Issue #12: Changes

🎵 Turn and face the strange ch-ch-changes… 🎵

There are always some changes in tech. They happen constantly; the metaframeworks ecosystem is not an exception. The magnitude of change is always different though. You can easily see it for yourself in this newsletter’s archive, for instance. The trickiest thing is to figure out if the magnitude is really that high as it’s declared, or it’s just pure hype/marketing/whatnot. In this week’s issue, we’ll have some practice in that.

The Good

The RedwoodJS team had come with a huge (and unexpected) split. As their new manifesto says (in a pretty cheesy nostalgic way but nevertheless),

It’s a personal software revolution. And it’s coming.

Which led the guys to create a new RedwoodSDK project dedicated to personal software exclusively. That’s probably a pretty smart move in the age of AI enablement for amateur software creation, and I empathize with that a lot. It’s just a bit unexpected for one of the most business-oriented metaframeworks around. Obviously, the startups-focused project part, renamed to “Redwood GraphQL,” will not be the main focus from now on:

Redwood GraphQL (formerly the RedwoodJS you use today) will continue to receive security patches and critical updates, ensuring you can confidently rely on it for your projects.

I think the framework authors try to find their place in the ecosystem with their simplified React Server Components support, as their “Quick Start” guide became much slicker than before, obviously. And there’s no “Before” anymore, as I can see, as all the official links lead to RedwoodSDK docs now.

Also, it’s not clear what destiny awaits the Redwood Startup Fund project (alas, I have missed the chance to come up with the Next Big Thing written in Redwood). As personal software rarely requires funding, the emphasis on startups support will most probably vanish. Which is not exactly bad; it’s just that it was a unique niche. Not the most efficient probably, as this story shows.

Anyway, I’m quite curious about what this all comes up to eventually and going to follow this romantic and contrarian (I mean, as opposed to the enterprise development world of Angulars and NextJSes) journey with my pom poms ready.

The Bad

The Express Technical Committee brought the news about making Express v5 the latest officially after more than half a year since it was released.

If you wonder why this topic is in the “Bad” part of the newsletter, I’ll tell you: because it marks the end of the epoch, and I feel even more nostalgic than after reading the “ode-to-hacking” personal software manifesto above. Express 4 was the go-to tool for the Node.js ecosystem in terms of building web application backends for 10 years, and for web development industry it basically means forever. And even if it’s not going to be EOL’ed for one and a half years more, it feels like a farewell to a friend. Especially in the new epoch of ubiquitous soulless Rust-based tools (OK, I’m kidding, I love Rust almost as strongly as Express 4).

Express is not used (directly) by new shiny metaframework tools, though it’s engaged by default in many ecosystem projects, and, for instance, became a basis for default SSR implementation in Angular 19 (which is actually a kind of metaframework by itself in this regard).

The v4-v5 migration path is quite daunting for large applications, especially taking into account the unlimited freedom Express always gave to developers. But I believe there will always be companies that will help to cope with legacy projects successfully. And in fact, maybe this will incentivize more developers to switch to newer solutions and alternative approaches, which is always good for ecosystem evolution.

The Noteworthy

And the evolution doesn’t make us wait for itself. The “neu wave” (the typo is intentional) of ambitious tooling projects is covering developers with the head, promising new performance highs and returning to standards, roots, and happy life enabling a lot of personal software opportunities (here we go again).

Nue, the metaframework-like tool which is not actually that new but still, is aiming to help you to “take HTML, CSS, JS, and WASM to their absolute peak”. As the docs are quite hip and scattered (yet?), it’s not clear how exactly should it happen, but the idea is to focus on performance through making the things the opposite way to how React does them.

Nue is for developers who wish to build on classic algorithms and data structures over React idioms.

The greatness of React is in the fact that its flaws and inefficiencies are the greatest drivers of the industry progress lately which we all should appreciate. I like everything that happens because of that (which is one of the reasons for this newsletter to exist). And I like and sympathize with Nue ideas pretty much, as (even being complex in terms of the implementation part) bringing the potential high-performance software enablement and focus on pure technologies (without different CSS winds of all sorts, for instance, you know).

It’s just that the reason why this project uniquely surfaced different hackernews-like headlines recently is this “Apps lighter than a React button” clickbait. Which, you know, to be quite fair, is not fair. But if it helps the guys to get their open-source traction, I think it will count for good. Of course, after they thank React for that.

Meanwhile, other metaframeworks didn’t make any significant announcements of that sort the last week. Of course, Astro (abandoned by its aficionado Ben Holmes in favor of some boring terminal app gig) delivered some regular goodies with v5.6 and summarized another cool month of the ecosystem growth (this project was new to me too BTW — quite nice for Astro fans). React Router went on with their new experiments and researches on making the best way of building React apps in v7.5 (some experiments bring breaking changes, be cautious!), along with competing TanStack bringing new versatile updates for their “one stack to rule them all” in v1.115. Analog deliberately goes to v2 through another v1’s minor, with another Angular’s alumni Quick City following their steps on the same way with v1.13. Humble Aral Roca singlehandedly delivered another patch v0.2.10 (which in fact is minor version too in the context of pre-v1 flow) for Brisa, complementing this list of hardworking frameworks authors delivering less invisible but highly efficient (and anticipated) updates on their way to major progress.

All in all, everything inevitably moves forward, in more narrow or wider extent depending on project ambitions, sizes, and development stages. This week’s list looks much nicer, more positive than the previous scary one, which is good by itself, even if there’s some pinch of sadness and nostalgia in the topics we discussed. But eventually, everyone finds their unique focus in this ocean of FOMO, and I truly hope I have helped you a bit with that today. Until next time.

👋