therepanic

I suppose Anthropic has relaxed its safety checks to such an extent that Claude now agrees to port iOS to Android phones?

minutes not hours

It depends on the goal. Right now, one thing I definitely wouldn't want to do is write js bullshit by hand - absolutely not.

But if we are talking about personal projects where the aim is to learn complex interesting things, then of course writing code by hand is the only way to actually learn.

I guess antirez agrees with this as well[0].

[0]: https://x.com/antirez/status/2105569301944238403

in that case it is truly amazing, because I have never seen anyone even with current ai capable of being so productive and truly understanding what they are putting out.

i see, but my statement does not become false because of it. his agent pushes several commits an hour in a continuous loop, meaning this respected developer is not involved in the process at all.

around 4k commits in a week.

i am not sure anyone actually understands what is going on under the hood of the project. looks like average slop.

>> Exactly a week ago (as a joke), I started writing THC,

you? i hope you have read at least something of what your agent wrote about the project.

It's funny how, just before that, they tactfully cut the limits on the $200 subscription.

As a 17 year old, I agree with this. Ofc I'm against all the hate directed at PG, I believe that all knowledge has value regardless of its economic utility, but I understand where the hate is coming from. Personally, I find LLMs boring for now, and I'm more focused on CS and electrical engineering.

This backend effectively does exactly that.

Because it's fun!

In our tests we don't store any metadata other than the ID and the title, and to be honest, I think that's beyond the scope of this project.

LeetCode problems are usually added after each contest, and as far as I know their statements generally don't change afterward at least the actual problem itself doesn't.

As for test cases, there are occasionally solutions that are technically incorrect but still pass because a particular edge case isn't covered. We maintain our own test cases, including a DSL for writing large stress tests, and try to cover all possible cases.

In the worst case, a solution might pass our local tests but fail on LeetCode. I expect such cases to be rare and one of the benefits of having the tests open is that when someone finds a missing case, it can be added so nobody else runs into the same issue.

I love open source and open systems in general. Popular competitive programming platforms aren't like that: not only is their source code closed, but they also don't publish the test cases used to evaluate submissions.

There are already projects that can submit your solutions through the LeetCode API and return the results. But I don't really like that approach, so I decided to build an open set of tests that aims to match the quality of LeetCode's tests as closely as possible. This way, you can run everything locally, without needing for example, an internet connection.

I see the project being useful for its most straightforward purpose: testing your solutions locally on your own hardware, without any submission delays and even completely offline. Since the tests are open, people can also build different tools around them. For example vim extensions and other integrations. Ofc it's not that simple, we currently support about 1400 problems (out of 4000~, including premium), and not all topics are covered.

Even at these prices, switching from subsidized subscriptions to the API just isn't worth it. Not even close.

You don't have to preface every criticism by saying you're a huge AI proponent. EVERYONE overdoes this.

neovim

So what should newcomers do who haven't yet gained enough expertise to ask LLMs the right questions?

Why?

I recommend reading this first to understand the author's point of view: https://antirez.com/news/169

the author does not believe in the technological singularity.

By all appearances, they did not succeed in doing so.

After Apple, Microsoft is now taking up the baton. Who is next?

> So how do you find startup ideas without looking for them? By working on projects with your friends. That's where the very best startups come from. Initially they're not even meant to be companies. They're just something people built because they thought it would be cool. That's how Apple and Google and Facebook all started. None of them were meant to be companies at first.

That’s cool and it’s a cool post, sure, but it sounds ridiculous when you look at how many dumb GPT wrappers there are in YC batches nowadays.

It looks quite unusual, I will definitely try it.

To be honest, judging by their repository, it doesn't look like they've stopped accepting third-party PRs.

To be honest, I share primeagen's view that LLMs handle translating code from one language to another quite well. As far as I know, they converted the languages file by file. This is what led to such a high volume of `unsafe` code. Although, in any case let's be honest, this is causing, and will continue to cause, various issues. I find it easier to live with this point of view.

Why should I pay more for a Chinese model's wrapper? Or is Cursor's subscription more subsidized than Kimi's? Anyone know?

No, not them, but those who provide subsidies to AI labs, which is why such people spend almost nothing

About 9 days ago, Jarred wrote that it was far from certain that this would merge and that it was an overreaction. Ironic.

Well, if that's the case, then in your concept the issue isn't what will happen to the programmers, but rather to all the work in general.