never_inline

Which reinforces my point that LLMs are really bad at class and module level design.

I would suggest not delegating the LLD (class / interface level design) to the LLM. The clankeren are super bad at it. They treat everything as a disposable script.

Also document some best practices in AGENT.md or whatever it's called in your app.

Eg

    * All imports must be added on top of the file, NEVER inside the function.
    * Do not swallow exceptions unless the scenario calls for fault tolerance.
    * All functions need to have type annotations for parameters and return types.

And so on.

I almost always define the class-level design myself. In some sense I use the LLM to fill in the blanks. The design is still mine.

I think this is just anthropomorphism. Sub agents make sense as a context saving mechanism.

Aider did an "architect-editor" split where architect is just a "programmer" who doesn't bother about formatting the changes as diff, then a weak model converts them into diffs and they got better results with it. This is nothing like human teams though.

> This isn’t just academic elegance, it kept phone switches running with five nines of availability.

Hmm....

> Erlang is the strongest form of the isolation argument, and it deserves to be taken seriously, which is why what happens next matters.

OK I think I know who wrote this.

> The problem isn’t that developers write circular calls by accident. It’s that deadlock-freedom doesn’t compose.

Is there a need to regugriate it in this format? "two protocols that are individually deadlock-free can still combine to deadlock in an actor system." This is the actually meaningful part.

> Forget to set a timeout on a gen_server:call?

People have pointed out its factually wrong in the thread. Eh

> This is the discipline tax. It works when the team is experienced, the codebase is well-maintained, and the conventions are followed consistently. It erodes when any of those conditions weaken, and given enough time and enough turnover they do.

I know this is an LLM tell, but can't point out. It makes me uneasy to read this. Maybe the rule of three? Maybe the reguggeiation of a elementary SE concept in between a technical description? Maybe because it's tryhard to sound smart? All three I guess.

I could go on, but sigh, man don't use these clankers to write prose. They're like negative level gzip compression.

Didn't want to hear the grok thing from handle named "dyauspitr". My day is ruined.

* gwern

* lesswrong

Amazon has been quite useful for me as a single bachelor living in an Indian metropolitan city.

1. I get very useful items at very good prices, many of which I would have to wander the city for hours to find, or couldnt find at all: - Eg: I got a pair of adjustable dumbbells at <2K INR. Some of you would call it a cheap knock. But it has been super useful and I would have not bought it if it cost 8k INR. I brought a whole bosch repair toolkit at good price and it has been invaluable for fixing electric/plumbing etc.. issues. I got a high volume travel bag - I didn't even know 40L travel bags existed and wouldnt have brought one if not for amazon. I could go on.

2. Amazon Fresh is usually cheaper for groceries and maintain consistent quality compared to local supermarkets. I will also avoid the need to walk long with the grocery bag.

3. Electronics are significantly cheaper on amazon and again the need to search.

Maybe all of this can be even better as you said. But bottom line is that their operations look pretty efficient to me. Their catalogue is pretty much unmatched. (They may be losing money on retail business - but that's not my position to care as a customer. As other commenter pointed out, it may not even matter much for stock price.)

Yeah its obvious clanker writing. I don't even mind using LLM for code but this rubs the wrong way.

Nice. Despite being anti-AI, He knows more about LLMs than most "thought leaders" and "AI enthusiasts".

Are there any car brands which don't do this fancy-useless-tech kind of thing? I have never driven a car, but a motorcycle with handle-lock would never be unlocked without the physical key.

Spinning 1K goroutines per request doesn't feel right to me for some reason.

Isn't trigram search supposed to be better?

https://swtch.com/~rsc/regexp/regexp4.html

+1 for userns, there's also proot (userspace chroot) and fakechroot (using LD_PRELOAD).

From Steve yegge's post

> Better UIs will come. But tmux is what you have for now. And it’s worth learning.

So brother has 2 claude code accounts and couldn't vibe code a UI, huh?

Poe's law &c.

I don't think he intentionally lied. He just didn't know how to check that and AI wrote

   - [tick mark emoji] implemented CSS and JS rendering from scratch - **no dependencies**.

In many enterprise environments, deployment logic would be quite large for bash.

Build a CLI in python or whatever which does the same thing as CI, every CI stage should just call its subcommands.

Someone tell the AI labs to stop training on tutorial code.

I was about to try it out, having heard good things.

But this leaves a very bad taste.

Guess I will stick to aider and copy-pasting.

I can't for the life of me tell what it's about.

To play devil's advocate: all the people using AI are not being significantly more productive on brownfield applications. If GP manages to find a Big Co (tech or non tech) which doesn't precisely bother about AI usage and just delivering features, and the bottleneck is not software dev (as is the case in majority of old school companies), he/she would be fine.

There's certainly an overlap between the crowd working on AI (since it's the current hype) and the crowd finding this sort of hyped up advertisement speak impressive.

It's pretty obvious that embeddings are not enough (and you need hybrid search or similar) for people who understand how they work. And yet there had been a flood of chat UIs in 2024 which are basically embed + cosine similarity, all having 10k+ stars. Now it seems to have slowed down.

All is because the taste for careermaxxing outweighs the taste for rigor in our industry. It's very sad.

The only AI bloggers who don't have something to sell seems to be simonw, the flask guy, and this redis guy. Any other blog recommendations from HN?

Can you suggest a best place to learn CSS in-depth, from first principles? (as opposed to, say, simple tutorials)

Didn't know about starship and need to check it out.

Shameless plug: I wrote a detailed git prompt in C which is similar to posh-git on Powershell: https://github.com/mahesh-hegde/promptsynth

> LLMs are eval(). Skills are programs. YAML is the motherboard.

This for some reason irritated me so much I wrote the comment.

An LLM could never write this.

> Bruce Schneier intercepts all your internal monologues by a man-in-the-middle attack.

Holy slop!

You can probably phrase this question as

* Can we enable the only features we want by toggling chat.disableAIfeatures and only selectively enabling copilot and chat?