BTW I think some anti-Rust people are more annoying than the worst Rust evangelists - seen some of them calling people not using Rust as “murderers”, because “memory leakage can kill at the right time” - but that’s due to them being evangelists to right-wing politics.
Imagine thinking the borrow checker is just “worse GC.” Please at least learn about a language before you try to spread crappy memes about it.
What? You list metaprogramming as a plus? And functional programming as a minus?
You should rewrite this meme in rust.
Did I change a nerve’s internal state instead of returning
nextnervestate?Rust is all about changing internal state. It even has structures (Cell and others) that specifically allow you to change internal state in otherwise immutable data structures. You should probably learn rust properly before making such claims about it.
Every foreign language is unreadable until it is not foreign. Acquiring programmer socks helps, too.
Changeable state in an immutable data type? Is… Is that a good idea?
I don’t know Rust, but I can tell that the criticism in these image is mostly bullshit. But changing immutable state sounds like a bad idea to me. Is there something I’m missing?
Rust is often treated like it has a split between mutable and immutable state, but it’s really a split between unique and shared state. Shared state can be mutated if certain invariants are held, which types that provide “interior mutability” as its called enforce.
https://doc.rust-lang.org/reference/interior-mutability.html
"We spent two decades making sure the compiler will never produce memory-unsafe code. It requires a lot of nigh-illegible boilerplate code to even compile and adds massive cognitive load, but the effort will be worth it.
Anyway, here’s a type whose only purpose is to disable all of that shit."
Cell doesn’t disable memory safety, though? It comes with additional restrictions such as being unable to share between threads (statically enforced) obviously, it’s not an unsafe feature. You also can’t read from it unless your type can be bitwise copied, etc.
Interior mutability is mostly for making immutable interfaces that for some reason or another benefit from storing a bit of mutable state, such as for lazy evaluation. It’s also used for cross thread communication in some cases since you have to use shared (immutable) references to share things between threads.
It requires a lot of nigh-illegible boilerplate code to even compile
this is the entire source code for an app that performs a rather complex function, note the absence of boilerplate
from https://codeberg.org/Mycellf/wordjoin
use std::{ fs::File, io::{self, BufRead, BufReader, Read, Write}, path::PathBuf, }; use clap::{Parser, ValueEnum}; use icu_segmenter::{ LineSegmenter, LineSegmenterBorrowed, options::{LineBreakOptions, LineBreakStrictness, LineBreakWordOption}, }; /// Insert utf-8 word joiners (U+2060) to prevent text from wrapping outside of whitespace characters. #[derive(Parser)] struct Args { /// Format and concatenate each file in stead of stdin file_paths: Vec<PathBuf>, /// Read stdin by line instead of all at once /// (sometimes worse for interactive use; files are always read by line) #[clap(short = 'l', long)] by_line: bool, /// See https://drafts.csswg.org/css-text-3/#line-break-property #[clap(short, long, default_value = "strict")] strictness: LineBreakStrictnessValues, /// See https://drafts.csswg.org/css-text-3/#word-break-property #[clap(short, long, default_value = "normal")] word_option: LineBreakWordOptionValues, /// Print this app's GNU GPL-3.0 license #[arg(short = 'L', long)] license: bool, /// Print this app's source code #[arg(short = 'S', long)] source: bool, } const WORD_JOINER: char = '\u{2060}'; fn main() -> io::Result<()> { let args = Args::parse(); if args.source || args.license { if args.source { println!( "Cargo.toml:\n{config}\n\nsrc/main.rs:\n{source}\n\nREADME.md:\n{readme}", config = include_str!("../Cargo.toml"), source = include_str!("main.rs"), readme = include_str!("../README.md"), ); } if args.license { println!("LICENSE:\n{}", include_str!("../LICENSE")); } return Ok(()); } let mut options = LineBreakOptions::default(); options.strictness = Some(args.strictness.into()); options.word_option = Some(args.word_option.into()); let segmenter = LineSegmenter::new_auto(options); let mut stdout = io::stdout().lock(); if args.file_paths.is_empty() { let stdin = io::stdin(); if args.by_line { join_text_by_lines(stdout, stdin, segmenter)?; } else { join_text_by_all(stdout, stdin, segmenter)?; } } else { for path in args.file_paths { let file = File::open(path)?; join_text_by_lines(&mut stdout, file, segmenter)?; } } Ok(()) } fn join_text_by_lines( mut writer: impl Write, text: impl Read, segmenter: LineSegmenterBorrowed, ) -> io::Result<()> { for line in BufReader::new(text).lines() { let line = line?; join_text(&mut writer, &line, segmenter)?; writeln!(&mut writer)?; } Ok(()) } fn join_text_by_all( writer: impl Write, mut text: impl Read, segmenter: LineSegmenterBorrowed, ) -> io::Result<()> { let mut input = String::new(); text.read_to_string(&mut input)?; join_text(writer, &input, segmenter)?; Ok(()) } fn join_text( mut writer: impl Write, text: &str, segmenter: LineSegmenterBorrowed, ) -> io::Result<()> { let mut segments = segmenter.segment_str(text).peekable(); while let (Some(start), Some(&end)) = (segments.next(), segments.peek()) { let segment = &text[start..end]; write!