TehPers

Beehaw defederated from .ml and I’m still seeing the image, oddly.

At work? Do what they want you to do I guess. If they are mandating you mass produce slop, do exactly that. I’ve tried fighting against it. There’s nothing you can do when your coworker opens a 785 file PR (not an exaggeration) and it needs to be reviewed today. Your carefully crafted code can never keep up with the constant waterfall of whatever the fuck Claude/ChatGPT/whatever calls acceptable these days.

Your best bet at work is to direct the agents others will use. Use a strict linter. Mandate a strict formatter. Add to the instructions that tests must be written and must pass. Do what you can to force it to emit semi-acceptable code (or fail if it can’t).

Outside of work? What’s a coding agent?

JavaScript takes yet another approach. It doesn’t have an equivalent of the Character.isWhitespace() method, but we can use the \s character class in a regular expression

\s might be the most obvious whitespace test, but the expectations are a bit more ambiguous when it comes to regex. Every regex runtime is different, so without looking it up, you’re going to end up with behavior you might not expect.

Instead, if you want to match Unicode whitespace, you should use \p{White_Space} in Unicode-aware mode (u or v flag).

I haven’t looked through any other runtimes, but something tells me the inconsistencies won’t stop at these three.

According to the docs, Rust’s char::is_whitespace checks if the value has the White_Space property, and regex does the same in Unicode mode (which is the default).

I liked Zed a bit, but I ran into some bugs that made me switch back to my previous editor. I think I’ll give it more of a shot in the future once it’s a bit more mature. Otherwise, I enjoyed how responsive it felt, and having a single toggle to completely turn off AI features was nice.

For Kate, I view it as an alternative to Notepad++. It’s quite a bit more powerful than N++, but to me feels like it fits in around the same place you described between simple text editor and advanced coding IDE. It’s a great default for people who tend to modify a lot of config files.

The LLM when you tell it to rewrite it in Rust.

The main problem I have with the bulk of Anti-AI opinions is that they are simply that: opinions, backed up with parroted “facts” many of which are baseless, untrue, blown out of realistic proportion, etc.

This is, to an unbelievabe extent, true of the “Pro-AI” crowd (in particular those selling AI). It’s to the extent that the economy is inevitably fucked once investors start pulling out. AI CEOs lie regularly about the capabilities of their products.

In any case, “pro-AI”/“anti-AI” is a political approach to AI, and politics revolves entirely around opinion. If you truly want “facts” then you have to be willing to accept facts that you disagree with as well.

It is a fact that many bugs have been found with AI, yes. It has provided value to a lot of people.

It is also a fact that for all the good AI has done for people, it is also an unregulated bullshit emitter. Shit, just the other day, I tried asking Kagi Assistant a question, and it gave me an answer in Chinese. My question was in English. It had nothing to do with languages. But if you want numbers, these AI companies publish their benchmarks publicly, so go look at those and consider whether “correct 80% of the time in a benchmark the model was optimized for” is good enough to always trust.

Finally, it is a fact that AI companies have done irreparable harm to the world on an enormous scale, socially, ecologically, politically, and so on.

Am I the only person that thinks fixing kernel bugs discovered with AI is a good thing?

No. A bug is a bug.

Even everyone saying Linus went full AI… Like… What? IMO he’s being practical. It IS a tool, and there ARE a lot of things it should not be used for, but using it to patch things before human verification happens is a pretty sensible thing to do.

Using AI to fix a bug is different than fixing a bug found by AI. Even in Torvald’s vibecoded patch, he introduced a new bug, then asked the reviewer to take over the patch and went on vacation.

Would love to hear from someone who has been programming for a long time and isn’t all just “AI bad because AI bad”.

While I get where this is coming from, being dismissive of people who you disagree with doesn’t exactly inspire healthy discussion. If you want genuine opposing opinions, those will only come from those you disagree with.

This seems like a problem that can be approached entirely in embedding space. Find embeddings for words that you want to avoid, then filter your dictionary to remove anything too close to those.

