

Race conditions I’m pretty sure aren’t language specific. Rust can have them too.


Race conditions I’m pretty sure aren’t language specific. Rust can have them too.


This was caught during review, so nothing was “introduced” quite yet. And there’s no indication that Linus wouldn’t have produced a race condition (a notoriously more easy to miss kind of bug) without the use of AI tools.
It’s only being exaggerated so much because he used AI.


Criminals don’t own datacenters running the models that actually pose a threat.


Official source this is based on is https://www.ncsc.nl/en/alerts/update-php-vulnerabilities-remain-a-reason-for-immediate-patching, which mentions they’re investigating signs of exploits and that they have at least one confirmed exploit that happened abroad.
It’s still a complex exploit, mind, but not an unusable one.


I mean, the national cybersecurity council here saw exploits happening in the wild. So it’s definitely possible and practical enough for it to happen.


At the architectural level, there’s definitely a difference yes. But at the smaller scoped implementation level the latest models are pretty good compared to humans. Not perfect by any means, sure, but it’s not all drivel anymore. It was 2 years ago, but today’s definitely different.
Race conditions are generally harder to catch, even for human devs. And you’re saying it as if Linus wouldn’t have introduced it had he coded everything by hand, which I don’t think is remotely guaranteed (like he never makes mistakes). And the review system clearly works just fine still.
Good developers can accelerate their work with agents, and you’re right that you can generate tons of LLM-powered slop with it too. But that’s why dismissing anything an agent wrote immediately is also wrong. The review determines if it’s good enough, whether a human or an agent wrote it.


Sure, that’s why I said “not necessarily”. LLMs can be misused and made to do dumb shit too. But I have seen plenty of bullshit come from human developers too.


Verifying LLM code isn’t necessarily harder than verifying a human PR. I hope that this also boosts automated test tooling so that it also gets easier to test these PRs, since the focus will shift to review rather than writing.
So don’t kinkshame y’know.


They’re required to make a copy to safeguard the data and prevent it from being wiped, but that doesn’t actually mean the request will be honoured and data will be shared. Thus far, police have said no data has actually been transferred pending the review of the request.


They haven’t actually shared anything yet, afaik. When the request comes in the authorities have basically no wiggle room regarding safeguarding the data, but it is only shared once the request is approved as legal.
In plenty of countries carrying knives in public is not allowed. You can own one for use in your home or work but don’t brandish it out in the open.
The reason few people know this is because it’s rarely ever needed, and the people violating that law are generally the people you want arrested anyway (e.g. because they’re trying to stab someone).


I think the rules written in Clean Code are mostly a response to the kinds of bad code that Martin had to deal with. I think compared to the bad examples he cites his rules are an improvement.
Ousterhout mostly pricks holes in these rules, based on the bad code he has seen. But he’s not infallible either since he wrote wrong comments on his own code, which is a bit of a self-own.
I don’t think it’s entirely possible to provide one set of rules that can be universally applied. You need to adjust your code to the problem you’re solving, your audience, your programming language, etc… You’ll always find counterexamples where doing something a little different works better.
Code should be self-explanatory, but a good informative comment can go a long way. Long methods should probably be broken up, but not always and not beyond the point of usefulness. I’ve seen examples where the UB version read better than the JO version, and vice versa.
The value for me in these discussions is finding out why people write what they do, and what they prefer to read (and why). That helps me think about how I can best structure my code for readability. But there’s no point in setting rules and becoming dogmatic about them.
Not sure the pose is the primary issue here…


I am, but you can use a custom analyzer to stop it from compiling if misused. So you can actually stop a caller from writing that and walking away :)


It’s not impossible, at least in C# I know you can mark objects that need to be RAII with some custom attribute and enforce correct usage patterns with a custom analyzer. You need to write it yourself but it’s not impossible that way.


If it turns out that it gets ruled as copyright violations, you can bet your ass they’re going to reform copyright law instead of rolling everything back.


Well that was also discussed, the person who suggested not having an AI disclosure was against it being used for advertising, but was also open to having a general one that didn’t allow the company/model. Not sure that discussion has concluded yet btw.
Perhaps you missed it, but the second affair that was uncovered was that he was fucking the head of Nvidia’s health divison. Who is most certainly his subordinate.