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

Google fixed more Chrome bugs in June than over the past two years, thanks to AI

blog.google

Post reply on HN
Print view View original

631–640 of 667 posts

Re: Google fixed more Chrome bugs in June than over the past two years, thanks to AI

#631

Post by fluffybucktsnek »

rustfreeforme wrote: post #628
fluffybucktsnek wrote: post #624

Earlier quoted context omitted.

The problem is that you think Rust programmers despise C or C++. We don't. We just think there are better alternatives for modern development. In fact, a lot of Rust programmers were experienced C or C++ programmers. And then you say: > [...] your belief system. What do you mean by that? What beliefs? This just sounds like projection. > Please do so at the earliest opportunity, so that your language can actually beco…

[dead]

> YOUR ENTIRE WORLD IS BUILT ON C/C++.

What a reductive view of reality. That's like saying our entire world is built on assembly. That's just missing the point.

> Saying Rust is the replacement to C/C++ is like saying Python is the replacement to C/C++. No, not quite.

With analogy such as these, I see where the "No, not quite" comes from. For starters, we can write actually low-level Rust code enough to write kernels without needing to interface with C code, only raw assembly. I don't think Python can do that.

> I don't go to Google for advice on how to design software, thanks.

Good thing they weren't giving advice, but reporting that most security vulnerabilities came from just bad memory access, which contradicts your counter assertion. None of that is subjective. Whether you think they are the ideal programmers or not is irrelevant.

Re: Google fixed more Chrome bugs in June than over the past two years, thanks to AI

#632

Post by rustfreeforme »

fluffybucktsnek wrote: post #629
rustfreeforme wrote: post #626

Earlier quoted context omitted.

[dead]

> In time you will learn about all of the faults and flaws in Rust, after you've suffered the pain of trying to rewrite the entire universe in it just to fix one single class of problem you are obsessed about--inevitably introducing many new bugs in the process, because that always comes with the territory. Good thing Rust doesn't just fix that one class of problems, but also provides better tooling to fix other clas…

[flagged]

Re: Google fixed more Chrome bugs in June than over the past two years, thanks to AI

#633

Post by insanitybit »

inigyou wrote: post #623
insanitybit wrote: post #591

Earlier quoted context omitted.

I can't take this seriously. The claim "you can look at SOME code and see how to optimize it" is trivially true, and saying "you can't look at ALL code and see how to optimize it" changes nothing.

Exactly, you have to find the slow code first, with profiling, and then sometimes it is obvious why that code is slow. It is unlikely you can just look at a project and find the slow code in it.

You absolutely can. I've given enough examples that are obviously true. If you can't read code and reason about its performance, that is a personal issue.

Re: Google fixed more Chrome bugs in June than over the past two years, thanks to AI

#634

Post by fluffybucktsnek »

rustfreeforme wrote: post #602
fluffybucktsnek wrote: post #554

Earlier quoted context omitted.

Quite frankly, this post is massive set of unsubstantiated claims. For starters, no one said the only problem with Chromium is C++. But it certainly contributes with some of the problems (specially those related to security). And, no, these bugs don't come from "noob programmers." They come from experienced engineers. Then you just mention that removed some "unnecessary and unwanted" code. According to what standard?…

[dead]

> Oh no, it seems my one brief post wasn't long and detailed enough to immediately and totally convince some Random Doubter on HN, therefore my 'claims' are 'unsubstantiated.'

No, you just didn't cite any actual data, only assertions. Don't come with this excuse of "busy day", you had enough time to create an HN solely for spewing anti-Rust nonsense.

Keep in mind, HN is a place for intellectual discussions, not for dumping your pet peeves. If you don't actually have anything actually productive to say, it's preferable to omit yourself before the mods or the flags do that for you.

> Nah. It doesn't.

Then try to prove it. I already cited a source elsewhere, which you just ad hominem'ed. The burden of proof is now on you.

> It's definitely not possible for fluffybucktsnek to just take my word here. I wonder what biases he or she might have.

Yeah, what biases does the name "fluffybucketsnek" indicates? "Rust free for me" doesn't sound as impartial in this discussion. It just sounds like you came here for an ideological battle.

> Yes, you should just assume that's what must have happened, and then move on, blissfully unaware of how wrong you are.

I didn't assume anything. If anything, I kind of wish you did/could prove me wrong. You could, at least, post the code of your fork.

Instead, you need us to assume that all your claims, none of which has any sources, are true. All I did was show a different possibility.

> Sure it does. There is extra machinery/interfacing for all the extra Rust crap bolted on, and it gets more and more complex over time as they integrate more of the crap.