No need to go through full text gen or anything, though if you want to go that route, a classifier like Jev (or OpenJev/etc) would be best.

the classic language models before LLMs existed

Which models are you referring to here? Earlier versions of GPT (like GPT-2)?

There are small LLMs like Phi 4 Mini that you could look at. Those were part of the “small language model” craze (SLMs).

If you want meaningful commit messages though, nothing beats handwritten messages. Commit frequently and the messages should be short and easy to come up with.

“Guns cause a lot of shootings, so let’s ban guns”

~ any reasonable country on the planet

Wouldn’t this also be true of the Windows version though? Windows applications usually ship alongside any libraries they use.

2GB seems way too excessive.

I also think the attitude toward it would be more positive if those in the spotlight for training these models weren’t actively creating global economic crises, bringing down random websites with sustained high-volume traffic, filibustering collaboration with low quality issues and PRs, and completely disregarding the licenses for the code they’re using.

In fact, were these models only used to better the world, I’m sure that many developers would be more than happy to support the effort even if it meant licensing the code for training under more permissive licenses (assuming they could get their contributors on board as well).

Instead, we all get to hear about how some egotistical billionaire is trying to become a trillionaire, or how the prices of everything skyrocketing is a good thing because some rich people got more rich.

It’s nice to see that business execs can be automated now though. Well, except the part where they openly bribe politicians I guess.

How are you accessing the website? Are you using a web server like miniserve or are you just opening the HTML files in a browser?

Beginning a path with / makes the path absolute for that origin. An origin is basically the URL’s scheme, domain, subdomains, and port (so https://www.example.com/foo/bar has the origin https://www.example.com/ - ignore the trailing slash, Lemmy keeps adding it…). If your URL begins with file:// then absolute paths will likely be absolute to your filesystem, not relative to your project’s directory.

If you can, try serving the directory with something like miniserve and see if it works. It’ll give you a better picture of what the website would look like if it were published for real. Otherwise, you can make the path relative to solve the problem without doing that.

This article does everything in its power to try to twist DHH’s words in some charitable manner while assigning (and subsequently defecating on) some arbitrary Rubyisms to an article that had nothing to do with Ruby.

Ironically, the article does exactly what it claims the others were doing by assuming those others were responding to his article based on uncharitable assumptions or poor faith.

The words a person chooses to write in a blog post are thought-out and explicit. He chose his wording to convey the meaning it did, knowing full well what the connotations of those sentences were. He reinforces those connotations with his later speech. And now he solidifies that message by accepting something of monetary value from Musk.

The far-right plausable deniability game isn’t going to work here. Evaluating speech solely on the semantic meaning of the words alone disregards the most valuable parts of communication. His message was clear, and it’s people like this who choose to ignore it (or who silently agree, though I don’t know this author so I have no clue which it is).

This is just speech-to-text, right? If so, advertising it as “AI”, while a historically popular term for it, is going to kill any interest in it before anyone has a chance to try it. I don’t use Ubuntu, nor do I have any interest in it, but STT tools are powerful accessibility tools.

To a repository? Might be good to put that in the post itself, honestly.

Is this just an ad? Or is there a repository somewhere that I can’t find?

made me realize do we still need documentation of how a software work when a AI can easily explain it?

I’ll respond for the sake of argument, but to put it bluntly: there are no stupid questions, just questions phrased stupidly.

Documentation comes in many formats. There’s prose documentation split into articles and chapters, there’s video documentation, inline documentation for code, tutorials, training, and so on. Most things are documented several ways. This is because people learn things differently from each other.

To remove documentation entirely and rely on only LLMs to document removes all of the ways that people might learn and narrows it down to only one form of documentation. Even if we ignore how fallible LLMs can be, having documentation in only a single format limits learning only to those who learn effectively through that format.

In other words, relying only on LLMs for documentation discards all of the millenia of teaching strategies learned throughout the course of human history and puts it in the hands of a stochastic magic box that is incapable of recreating those strategies.

Yep. It’s the only way to do it sometimes.