfortunately) desire to pivot on the humane stupidity during these harsh times.

On the positive note, it (the official website) is built up on Analog which is a cool metaframework choice for such project.

The Noteworthy

No matter how you refer to the (comparatively) recent updates in RedwoodJS and the team turning it into a personal-software-dedicated SDK, the project is evolving in a pretty fast way and just recently reached the first notable milestone of version 0.1 which is (unofficially but in fact) can be considered v1 for such kind of tool (I wish I had the level of productivity this guy Justin has). It brings some breaking changes but if Cloudflare is your thing and you support the project’s philosophy itself, it will be worth it.

The other recent new releases that deserve to be mentioned are the 11ty’s v3.1.2 patch (whose scope a good Next’s minor would envy, FWIW), TanStack’s v1.122.0 with minimal but interesting fixes and improvements, and the update with v1.19.0 from Analog guys, pretty tightly coupled with the main topic of today’s issue.

On the side note, I have just discovered that Jest had got its version 30 which is very anticipated (first major in three years!) and very cool simultaneously. In the metaframework world we got used to betting on Vitest but Jest is the tool that made me eventually cope well with unit testing many years ago when it emerged. The release notes promise a lot of breaking changes and new limitations but at the same time some huge performance improvements, so it’s very enticing and scary at the same time but I personally, as an enterprise software developer, would strongly consider the migration, even through hurdles.

Last but not least, I mentioned in the previous issue that the Astro team had released the promising experimental feature of live content collections in the latest release. Now if you’re still intrigued by that (as you should be!) there’s the deep dive from Matt Kane on this topic, which explains the reasoning and mechanics and helps you play with this new awesome feature today.

Which I highly recommend you to do now, after you reached up to this point (but just after you update all your projects to Vite 7 and Jest 30, of course). I’ll ask about the results the next time.

👋