About Rust in Webdev

4 points by EugeneOZ ↗ HN
The best programming language right now is Rust, but we need something better. Readability is worse and worse, damned “?” was just the first bullet. Still zero progress on custom lifetimes, no efforts to improve readability. Wasm is losing dev's interest, because the learning curve is steep and the benefits are not so obvious. It's almost impossible to find a job in webdev, requiring knowledge of Rust - language becomes more and more “system”-focused, more and more “enthusiastic”. Most of the libraries are “forever 0.x”.

The Rust-for-web ecosystem is sporadic. Sometimes you can find what you need, and in 99% of cases it will have version 0.x. “We use beta, because it's better than nothing” - it’s 100% about Rust right now. I wish I could say something better.

Looks like webdevs are not interested in wonderful languages, they are just looking for the shortest and the most simple solution they can find. It's depressing, but it's true.

Without any doubt, I sincerely think that the best language for API is still Rust. But what is our choice of frameworks? Forever 0.x Hyper, forever 0.x Rocket, that's all. Please don't even start with Actix - I hope this zombie will finally die soon. My only hope is that The Rust Foundation will acquire enough attention from bug tech to change the trend and improve Rust positions in webdev.

11 comments

[ 5.0 ms ] story [ 38.0 ms ] thread
"... they are just looking for the shortest and the most simple solution they can find"

Why is this problematic?

Humans should have ambitions to create something better than “it works”.
Simplicity is a high value in engineering.
And still, we have more than BASIC for a reason.
I completely reject the idea of “the best programming language”. It’s just a tool for a specific job. That’s all there is to it.

It’s not going to solve every problem and it never will.

What problem do you have with actix-web?
> Looks like webdevs are not interested in wonderful languages,

I'd guesstimate 99.9% of web developers have no experience with native non-GC development. Asking them to embrace non-GC is a big ask when GC is not a performance bottleneck for common cases.

Save Rust for serious apps by elite teams. You don't need Rust for average Joe web services that do little more than ferry data around.

You don't need it, but plenty of devs prefer it. Our shop switched from Python to Rust. We almost went Go, but in general the language didn't feel like it was solving enough problems over Python.

Rust gives us a nice type system and performance, in an enjoyable package.

We certainly don't need it (well, maybe in a single service), but it's our choice. The decision was largely Go vs Rust, but again, Go just didn't offer enough to us.

I agree though, a GC doesn't prevent us. Ours was a language choice, not a performance/GC choice.

Catching a bunch of bugs at compile time without having to write tests for them is real nice
As a programmer battling a C++ dependency, being able to write tests without having to bring in a 3rd-party testing framework is real nice too.