(writer, "{segment}")?; if end < text.len() && segment .chars() .next_back() .is_some_and(|end| !end.is_whitespace()) { write!(writer, "{WORD_JOINER}")?; } } Ok(()) } #[derive(Clone, ValueEnum)] enum LineBreakStrictnessValues { Loose, Normal, Strict, Anywhere, } impl From<LineBreakStrictnessValues> for LineBreakStrictness { fn from(value: LineBreakStrictnessValues) -> Self { match value { LineBreakStrictnessValues::Loose => LineBreakStrictness::Loose, LineBreakStrictnessValues::Normal => LineBreakStrictness::Normal, LineBreakStrictnessValues::Strict => LineBreakStrictness::Strict, LineBreakStrictnessValues::Anywhere => LineBreakStrictness::Anywhere, } } } #[derive(Clone, ValueEnum)] enum LineBreakWordOptionValues { Normal, BreakAll, KeepAll, } impl From<LineBreakWordOptionValues> for LineBreakWordOption { fn from(value: LineBreakWordOptionValues) -> Self { match value { LineBreakWordOptionValues::Normal => LineBreakWordOption::Normal, LineBreakWordOptionValues::BreakAll => LineBreakWordOption::BreakAll, LineBreakWordOptionValues::KeepAll => LineBreakWordOption::KeepAll, } } }
I have a guess that it was added as a workaround the moment Rust was marketed as a general C++ replacement, and as people started to realize overusing the FP paradigm also has its downsides, it’s not unique to OOP.
Pure OOP without first class functions sucks fucking ass.
Yes, but wat kind of programming socks? Thigh highs, with white, pink and blue stripes?
thinks functional programming = bad

“Don’t look a gift horse in the mouth”
No, you probably should look into that horse’s mouth to make sure there are no Greeks hiding in it.
“Beware Greeks bearing gifts”
Again, no. You should be wary of gifts bearing Greeks.
Tbh the borrow checker isn’t a problem for 75% of cases. If you actually need the performance/memory optimization then yes you will have to deal with it… Otherwise just
.clone()And if you find the borrow checker annoying in async rust, that’s mostly a tokio issue. Look into smol-rs as it offers alternatives
If you want real cons…
- Compile times
- easy build time arbitrary code execution
- trait bounds spaghetti
- viral async
- viral lifetime annotations
🫣
Viral async?
I’ve done quite a bit of async programming and I can’t quite figure out what people are complaining about here. Best I can tell, they just don’t understand what async functions actually are.
Most of the time, async tutorial makes you learn tokio, not async. If your program can run with only tokio::main, then you learned async. If not, you learnt tokio (except if you are spawning a future that should never stop)
For example, my pet project only uses tokio::main to do async stuff. The only instances of tokio::spawn is make sure some SQLite transactions get polled to completion. I do need to replace them with a proper mechanism now that sqlx supports smol-rs
When people say “async is viral” in Rust, they mean that once you make one function
async, that change tends to ripple through the rest of your code. Any function that calls it usually has to becomeasyncas well so it canawaitthe result. In turn, the callers of those functions often need to becomeasynctoo.This propagation can continue all the way up the call stack until you reach your application’s entry point. The main exception is when you introduce an explicit synchronous-to-asynchronous boundary, such as by using
block_on, which drives the future to completion without requiring the caller itself to beasync.Yeah but it’s not really a problem with rust but how the language pattern is made. It’s the same in JavaScript/typescript, and Python IIRC
It is Rust implementing the pattern. So… :P
along with most modern languages… it’s the way we deal with async when you don’t want callback hell. it’s just a complex problem domain
like… what… JITs are complex so that’s a problem for V8 specifically?
And if you find the borrow checker annoying in async rust, that’s mostly a tokio issue. Look into smol-rs as it offers alternatives
This is great until you want to use a library which is tokio exclusive, which is most of them.
Well you just need that crate :3
Huh, that pretty cool actually. I need to play around and see if this works with gtk-rs, channels get fairly annoying if you need to use them a lot.
Keep in mind that it doesn’t remove tokio from the stack tho. Don’t use this to try to improve compilation time
I know, I read the description. It just looks like a nicer syntax around setting up a tokio runtime and sending code between runtimes. It’d still be nice to have a non-tokio options so stuff could be single threaded.
A lot of the time it’s not about options. It’s about not messing up the async pattern.
If you have something that either:
- requires a lot of CPU time
- requires to run permanently, independently to the caller’s future polling. Then you can spawn it on a global tokio executor.
If not, just use future polling tricks like the
futures::join!()macro or a stream with.buffered(). It won’t be slower. The bottle neck is IO. Not the program.Personally I even try to replace the heavy
reqwestlibrary withureq+blocking, and it works perfectly and compiles faster (you can see that in theapi_bindiumcrate)
A forbidden compiler
Compile time borrow checker is equal to runtime GC, right…
being annoying online
hmm








