Frontend development, 2022 baffles me. I needed a "landing page theme" (open source).
What I need: index.html, style.css.
What I find: Yarn, Vue, Babel, Sass, tailwind, react, nextjs, webpacker, wasm, vite, pug, typescript.
And a very few "html+css".
What happened to KISS? It seems frontend-crowd is solving problems by stacking stuff on top of stuff, which then causes problems. Which then are solved by stacking more on top.
A single page is still just an effing single page.
@berkes Most frontend devs building these things should have been rocket engineers, astronauts, brain surgeons, etc.
@simeon I suspect that quite some started off as "I need a fun project to test [this new tech everyone is raving about]". Followed by "why not open source it", and "why is it getting so much stars and attention".
And not as a well-thought-out problem-solution fit.
@berkes
@simeon I would love to take you to coffee some day and explain the whole sordid mess, but the short version is almost entirely the opposite of what you suggest. What happened was that some very smart people came up with a very specific solution to a very specific problem, released that solutions to the web, and the idiots on the web decided that these are the solutions to EVERYTHING.
React was a solution for Facebook internet explorer problems. Angularjs was a solution for consistency at Google. Angular2 was a solution for Google because angularjs was kind of lousy and everyone would rather write react. Vue was a solution for the fact that UX specialists always hated react. And developers responded with enthusiasm because they'd rather write this stuff in pure JavaScript than learn the polyglot, multiparadigm browser.
The problems, fundamentally, are that html is not composeable, css is contemptable, state is hard, and the internet grew too fast for its own good.
> fundamentally, are that html is not composeable, css is contemptable
I agree wholeheartedly. I've always said, every since the web2.0 craze started, that the web is fundamentally unfit to run (desktop-)applications.
Now, I have been proven wrong by the myriad of applications that I use on a daily base, from the mind-blowing figma (or penpot!) to a more boring RSS reader.
But, I'd say those came *despite* the web being fundamentally unfit for development of such apps.
More practically. Web (HTTP/HTML) was meant for an (interactive) document system: a way to distribute, explore, read, publish and discover data and information.
The discoverability part is *so* strong, and so powerful, that ever since, people have been trying to build rich applications in this document system.
Just like it's possible to build large desktop apps in excel, it's neither easy nor a good idea; generally. Same with Web. Yet here we are. Without alternatives.
@berkes @braindouche What is the alternative to the web for building and running applications?
Is there a turn we missed that would've gotten us to a better place?
I'm so entrenched it the "web way of things" that it's difficult to think about alternatives.
@simeon @braindouche I don't think there is today. But there could be.
Web apps, I think, are the norm because of discoverability (app stores somewhat cover this, but never as good as web) sandboxing guarantees (we know it'll be safe) and standardized runtime env.
Maybe WASM+canvas will fix some? Idk. Maybe the app store idea must be expanded and opened? Maybe we need better sandboxing (Docker, snap, and iOS, Android alike?).
@braindouche @berkes Good points!
I also remember the hype about Angular's two-way binding. It seemed very magical at the time but seeing where it took us puts it in a different light.
Luckily there seems to be a return simplicity and HTML-over-the-wire as pushed by https://hotwired.dev and https://htmx.org.