Post

What do you predict will happen with web development in 2025? What will pop? What will stop? What Frameworks, tools, and libraries will become popular? What will we fall out of love with? Will AI take our jobs in 12 months? Give me all your takes for the yearly @syntax.fm podcast

7 reposts105 likes

- @deno.land will begin to gain real traction as teams start to understand and make use of their workspaces implementation - Angular will continue to rapidly evolve and have a bit of a renaissance - Tailwind will remain king as v4 is released - Vitest will become the predominant testing tool

I'm here for the Angular renaissance. Ever since the switch to v2 they've been making steady well thought-out progress. I've kept my large corporate project up to date over the last few years and each update took me less than a day. There's a place for Angular in the ever-changing fragile FE space.

As a closed source framework maintainer, I just don't see deno gaining much ground. There's too few upsides and a whole lot of downsides to the argument. What if Deno Inc closes shop? I now have a closed source runtime for which I won't receive security updates. It's just too risky.

Fair point! I’d be interested to see how they approach this (imo quite valid) concern. I’ve been thoroughly impressed with it since v2 came out though, so I’m selfishly hoping they have thought about this.

It’s going to be the year of make or break for RSCs. Either they can make the model friendly and grokable enough to be picked up, or devs will decide that the marginal benefits of preventing waterfalls is not worth the churn and use simpler naive approaches like loaders or tanstack query/start

I’ve been building something on RSCs for the first time. It’s a bit weird and I don’t like things like redirects being done by throwing errors but there’s a lot of good stuff in it.

I don't think much is going to change in a year when it comes to RSCs. My guess is that Next will continue to be a popular framework and more Next projects will be using App Router. Also, other frameworks will adopt RSCs and make it available even for SPAs. They aren't going anywhere.

I could see Nitro having a big year. Nitro 3 is incoming and I’m not sure if folks realize how much it’s powering: Nuxt, Solid Start, Tanstack Start, Analog.

Hmm, I’m not convinced it will get a lot of attention. I’m a big fan of Angular but the framework hasn’t grown with other frameworks at the speed it should have. I mean, it still doesn’t support HMR. We moved our code over to Next but now are looking very closely at Svelte.

The quality of swipe-driven UI patterns implementations, like bottom sheets, sidebars and full pages, will finally come close to that of native implementations, and will start to be widely adopted on the web.

Simpler frameworks like tanstack start will get more traction and nextjs will start decelerating. Local-first will get more buzz and more startups will offer services for it, but won’t get the expected traction. React 19 and the React Compiler will be generally available. Big milestone.

AI tooling will boom. Server-first frameworks like Next/SvelteKit/Astro keep rising. Tanstack Start leaves beta. Bun challenges Node. Big tech shrinks teams as each dev+AI builds more, but small companies & startups multiply - more jobs overall, just distributed differently. 🚀

I'll be poppin on: Windsurf IDE, Astro and NPM Monorepos, Go & Sqlite micro backends, CSS 4 print, Frameworks that are easy to ingest for AI will prevail. Stoppin on: Wordpress, WebDevelopment, Facebook is still not dead.

People will start using AI like Jarvis A simultaneous rise in layoffs and rise and startups do to personal productivity Vercel is will add a couple more 3 letter acronyms to their repertoire Local First will become a legitimate way of building software

DB as a service will start providing and acquiring sync services Hypermedia approaches will get more popular for traditional server client models

I think AI making us more productive will have an impact, it will leave a lot of time for developers to improve and focus on other things they thought they'd never focus on before, and in light of that newly gained time from AI assistance, many new great things will happen. :) Still not sure what

I feel extremely strongly that ai is making more possible than before, and hence it will lead to more human work not less. Things that weren't possible before are possible now. This means restructuring though. As in jobs lost some places and jobs created other places.

- RSC will be put to the test with other frameworks implementing it. - Astro and Svelte will grow. - AI bubble will burst with the next openai model. It will be a usefulish tool but the inherent randomness models is a roadstop for this generation.

Some of my favourites from its toolset are definitely being runtime agnostic, same code running on vercel, deno, cf or good old node, and the amazing tools coming from the unjs ecosystem: websocket, storage, database, session management, data conversion, dynamic configurations and so on

Rspack will gain traction. Let's be honest, most projects are still using webpack because migrating to something else is way too much effort. Rspack is a simple migration to get huge perf boost

I just migrated our app off of webpack to rspack. Doesn't play well with HMR, and the Typescript/ESLint with caching (so I had to disable caching in dev, which is still good enough to be worth the upgrade), but rspack has been a big win for little friction.

- Better routing for RSC: better TS DX, client-based routing fragments, easier revalidation - Debate of going full on complex selectors, long list of classes, and css vars (TW v4), vs compiled and simpler css (StyleX) - JS frameworks understanding that having only one loader per page doesn't scale 🙏

That last point is where I’m at as well, but no one wants to waterfall the requests or figure out how to hoist the queries. Have you seen much in this are aside from RSC?

