Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Learnings from 100K lines of Rust with AI (2025)

zfhuang99.github.io

Post reply on HN
Print view View original

131–140 of 225 posts

Re: Learnings from 100K lines of Rust with AI (2025)

#131

Post by johnnyanmac »

embedding-shape wrote: post #129
johnnyanmac wrote: post #128

Earlier quoted context omitted.

> Unless you can somehow provide some arguments against it, We're year 4 into this discussion and camps have only gotten more bifrucated. There's no 1-1 discussion to have about this as of now, at least not before the crash. Your only hope in such discourse is not trying to convince the other party how wrong they are, but appealing to an as of yet undecided party. Be it with reason, or simply pointing out how absurd…

> Your only hope in such discourse is not trying to convince the other party how wrong they are I don't care about convincing anyone, the ones I reply to or others, but if you take the time to leave a comment, at least make it something to read and think about instead of soundbites like "This is astrology for devs", it's plain boring to read and makes HN worse.

>I don't care about convincing anyone

That's fine. Others will care for you.

>it's plain boring to read and makes HN worse.

I chuckled at the joke. Surprising amount of layers to it.

Though I never strove to be a comic nor writer, that kind of terse, compact punch makes me envy those of such literary talent.

Re: Learnings from 100K lines of Rust with AI (2025)

#132

Post by pjmlp »

The moment a language is the output of a natural language compiler, the language itself is kind of irrelevant.

Change the skills, ask the agent to do exactly the same in something else.

I am slowly focusing on agent orchestration tools, which make the actual programming language as relevant as doing SOA with BPEL.

Re: Learnings from 100K lines of Rust with AI (2025)

#133

Post by throw-the-towel »

pjmlp wrote: post #132

The moment a language is the output of a natural language compiler, the language itself is kind of irrelevant. Change the skills, ask the agent to do exactly the same in something else. I am slowly focusing on agent orchestration tools, which make the actual programming language as relevant as doing SOA with BPEL.

The language may be irrelevant, but the hard guarantees it offers are not. Agents are still very stochastic, they need something deterministic constraining their output.

Re: Learnings from 100K lines of Rust with AI (2025)

#134

Post by dimitrismrtzs »

motoboi wrote: post #61
chadd wrote: post #54

We're working on a large Rust codebase, heavily assisted development with Claude and Codex, and one critical workflow is after you have written a spec, have the other LLM critique it thoroughly. This back and forth will take quite a while, but the resulting implementation plan will be 10x better than the original. You can automate this by giving Codex a goal, and a skill to call Claude to review the implementation sp…

I strongly believe you don’t need to call another model for that. The same model can do result fine. Just not as part of the same context. I mean that if you ask codex on gpt 5.5 to submit to a plan reviewer subagent that uses gpt5.5, this is enough to have a very good reviewing and reassessment of the plan. My hypothesis is that it’s even better than opus. The reason why submitting the product of one LLM to another…

[dead]

Re: Learnings from 100K lines of Rust with AI (2025)

#135

Post by pjmlp »

throw-the-towel wrote: post #133
pjmlp wrote: post #132

The moment a language is the output of a natural language compiler, the language itself is kind of irrelevant. Change the skills, ask the agent to do exactly the same in something else. I am slowly focusing on agent orchestration tools, which make the actual programming language as relevant as doing SOA with BPEL.

The language may be irrelevant, but the hard guarantees it offers are not. Agents are still very stochastic, they need something deterministic constraining their output.

That is where formalisms come into play.

Also it is kind of interesting that there is so much enthusiasm to use Claude and Claw all over the place, yet lack of vision on how much the whole infrastructure will improve.

Even when it finally bursts and we get into another AI Winter, what was already achieved isn't going away.

Re: Learnings from 100K lines of Rust with AI (2025)

#137

Post by thesz »

CrazyStat wrote: post #126
dxxvi wrote: post #125

Earlier quoted context omitted.

> It's not surprising that an LLM would produce different specs for the same work on different runs This is what I don't understand: AI is a computer program with its own data. If we give the same input to that computer program every time, why does it produce different outputs every time? Or does the input include LLM data + our prompt + some random data that computer program picks from its Internet search?

