The C++ takedown video was amazing, this video seems like it was made by a person who couldn’t gather their thoughts and just goes off on tangents. Is there a problem with the rust memory model? Is borrowing theoretically incorrect? What actually is the purported problem? I guess we won’t know until they make a new structured video.

Bloody hell, can this guy just shut up? Fucking edgelord.

Where’s the code?

Same as you can say something seemingly obvious but still wrong 🤷

“If you don’t paint a picture with this specific paint, it’s not a painting”. So if someone were to format the code differently, say put each right bracket on a new line, the compiler would fail?

Haskell and Lisp are languages that by nature pull CS and math wizards

They pull them because he learning material is written for them. If the learning material were written for non-wizards, it would attract non-wizards.

“Learn you a Haskell” is not a good book, IMO. It teaches exactly in the style mentioned. Just look at the part on applicative functors.

fmap :: (a -> b) -> f a -> f b It says: give me a function that takes an a and returns a b and a box with an a (or several of them) inside it and I’ll give you a box with a b (or several of them) inside it. It kind of applies the function to the element inside the box.

He has to describe what the function does because the function declaration doesn’t use clear variables or type names. He says the word “box” but writes “f”. Why not just write “box”?

The entire CS and maths field is filled with people who suck terribly at explaining anything to anybody who doesn’t think like them. Monads aren’t terribly complicated, but the way they are explained is. Every book trying to explain them makes the mistake of going deep first and then zooming out. They explain in excruciatingly minute detail how monads are built, what other complicated concept they are like, and use single letter names for everything in that explanation.

 class (Functor f) => Applicative f where  

     pure :: a -> f a  

     (<\*>) :: f (a -> b) -> f a -> f b

Same here. Then the examples use Just which doesn’t help with readability either. It would read much better if the code were actually literal instead of using shorthands for everything. More mental energy has to go into keeping a mapping table for every letter in the alphabet instead of it being written explicitly, right there.

It’s why I hate reading Haskell declarations and old code. Programmers back then just used 1-4 characters for everything. char, bool, m_lnr, T, R, I, f, g, h,… As if they were constantly on the run from the last person who had to read and maintain their code.

You might be a Haskell teacher, but that doesn’t say anything about the quality of your teaching. I play basketball and could even coach a team, but there’s nothing that will say whether I’m a good coach. If I get a team of people who already play well together, speak my lingo, and jive with me, they might think I’m a good coach. But if I get a team of beginners who have never heard my vernacular, can’t understand my mannerisms, and think “tough love” and the cold shoulder make me a dick, they’ll think I suck.

“They’re wrong, that’s the way it is”. Great argument. Very convincing.

Why should LISP always be formatted in a way that’s unreadable? That can’t that be the defining property of LISP, can it?

Haskell unfortunately has literature which doesn’t read well to me. All examples are jumbles of letters f(a, b, c) with “as one can clearly see”. It’s as if I’m reading a math textbook. It’s like nobody writing a Haskell book is an actual educator. Python’s documentation is a better intro to the language than any Haskell book I’ve read.

Never heard of rhombus, but I’ll check it out, thanks.

Why not? Is LISPs only, and major quality, polish notation?

Did you read the article? This has nothing to do with familiarity.

I wanted to learn lisp, but the default formatting is just terrible. If it had a different formatting and supported infix notation, it would be readable to me.

I can’t say much about hardware. I imagine the code that came out is at the level of a junior firmware engineer? Would that be correct?

Agreed, no harm in letting it write stuff. It’s your money and your time. If it wrote something you weren’t looking forward to writing or gave you a starting point, no problem.

Most of my experience with AI agents comes from work. I don’t use it much in my personal projects. But at work, since we can’t use Chinese models nor European ones, I rotate between Anthropics, OpenAI’s and Cursor’s own models (well Cursor fine tunes Kimi K and other open weights models, so practically I am using a Chinese model).

The experienthas been alright, even though we are pushed to output faster. But I’ve seen the effect it has on colleagues (and probably me) : they just accept that something runs alright and send it off for review.

The quality of the code is terrible though. My colleagues don’t understand tests, can’t seem to see architectural issues that can lead to inefficiencies, and other problems. Nearly every PR I review requires an addition to the project’s rules because the AI has succeeded in introducing something new and awful I hadn’t thought of.

So to me, LLM output is fine as long as you don’t look too closely. Otherwise, it feels like hack upon hack.

Do you have a similar experience? Maybe I am just being too critical of code, I dunno.

How is saying “Software engineering has always been bad. People just copy pasted stack overflow.” a good faith argument? Is hyperbole a good faith argument? “Honey, you never put the dishes into the washer the correct way, you just copy different styles from youtube.” Is that good faith? Don’t try and gaslight me.

If you want to have a good faith argument, we can do that.

I think the engineering part is misleading. If we’d built bridges like we build software. I’ve seen my fair share of legacy code in companies and it is far from what people think of it seems. Actually, that would be critique I have of LLMs. That they copy the style of a project, even if it is horrible.

Sure, actual engineers that have to pass a state exam in order to be called engineer probably do smirk at the term “software engineer”, and yes, code that comes out of companies is often hastily slapped together to achieve the goal being first to market, delivering something that works, not properly thought out to be future safe, and so on. But there is quite a different between a code-base written by juniors in the language or field and seniors in the language or field.

Simply waving away those differences and reducing it to “copy pasted stack overflow” which is now automated, while obviously hyperbole, is IMO pretty far from reality.

That they copy the style of a project, even if it is horrible.

My experience has not been that way. LLMs have so far introduced whichever style of code they were trained on, unless given rules and forcefully linted into adopting a better style. But there is only so much a linter can do.

That said, good models absolutely are helping with real world problems and Lemmy claims it’s literally the spawn of Satan.

I agree, LLMs have their place. But there are worlds between “LLMs = never”, “LLMs sometimes” and “my job consists of prompting AI”. Maybe we do agree, but your initial message was, to me, unnecessary hyperbole, which prompted my reaction.

LLM won’t destroy OSS for me. It will actually be the only thing left to use because proprietary software will be the first place for LLMs to take over.

If you seriously think that’s what software engineering was and still is, I feel sorry for you. You think opensource software is just a series of copy-pastes from stack overflow with no understanding?

We’re going towards a world where we can’t trust code anymore than we can trust images or videos. People all over the place are entrusting their entire jobs to AI and consumers have to deal with the output.

The responses to your question makes me fear for the future of products.

I use LLMs for scaffolding, boilerplate stuff, and basically things I can trust a junior with. For anything else, I have been forced to write it myself or rewrite it because the agents were so goddamn awful at it.

Reading that people even think programming has nothing to do with code and is just solving problems is just nuts to me. Code is a spec for a problem solution. If someone thinks the spec is neither important nor that being able to understand the spec is important… It’s like saying “I don’t care how this bridge is designed, it’s built”. Say that again and tell me it doesn’t sound nuts.

Are you a bot? It doesn’t seem like you understand what you’re talking about.

VaultWarden literally is a fork, what are you on about?

VaultWarden != BitWarden!