Hacker News: wrs

New comment by wrs in "I hate soldering"

wrs — Tue, 12 May 2026 17:40:28 +0000

I understand the skepticism, but it's really just a simple application of negative feedback. There's a 200W-ish variable power supply, a thermocouple right at the tip, and a software PID controller. So it's no mystery why it works so well.

Live demonstration: https://youtu.be/BepVEStPmJk

New comment by wrs in "Fake building: Claude wrote 3k lines instead of import pywikibot"

wrs — Tue, 12 May 2026 02:56:56 +0000

Hmm, I don’t see this much anymore. I typically start a project in plan mode and tell Claude to do some research to bring me 2-3 alternatives. Then we talk about the pros and cons before deciding on the libraries, etc.

On the other hand, if you just tell it to do a thing, I could believe that it would just do the thing. It is pretty bad at high level design judgment. Human guidance on architecture choices results in much better output.

New comment by wrs in "I hate soldering"

wrs — Mon, 11 May 2026 21:56:55 +0000

The new breed of irons with temperature measurement built into the tip (invented by JBC, cloned by Geeboon and similar) is amazing. The tip heats to exactly the temp you want in 3 seconds, then cools down to avoid damage when you put it back in the stand. As you solder, the power is automatically controlled to keep the tip at the specified temp regardless of the load you put on it. I never thought I'd replace the Weller station I've used for 20 years, but I'm glad I did.

Edit: For a specific recommendation, look for the Geeboon TC22 on AliExpress or Amazon. Don’t forget the tips, you may need to get them separately.

New comment by wrs in "CUDA-oxide: Nvidia's official Rust to CUDA compiler"

wrs — Mon, 11 May 2026 18:02:27 +0000

This is explained in some detail in the docs. There is a safe layer, a mostly safe layer, and an unsafe layer. Some clunkiness is needed for safe-yet-parallel work that they couldn’t easily fit into the Rust Send/Sync model.

New comment by wrs in "SpaceX wants to launch a million satellites"

wrs — Sun, 10 May 2026 21:35:08 +0000

The people who want to put data centers in orbit must be either much smarter than me, or much dumber than me, because I just don't get how that makes any technical or economic sense.

Of course, it's possible nobody actually wants to do this, they just want to get funded to do it. (Old joke: "I wish I had enough money to buy an elephant...")

New comment by wrs in "I'm writing a history of Visual Basic, Chapter 1 is up"

wrs — Sun, 10 May 2026 01:52:00 +0000

My first impression was that the text style is dismayingly like either an AI wrote most of it or (to be charitable) the author’s writing has been heavily influenced by the current generation of LLM output. So the image style goes perfectly with it.

New comment by wrs in "US Government releases first batch of UAP documents and videos"

wrs — Fri, 08 May 2026 21:43:18 +0000

"war.gov" -- give me a break. Are they going to try to executive-order a .war TLD to replace .mil next?

New comment by wrs in "Train Your Own LLM from Scratch"

wrs — Tue, 05 May 2026 16:14:58 +0000

Point taken, but I think to most ML folks, PyTorch basically is the stdlib.

New comment by wrs in "Async Rust never left the MVP state"

wrs — Tue, 05 May 2026 16:11:06 +0000

Also that you have to rewrite the entire standard library, because the kernel knows how to suspend kernel threads on syscalls, but not green threads. (Go and Java already had to do this anyway, of course.)

New comment by wrs in "Mouse Pointer as a Mere Mortal"

wrs — Tue, 05 May 2026 16:04:03 +0000

It’s an established thing in CAD programs (since a time before there were any personal computers) to move the cursor for you sometimes. I seem to recall this is one of the points of contention between KiCad and Wayland at the moment.

New comment by wrs in "Zig → Rust porting guide"

wrs — Tue, 05 May 2026 03:09:33 +0000

It was and is a perfectly good term, but people started using it without regard for its definition. I don't know why people wouldn't misuse a "better" term the same way.

New comment by wrs in "Tesla owner won $10k in court for Tesla's FSD lies. Tesla is still fighting him"

wrs — Sun, 03 May 2026 01:26:56 +0000

That idea of a simultaneous small claims day is brilliant. I hope somebody is vibecoding that site up right now.

New comment by wrs in "Waymo Drives Off with South Bay Man's Luggage"

wrs — Sun, 03 May 2026 01:22:43 +0000

Why couldn’t they just stick the suitcase in a Waymo and send it to him?

New comment by wrs in "When Dawkins met Claude – Could this AI be conscious?"

wrs — Sun, 03 May 2026 01:19:30 +0000

Not simply by reading every word ever written by a conscious being and learning to reproduce them with high probability.

At least, that’s certainly not how I got here.

New comment by wrs in "Rise of the Forward Deployed Engineer"

wrs — Wed, 29 Apr 2026 18:18:25 +0000

Of course, Palantir's sudden jump in stock price since November 2024 could also be explained by entirely different factors that have nothing to do with engineering.

New comment by wrs in "Claude.ai unavailable and elevated errors on the API"

wrs — Tue, 28 Apr 2026 18:44:46 +0000

On the other hand, the status page is blaming the authentication system, which one would think is not a frontier-class problem.

New comment by wrs in "Integrated by Design"

wrs — Tue, 28 Apr 2026 18:11:32 +0000

> For the amount you’re charging, you could have hired a human translator.

Citation needed.

New comment by wrs in "Vibe Coding Will Break Your Company"

wrs — Tue, 28 Apr 2026 16:47:50 +0000

The point of the article is exactly that you shouldn't hand over software development to people with no expertise in software development. So he sounds smarter and more humble than a lot of CEOs at the moment.

New comment by wrs in "Microsoft and OpenAI end their exclusive and revenue-sharing deal"

wrs — Mon, 27 Apr 2026 15:55:54 +0000

It’s a system that generates $100 billion in profit. [0]

[0] https://techcrunch.com/2024/12/26/microsoft-and-openai-have-...

New comment by wrs in "Could a Claude Code routine watch my finances?"

wrs — Sat, 25 Apr 2026 03:28:42 +0000

Yes, Plaid clearly has different levels of integration with different banks.

When I connect something to Chase with Plaid it is clearly a cooperative system with an OAuth-like permission dialog, and the Chase side even mentions they're tokenizing the account numbers so Plaid can't see them.

When I connect to the little bank down the street I just get a username/password dialog. Their web banking system is so primitive I'm pretty sure Plaid is just scraping it. When they introduced 2FA, Plaid became quite flaky.