LLMs have a temperature parameter. At zero temperature they are deterministic: they always choose the most likely next token at each step based on what came before and the model weights, and they will always generate the same output given the same input. As you raise the temperature they will start (pseudo)randomly choosing tokens other than the single most likely token (though that one will still be the most likely…

  > LLMs have a temperature parameter. At zero temperature they are deterministic: they always choose the most likely next token at each step based on what came before and the model weights, and they will always generate the same output given the same input.
https://en.wikipedia.org/wiki/Softmax_function

"A value proportional to the reciprocal of β is sometimes referred to as the temperature: β = 1/kT, where k is typically 1 or the Boltzmann constant and T is the temperature. A higher temperature results in a more uniform output distribution (i.e. with higher entropy; it is "more random"), while a lower temperature results in a sharper output distribution, with one value dominating."

"Temperature" in the context of softmax does not change a "winning" token, it changes how much probable (in the sense of softmax distribution) winning token will be. If the winning token is "New York", it will be a winner with temperature close to 0 and with temperature of 1e9.

The actual selection of the random token is done separately by using inputs outside of the softmax distribution, for example, by using random number generator. I believe most of LLM configs have a seed for the random number generator.

More than that, generation of code in most programming languages is done with the more guardrails such as beam search guided by schema, syntax and semantics.

Re: Learnings from 100K lines of Rust with AI (2025)

#138

Post by rixed »

dist-epoch wrote: post #96
olafmol wrote: post #73

Earlier quoted context omitted.

An LLM should not "generate specs", a human should. The LLM can work from the specs. It can never infer meaning from a vague prompt. If so, it will start guessing. Every human that ever did functional specification or information analysis at some point knows this. Or has learned the hard way, something with assumptions and asses ;)

The guessing of a LLM for a vague prompt is better than the one of your average developer. A prompt like "write these two files on disk" will very likely make the LLM do some sort of an atomic write/swap operation, unlike the average developer which will just write the two files and maybe later encounter a race condition bug. You can argue the LLM output is overkill, but it will also be more robust on average.

What kind of race condition do you have in mind?

Re: Learnings from 100K lines of Rust with AI (2025)

#139

Post by dnautics »

claytongulick wrote: post #123
dnautics wrote: post #84

Earlier quoted context omitted.

LLMs do reason (they just sometimes don't reason well ). I assure you I've met many devs and "engineers" that reason less than LLMs, and are black boxes, especially in terms of the code they write.

> LLMs do reason No, they don't. They are token predictors that use statistical techniques to emit the randomly weighted next most likely token given the previous token list. The result is a strange mimic of human reasoning, because the tokens it predicts are trained on strings that were produced by humans that were reasoning, but that's not the same thing. Human cognition is complex and poorly understood, and the na…

You have moved goalposts from reasoning to "human cognition". I won't tolerate that sort of slippery wordplay.

Reasoning is making analogies between logical patterns found in conceptual space, with a direction of time (statements precede conclusions). For example. A => B and B => C. You may now deduce A => C. For something fuzzier, A~D and B~E, you may now deduce that D~=>E. This is the sort of thing that higher layer attention mechanism is capable of doing.

> This is a box they can not escape.

Would you say that Helen Keller was less capable of abstract reasoning because she had more constrained access to sensory input?

Re: Learnings from 100K lines of Rust with AI (2025)

#140

Post by NobleLie »

b40d-48b2-979e wrote: post #72
CrazyStat wrote: post #69

Earlier quoted context omitted.

If you ask 10 different humans to produce the spec with the same information (prompt and context) they will also produce 10 unique answers that will contradict each other and (depending on who you asked) may be just as confident. There are real decisions to be made when going from a vague prompt to a spec. It's not surprising that an LLM would produce different specs for the same work on different runs. If the prompt…

LLMs aren't people. They don't reason. They're token generators, a black box. Your analogy falls on its face with any scrutiny.

It appears they don't need to reason or be intelligent to be able to produce working solutions for code. But sure let wild and unmonitored? I wrangle my LLMs like the code monkeys they are. They help materialize code and then you need to sculpt it (and test harness of varying sorts)

It really can be useful. It's very different from old world programming.

Post reply on HN