Where will we all be(e) next year?

Chris Remington · 7 months ago

Where will we all be(e) next year?

@maegul · 7 months ago

Because activitypub is a hot mess and Lemmy’s implementation on top of it, is problematic.

I believe you, but that doesn’t quite answer whether it’s worth it or not, software often being a hot mess to some extent anyway (said by someone who’s never developed against AP/Lemmy)
I’m genuinely interested in your critiques … do you have any prior posts of your or someone else’s that you could point me to … or a rant you’re happy to drop here?

PenguinCoder · edit-2 7 months ago

EDITL Lemmy-UI deleted my !@#$#% links because I used the standard Markdown of less than and greater than symbols for a link.

1.)

2.)

My comment on a deleted post

These are my opinions, probably not shared with others and surely not the end all be all to ‘why not Rust?’

Rust is hard. Rust is slow to iterate with and compile. Here’s a highlevel overview of the things you’d need to learn to effectively understand Rust code; not even speaking to learning enough of that to write it. Rust gets touted as secure, and it is more secure than other low level languages like C/C++. That does not make it invulnerable to exploits. It gives a false sense of security to developers who think My app can’t be hacked, it’s written in Rust!!. While Rust is better at memory management and protecting against run time errors related to memory issues, that doesn’t make it 100% safe. That removes almost one class of potential exploits, not all of them, and can’t protect the developer against other developer created issues. Rust code is still written by people. People still write insecure code even in Rust. Computers are dumb as hell; but fast. People are smart as hell, but SLOW. Combine the two and you get stupid things faster; not smarter things better.

Rust development is hard, hard to learn, very hard to do right
Rust is not suited for a web application. Would you tell someone to go write the web application and web page in C/C++? Nope. There’s a reason for that. Not suited to it. Square peg, round hole
There’s always exploits being discovered. Rust handles some things better, but uhhh… Go look at some Lemmy Rust code. Definitely still has vulnerabilities; Rust won’t save you from yourself.

Something like Golang is much better choice for development of a web service that has sane API handling. By the time to add in more error handling, more type checking, more passing around of this function or that data, and more checking it on the sender and receiver side…etc. By the time you’re writing Rust safely and correctly; it is much slower than what you may be lead to believe.

Go is primarily designed for speed of development (including compilation times), rather than speed of execution. Go programmers tend to value clear code over fast code. Why does a microsecond longer matter for a website? Maybe in some backend PLC land there’s a damn good use for that performance. For a networked web application; it’s a pointless metric. That is not going to be your bottleneck.

Rust is hard to understand just reading it let alone determine why it’s doing something.

Rust

fn does_what(n: u64) -> bool {
    match n {
        0...1 => false,
        _ => !(2..n).any(|d| n % d == 0),
    }
}

Golang

func doesWhat(value int) bool {
    for i := 2; i &lt;= int(math.Floor(float64(value)/2)); i++ {
        if value %i == 0 {
            return false
        }
    }
    return value > 1
}

Not talking about the functionality merits of the two, but in comparing the _code itself. One of those is much easier to determine what’s going on and what it should be doing versus the other. I don’t feel like fighting my code. Programming is a tool to help. If it takes more work to use the tool to achieve your goal, it’s the wrong tool. Rust is the wrong tool for web development.

@EthicalAI@beehaw.org · edit-2 7 months ago

Meh, I’ve programmed in both. Rust is “hard”. I wouldn’t ask a company to write in it, because it might be hard to get devs for it. However, open source is different. Rust is not hard enough for most developers to learn, and most developers love it when they learn it. On top of that, GoLang is practically an expert in hidden, annoying bugs that rust almost categorically eliminates. Golangs panics don’t backtrace unless you write them in a certain way, you have to know the golang “culture” of error handling, and then without a good match statement or ? macro you are left with ifs under every goddam line of code to do your own manual error checking. Golang goroutines are not as intuitive as one might think with how they close when the scope they come in from closes and their channel patterns. And the “context” passing takes a long time to learn how to do right. It’s an intuitive language at its core, its docpage being one page, but it’s culture is like python’s, needing a year or more to really know what best practices are. I tbh think they are just about exactly as hard as one another, but one, golang, leads to more bugs. Compile time is not that important when you can ensure that at compile time the thing will run.

Refactoring rust sucks, but by keeping your structs small you can usually avoid it.

@maegul · 7 months ago

Thanks!

@potterman28wxcv@beehaw.org · edit-2 7 months ago

About your specific example I find the Rust code to be much simpler to understand than your equivalent Golang code…

To understand the Rust code I just have to understand each case. 0…1 returns false. Ok. 2…n returns true iff no divider was found between 2 and n-1. Ok the function is primality test

The Golang code is much harder. I do not take into account the division by 2 because its not part of the original Rust code.

A for loop starting at 2 that look for divisors. Then the return value > 1. Why is it OK to just return value > 1? Oh that’s because the loop did not return. Why did it not? Either no dividor was found or we did not get into the loop at all. If value > 1 we have the guarantee the loop was executed so it’s really a primal number. If value <= 1 it is either 0 or 1 which are not primal. Ok, so we return value > 1.

I think people dislike Rust because it has a lot of functional languages constructs and people are not used to code in functional languages.

Whatever you fight in Rust you end up saving time by avoiding runtime bugs that would have plagued your productivity anyway. I’d much rather have a language with a hard entry but with solid and maintainable code rather than fast-written spaghetti that no one knows what it is supposed to do 2 years after.

@EthicalAI@beehaw.org · 7 months ago

Reading your other links I don’t understand why you take this stance

The problem with forking Lemmy is in starting from all the bad that is inherently there, and trying to make it better. That is way more work than starting fresh with more developers. IE, not using Rust for a web app and UI, better database queries from the start, better logging/functions from the start; not adding on bandaids. A fork of Lemmy will have all of Lemmy’s problems but now you’re responsible for them instead.

Assuming you are an experienced developer, as am I, I have said this almost a thousand times. It’s almost always wrong. Lemmy’s codebase is decently clean and organized, and seems to be around 50,000 lines. All you are going to get by forking is having to rewrite a bunch of CRUD.

The other option is writing tools and plugins to interface with Rust’s API. Lastly, as long as you keep your history clean on your fork, you can continue to rebase onto the original. I think rewriting is just a terrible idea.

Anyway, if you do end up leaving lemmy, why not just go to one of the many standard forums? Discourse is nice.