Is it really worse now though? As a full-stack developer since 2013 (Django on backend since the start, React on frontend since 2015), I think it's finally getting better in the past couple of years, and I think it was worst between 2016-2019...

After years of not knowing what to use because the libraries I've used 2 years ago are no longer maintained or their APIs changed 3 times since then, there's now Next.JS which seems like a well supported, batteries included, opinionated framework with good documentation...

with vite and esbuild on the rise, the days of fiddling with webpack and other complicated build configurations may soon be behind us...,

typescript vs. flow seems to have ended with typescript being the clear winner and having great support in most libraries, frameworks and IDEs... (although I'm a bit scared that the JS native type annotations proposal may again fragment the typing world here...)

browser-side APIs are no longer evolving so rapidly, IE11 & EdgeHTML are dead and there aren't that many features/bugs specific to Firefox/Chrome/Safari anymore...,

Does Nextjs make you use Django less? I think my basic process going forward is going to be Nextjs + something like Supabase as default and only add a more complex backend as needed.

At work we don't have enough manpower*time to convert our React spaghettis to Next.js, but I'm starting to experiment with Next.js & prisma & postgres aside and plan to use it for my next hobby project which I'm never gonna finish just as the rest of them..., If I'm gonna like the approach and not miss some of Django's features right from the start I'm gonna think about letting Django in the past...

Supabase could be enough for a simpler project where you know you won't need any advanced features.., I wouldn't just start with it if I knew the project's gonna get huge in the future, but I did a couple of smaller projects with Firebase alone..

> The situation is even worse now.

Is it really? I have the exact opposite opinion. I mean, I feel like the industry has pretty well standardized on React in Typescript for the front-end on web apps. Sure, there are other technologies that do different things (e.g. Svelte, and someone else mentioned Phoenix LiveView), but for the standard "I'm building a CRUD-focused web app", there are simple choices to make and it's easy to "do the right thing". I contrast this with 2016, when things were still in flux so it was much easier to make what turned out, in hindsight, to be the "wrong" choice (e.g. Angular or Flow). Plus, tooling support is much better now.

React is used a lot, but it is certainly not the standard. Angular is still used extensively, especially in enterprise

I'd still argue that React is very much the default for new applications where legacy interop or existing team familiarity isn't a factor.

I mean, this is obviously an extreme example, but COBOL is also still used extensively in the enterprise, yes it's still not really used for any new projects.

I'd say you are still wrong. Anecdotal evidence: I have recently started on a new green field project in a large enterprise and it was Angular.

React is just utter trash for organisations where maintainability is a concern.

And you are still wrong. Bigger companies often pick Angular.

I would say things are improving. Through WASM we are now getting more languages running in the browser. They are starting to bring alternative frameworks and ways of doing things to the table as well. Instead of this frontend / backend notion, we should start thinking again about the notion of networked applications; like we used to do before the web became a thing.

Inevitably you want some computation to be close to the source data (for efficiency) and some other computation to be close to the point of interaction (for responsiveness). It's not an either / or proposition. You can do both. And phones and browsers can do a lot locally these days. So there's no need to pretend that it is still 1999 in terms of browser capabilities. They can do so much more now.

Instead of AJAX, we now have companies like Tailscale doing all sorts of funky networking stuff in a browser. Likewise, people are running entire 3D games, photo and video editing tools, or design tools like figma, etc. in a browser. All enabled by WASM. Most of that stuff does not involve a whole lot of css, javascript, or html. That stuff is increasingly optional. Browser application development and desktop application development are finally merging after being considered completely separate things for more than 2 decades. It's all just application development. It may or may not involve talking to servers via a network. You don't have to limit yourself to HTTP when doing that.

How is WASM different from what we had with Java applets? (haven’t looked deeply into WASM yet, so seriously asking)

At this point web development has become so complex and fast-changing that by the time you're done explaining how everything works the whole world has moved on to the next generation of complexity. We're almost at the point where web development is becoming "unknowable"; like knowing the position of an electron... You can know where it was at some point (maybe) but really the best you can do is estimate its rough position in the cloud.