Every FFI introduces some manner of complexity, specially when one language (Rust) is more strict than the other (C++). I don't say C is complex because JNI is a hassle.

> Furthermore they vendor the entire Rust compiler, library, etc, all of which has to be compiled first before the browser.

That just seems like a Chromium project issue, not a Rust one. It's weird that for issues with C/C++ in Chromium, you blame the project itself, but for Rust, you blame Rust. Just like you don't have to vendor LLVM, you don't have to vendor Cargo nor Rustc along with the project itself. Even for compile times, similar strategies for cutting C++'s work for Rust as well.

Re: Google fixed more Chrome bugs in June than over the past two years, thanks to AI

#635

Post by Aeolos »

rustfreeforme wrote: post #632
fluffybucktsnek wrote: post #629

Earlier quoted context omitted.

> In time you will learn about all of the faults and flaws in Rust, after you've suffered the pain of trying to rewrite the entire universe in it just to fix one single class of problem you are obsessed about--inevitably introducing many new bugs in the process, because that always comes with the territory. Good thing Rust doesn't just fix that one class of problems, but also provides better tooling to fix other clas…

[flagged]

It is obvious that you are just trolling, but for other readers: our production app went from 50-100 unexplained sentry errors in typescript to 0 sentry errors in rust.

The language absolutely helps fix logic errors.

It's something that becomes immediately clear as soon as you start using the language in earnest.

Re: Google fixed more Chrome bugs in June than over the past two years, thanks to AI

#636

Post by userbinator »

calmingsolitude wrote: post #556
murukesh_s wrote: post #555

Earlier quoted context omitted.

If that is the case, isn't that the failure of your testing strategy/setup? If your style doesn't fit full vibe coding - you can do the path of AI generate code - human reviews and tests all edge cases flow - I think in any serious software - that should be the flow until next few iterations. But either way, don't see a going back to coding every line by hand now, those days are gone. Was fun while it lasted!

No. I’ve heard this argument multiple times before - “let the human review and write tests” - and sure it might work, but this is almost never practiced. AI makes coding a lot faster, so no one is really ready to spend time on manual review and writing tests when LLMs can do that as well and take you most of the way.

The AI will not only write you subtly wrong code, it'll also write you as many tests as you want that are also subtly wrong and pass when run.

Re: Google fixed more Chrome bugs in June than over the past two years, thanks to AI

#637

Post by fluffybucktsnek »

rustfreeforme wrote: post #632
fluffybucktsnek wrote: post #629

Earlier quoted context omitted.

> In time you will learn about all of the faults and flaws in Rust, after you've suffered the pain of trying to rewrite the entire universe in it just to fix one single class of problem you are obsessed about--inevitably introducing many new bugs in the process, because that always comes with the territory. Good thing Rust doesn't just fix that one class of problems, but also provides better tooling to fix other clas…

[flagged]

> Wow, what a magical language! [ waves hands vigorously ] will fix all the problems which are inevitably introduced by auto-porting from one language to another, or by manually rewriting everything from scratch. Rust doesn't obey the same laws of the software universe as other languages, it seems. Or is this a "just use magic AI bro", type situation?

Nope, that's just your own prejudices speaking. We already cited features (seen in and inspired by other languages) and you handwaved them away with "I don't care". You are the one fixated on C and C++. You are a C evangelist, if I were to make a guess. And that's on you.

> In the latter case, why don't I just use the to auto-magically solve my C/C++ problems, rather than rewrite the entire universe from scratch? You know, like Google claims to have done for Chromium C++ sources in this very news article?

You certainly can. But to spin this the other way around: in lieu of the Bun case, why not use it to rewrite the project in Rust, Zig, D or any language of your choice (yes, even C)? Given the role harnesses have in keeping agents in check, it seems plausible that languages with stricter safety features help agents avoid common issues.

> This is the type of shit noobs are always wringing their hands about as they need maximum guardrails to cover for their ineptitude.

You seem to be forgetting that proper software engineering is just as much about managing developers and their skills as the actual code structure. If a tool helps noobs produce better quality code, specially ones without as many vulnerabilities, why not employ it? Also, it's not like experts are immune to mistakes either.

Re: Google fixed more Chrome bugs in June than over the past two years, thanks to AI

#638

Post by naveen99 »

iknowstuff wrote: post #271

This is a tacit admission it was Mythos, not Gemini, despite alluding to experiments with Gemini earlier on: > Added support for model interoperability to leverage the unique strengths of both open-weights and proprietary models

Weird, why would they use open weight models.
Post reply on HN