Found only this unimplemented idea for Remix github.com/remix-run/re... The proposed APIs doesn't seem great tho. I wish there was something like: createRoute({ slots: { right: Feed }, component: () => { <Route.Outlet slot="right" /> } }) Feed = createSlot({ loader: ... })

github.comSibling Routes (aka Parallel Routes or Aggregated Routes) · remix-run remix · Discussion #5431Proposal Add sibling routes support to Remix and React Router to allow matching multiple routes in the same segment but being independent chunks and with a different data URL. Background The propos...

Most gainfully employed developers I know are all writing Rails and Laravel and boy do they love Inertiajs. I think there will be a big push towards classic server-side web technologies and something like Inertia connecting a thinner client technology.

-Pop: No Code + Design+ AI -Stop: monetized Boilerplate , 5$ VPS, paywalled blogs or content sites (medium), Quora, overemployment -Tools/Frameworks/Libs: RN, Vite, Tanstack Start, Remix/RR, Angular (!), Astro, Nuxt (!) Falling out love with: The edge, X, Dark themes,Wordpress, AI code editors

As i've personally spent the last 5 years thinking next js was a mistake, I'm hoping for a slight shift back to SPAs and REST. Dagnabbit

I’m actually interested in seeing in what comes out of different React routers. Between the Remix -> React Router 7 migration, TanStack router, and Next, I’m really curious as what will come out of this space

It's been said already, but I think Next.js will continue to decline in popularity. It's far too complicated, and achieves the same thing as much simpler frameworks. It's also too closely aligned with Vercel, giving the feeling of vendor lock in, even when there isn't.

I don't think it's a software advancement, but more an excuse by companies to pay less. "We've had a lot of trouble hiring a senior dev for $$$$. Can we just hire a junior dev and ask them to use chat GPT for $$?"

Companies that think like that aren't going to be very successful so I'm not too worried about that impacting pay generally

Server rendered HTML will continue to regain popularity as more people start to rediscover the simplicity of progressive enhancement and unobtrusive libraries like htmx. Not every site needs a heavy framework!

Tanstack Start is gonna be big I think. More disgruntled JS devs will explore Laravel and Rails. Here's my 🌶️ one: no significant forward steps for LLMs in 2025, we've reached the iPhone 4 era

To counter the Laravel and Rails thing, I wonder if there’s any chance of a batteries included next/tanstack/whatever framework/assembly of other frameworks as opposed to “starter kits”

Laravel has starter kits for Vue/React/etc but I guess you mean more like a higher level of abstraction? A single framework you can install that has a folder for frontend, a folder for backend (whether that be PHP, Python, C#, whatever), and no boilerplate to deal with.

I mean a react framework with all the stuff you get with laravel. Like I can just do config.db=postgres, jobs.persistence=database or whatever you do in laravel world that means you can get on with building stuff instead of choosing a library.

Maybe but it's taken years and years to get Rails and Laravel into the shape they're in. It's not something someone can just cook up in a year. I'm also not sure there's as much demand for batteries-included in the JS world. People still seem to value flexibility.

Yup I can totally see that. I still think their use is limited by the reliability issues but there are a ton of interesting uses that haven't been sussed out yet

As the JS frameworks are moving to the server, I think people will start looking to the established server technologies (go, .net, etc) for a better experience. Also web components (I hope)

Pop: - the Vite ecosystem and tools betting on it (Astro, Tanstack, Storybook, Remix, etc.) - CSS is good now, actually Stop: - Wordpress will start having a hard time given this year’s debacle Fatigue: - Vercel/Next/RSC seems like it’s starting to be met with some skepticism

Basically at Matt Mullenweg (Wordpress founder) er i juridisk tvist med WP Engine (stor WP hosting provider og plugin) og afskåret adgang til nogle ressourcer og overtaget WP Engines kode (som jeg forstår det). En masse WP ansatte sagde op.

-Tanstack Router/Start will get much much more popular -Pendulum will swing towards SPAs from SSR complexity -Local-first will stay niche -React 19 will hopefully ship 🤞 -Tailwind will continue to dominate but nextgen css-in-js will catch up -Use-the-platform folks will continue to be angry

- move towards compiler-based frameworks and signal-based reactivity continues - tailwind remains dominant (esp post v4) - astro and svelte keep gaining ground on react/next - big push on smaller dependency graphs and universal ESM adoption - move continues to platforms like cloudflare workers...

- sass and other pre-processors keep losing favor - deno gains adoption but remains very niche - 327 new web frameworks and 41 new build tools are launched in 2024

Hope/prediction Framework with a cleaner approach will catch up and take some of Next.js share Such as Sveltekit, Tanstack Start, react router v7, …maybe Solid (start) Am I the only one fed up of Next.js?