People who don't like #Rust: why specifically don't you like it?
I'm in the process of learning it now. There are definitely some things about the language that I can see some as finding irritating (i.e.: the borrowing system). Personally though, I'd rather have a dozen complie-time errors than a single runtime error. This is the reason I tend to gravitate towards Haskell, for instance.
It's certainly not the right language for everything, but if you want better safety in code that needs to be highly efficient, it seems a reasonable alternative to C/C++.
reshared this
Jonathan Lamothe
in reply to Jonathan Lamothe • •Jonathan Lamothe
in reply to Jonathan Lamothe • •Found my first point of irritation. Crates.io requires a GitHub account.
I have one, but I don't like to use it.
That said, creating a crates.io account seems optional-ish...
glyn
in reply to Jonathan Lamothe • • •Non-Github account creation · Issue #326 · rust-lang/crates.io
GitHubJonathan Lamothe likes this.
Jonathan Lamothe
in reply to glyn • •Mo :ferris: :tux:
in reply to Jonathan Lamothe • • •We are in the same boat. I don't like the current state of having a single authentication option with Github.
But the good thing is that Github is only required for authentication. The code of all my published crates on crates.io is hosted on Codeberg 😁
They are open to contributions to implement authentication via email, but sadly it is not a priority.
Jonathan Lamothe likes this.
soc
in reply to Jonathan Lamothe • • •I kinda like Rust, but the community is highly irritating:
Incredible amounts of zealotry, usually paired with a poor understanding of the subject matter, the design space, and possible alternatives.
Discussing technical topics with Rust people is simply not a good use of time.
Jonathan Lamothe likes this.
Jonathan Lamothe
in reply to soc • •@soc Ah. I've had little to no interaction with the community at this point. My evaluation thus far has been on its perceived technical merits alone.
That said, I don't know enough about it yet to have a well-informed opinion on even that yet.
Edit: To clarify, community is important, and having to deal with a toxic one would be a deal-breaker.
Edit 2: typo
randygalbraith
in reply to Jonathan Lamothe • • •Digital Mark λ ☕️ 🕹 Z?
in reply to Jonathan Lamothe • • •As a #Scheme hacker, #Rust has nothing I need:
Exploratory, interactive programming, with a REPL.
Dynamic types, I can do an (assert (Foo? x)) if needed, but having to write Foo x, or Foo<T:Bar> x, everywhere sucks.
My errors are never caught by strict typing or borrow checking. I make much higher-level logic errors.
Garbage collection or ARC equivalent is the only way to safely manage memory. STOP manually doing it. Even in C, you can use Boehm GC!
Scheme compiles to fast binaries.
screwlisp reshared this.
Digital Mark λ ☕️ 🕹 Z?
in reply to Digital Mark λ ☕️ 🕹 Z? • • •robinm
in reply to Digital Mark λ ☕️ 🕹 Z? • • •GitHub - evcxr/evcxr
GitHubJonathan Lamothe
in reply to Digital Mark λ ☕️ 🕹 Z? • •@Digital Mark λ ☕️ 🕹 Z? It's all about what you're looking for in a language. I don't have much Rust experience yet, but I can tell you that Haskell's type system has saved me from making errors on multiple occasions (most frequently when a function can return some sort of error (including null (or equivalent)) and I forget to check for it).
At the end of the day, if Scheme works for you, then that's the language you should use. That's totally valid.
Edit: paren balancing
Reid
in reply to Digital Mark λ ☕️ 🕹 Z? • • •Mo :ferris: :tux:
in reply to Jonathan Lamothe • • •If you gravitate towards Haskell, Rust is definitely for you.
I really enjoyed learning functional concepts in Haskell last semester after knowing Rust. There are many similarities, but (sorry Haskell fans) Rust is practical ;)
About borrowing being irritating: This is normal at the beginning, but the theory is well explained in the official Rust book.
BTW, could you please use the RustLang tag? 🥰
Background: https://fosstodon.org/@mo8it/112056453394255413
Mo :ferris: :tux: (@mo8it@fosstodon.org)
FosstodonJonathan Lamothe
in reply to Mo :ferris: :tux: • •@Mo :ferris: :tux: Oh, I understand perfectly well why it exists. It makes the kind of checks I had to do myself back in my C/C++ days. The difference is, as a human, I can fail to look for these things.
I actually appreciate the borrow checker. It gives me similar safety guarantees to Haskell without the runtime costs. I just also understand why some would find it irritating.
screwlisp
in reply to Jonathan Lamothe • • •Digital Mark λ ☕️ 🕹 Z?
Unknown parent • • •Probably, you'd have to be careful with PG's free rules, but Boehm can be per library.
But you know what's really great? Whataboutism by randos who don't care about the conversation, they're just RETF 🦞
(I should note, I'm not *saying* that's you, just… implying it from the weird edge case.)
Digital Mark λ ☕️ 🕹 Z?
Unknown parent • • •Reid
in reply to Jonathan Lamothe • • •the borrow checker is such a big part of the language it's not just slightly irritating, it's like having a non-consentual finger up the ass every time you open some Rust code in your editor
And the fact Rust is always staticallly linked and lacks any sort of reproducible builds don't help, even the compiler itself only compiles with an n-2 version of the compiler, if you skip updating the compiler for a while and want or have to keep using sources then have fun compiling every version since you last updated the compiler
Its type system is also like a borrow checker: non-consentual fist up the ass, want to add an u8 to an u32? Nope, can't, have to manually cast everything because that's why we do programming languages instead of writing Assembly, to do all the fucking busy work ourselves
Oh, and Cargo is its own can of rotten worms
LisPi likes this.
LisPi
in reply to Reid • • •@Reiddragon > And the fact Rust is always staticallly linked and lacks any sort of reproducible builds don't help
That is excusable in languages where source-only distribution is normal and expected. (Indeed, compilation should be a transparent caching step and artifacts of such shouldn't be commonly shared.)
That is not the case for Rust.
> even the compiler itself only compiles with an n-2 version of the compiler
That's also a problem, Rust's bootstrap story sucks.
Ada's might suck as much, I'm not sure, I have found a few interpreters when I last looked...
> I'd rather have a dozen complie-time errors than a single runtime error. This is the reason I tend to gravitate towards Haskell, for instance.
There should be no meaningful difference between runtime and dev-time for the majority of devs. Dead languages aren't necessary. And punchcard retrocompatibility can be preserved without prioritizing a development process that is optimized for that workflow.
As for typing static vs dynamic, there's a thing called "gradual typing", and it is very possible to tie the type-checker into a REPL.
mhd
in reply to Jonathan Lamothe • • •I've got a few things with Rust that make me dislike it a bit -- note that that doesn't mean that I think it's generally a bad language. (They're all good langs, Brent.) But here we go:
Another single implementation standard-less language. No solid standard library, everything done by downloading the internet. Very un-Turbo-Pascal-ish compile times and memory usage. Annotations. BCPL-ish syntax with too much line noise (and hey, I used to program Perl). Tied a lot to the worst things in IT (browser engines, crypto). Fanatical community. Overly complex async to save me from writing threads. Both functional purists and micro-optimizers. The word "Rustaceans" alone.
It's not a language that I'd like to use recreationally, but I wouldn't quit jobs if I have to do more work with it at work.
Jonathan Lamothe likes this.