If an LLM can generate 90% of a HTTP API correctly, why would you want to do it manually?

Because figuring out which 10% it did wrong and then fixing that will take longer and be more effort than just doing it from scratch myself.

But the only ones doing it are bigger companies anyway, they’ve money to do it.

Yeah… Compared to the expense of buying literally all the computer hardware in the world, paying reddit for API access is nothing.

Oh, it will ‘find bugs’ alright. And then flood FreeBSD’s bug report system with bullshit bug reports that turn out to be nothing, but require expert human review to discern that.

I find it works best when you give it a very small/simple task to do.

If it’s a small/simple task, why do I need help at all?

What I don’t get, though, is how the vibe code bros can’t discern this reality.

How can they sit there and not see that their vibe-coded app just doesn’t do what they wanted it to do? Eventually, you’ve got to try actually running the app, right? And how do you keep drinking the AI kool-aid when you find out that the app doesn’t work?

Stuff that can’t be reproduced and “only” comes up because of some timing issue/race condition is often the most shit to hunt for

Ah, but if you can’t reproduce it, you can just put in an entry of ‘could not reproduce’ and close the bug report. Case closed. Go home and enjoy a nice beverage.

I like using bullet points sometimes. It’s a good way to organize lists of things.

anyone can scrap Reddit or Lemmy just fine to train on LLM.

Well, until you hit flood limits and reddit keeps giving you ‘prove you’re not a robot’ screens or just timegates you from loading new pages at all.

Plus, it would be a lot more convenient to have API access to automatically provide your AI with only the text of posts and not have to scrape/strip entire pages.

You could scrape reddit without paying for it, but it would probably be a much slower and more annoying process.

You see, that’s why when I need to use Discord, I just use it in the browser anyway. No need to install an app that’s just going to be a browser tab in disguise.

Needing to run a full-fledged browser in the background in order to display your html/css frontend adds a lot more performance cost than necessary, making the app eat up far more RAM and CPU than necessary. It probably also introduces a lot more security vulnerability concerns that an otherwise simple app shouldn’t have to worry about. And then there’s the dependency chain you’re introducing – now your app needs to be updated every time the underlying browser gets an important update … and maybe needs to be tweaked/rewritten to accommodate that browser update if it changes the way the browser interacts with your app frontend.

There are plenty of other GUI frontend frameworks that are also expressive, simple, and well-known, without all of these potential problems associated with them.

My 2nd most hated trend in modern programming. (Behind AI forced into everything.)

Well … it’s not terribly useful without good roads.

Or, you could use it on bad roads, but only if it had good tires.

Maybe they could have used it to a limited extent within cities, but that’s about it.

This fun but impractical “mechanical horse”

I don’t think it’s actually that impractical.

It would still be better than walking in some situations. Especially if you modernized it with some nice rubber tires, and maybe a bit more stable geometry.

I’d be interested to try their distro when it’s done…

I’ll just need to remove the french language pack first. Let me look up instructions for that real quick… Oh no…

With the resources they have – and the goal of having security and control over their digital supply lines – I don’t see why they wouldn’t roll their own distro. Maybe heavily based on an existing distro (looking at you, Debian) but something that’s completely under their control, so they can choose for themselves exactly what packages are included, what security settings are used as defaults, when and if to update things, etc.

Edit: after (gasp) actually reading the article, it seems that each department within the French government will be responsible for their own implementation, which means they might all end up using different distros in different departments.

And cats don’t weigh anything

*unless they’re standing on your chest while you’re in bed, in which case, they weigh several hundred kilograms. But that’s not the case here. So, yes, we can work with the approximation of a weightless cat in this scenario.

Exactly. This only works while falling.

So, no, you’re not creating infinite energy, you’re just converting gravitational potential energy to electricity (for a short time).

Yeah, that’s why I included the word ‘openly’ in there.

more-and-more a hyper focused bubble of what you want to hear, and only what you want to hear

*and only what Elon Musk wants you to hear.

There are plenty of mainstream platforms to reach such people on that aren’t owned by (openly) Nazi billionaires.