christophilus

Competition is here. I personally prefer Codex. Opencode with a variety of Chinese models is also just fine for 80% of my use cases.

Agreed completely. It’s my favorite language in its class. I need to build something serious in it someday. So far, I’ve only used it for building personal tools.

Niri is a perfect window manager approach, in my opinion. I don’t need tmux. Just Niri + Foot (terminal) and it’s perfect for the way my brain works. If you haven’t tried it, I highly recommended it.

> alarming

Literally. It’s the windows visa-ification of cars. I hate it.

I doubt the thousands of lines per day claim unless the lines are trivial. It’s quite fatiguing to read hundreds.

Hm. I must be holding it wrong. I hit $20 easily in 5 hours with deepseek in opencode.

Honestly, I hate that Bun is vibe-coded and seems amateurish in many ways, but it’s still an excellent tool.

> static electricity typing

I assume that was an autocomplete typo?

As in inverse correlation? The best programmers I know have barebones text-based sites.

ex-Elastic here, too. It was a great place to work pre-IPO. It seems the culture has shifted a lot since the IPO, though.

I've been working with Deepseek V4 Flash (with opencode as the harness). It's been almost indistinguishable from Codex / Claude Code for me. I'm sure I'll run into problems when I get to a stickier ticket to tackle. But so far, it's been quite good, and I find it writes straightforward code.

I do think the Chinese models are good enough for an 80/20 rule use case.

I agree with that. You can use Postgres as a message queue / task manager backing store without a database function, though, and it works quite well at the small scale that most sites / SaaS products operate at.

Hm. I’ve been way, way underselling myself this whole time.

Woah. I missed that one. Had to look it up. That’s one hell of a surprise bill.

To me, the justified text makes it an effortful read.

Well, everyone's bashing on OpenAI as well they should, but just a reminder, unlike Claude Code, Codex is officially available to customize here: https://github.com/openai/codex

It's fairly easy to patch.

I don’t trust any agent to respect any boundaries. They might today. But tomorrow’s vibe coded slip update might break it in subtle ways.

My solution to this is to only run agents in a sandbox of my own making (a locked down Podman container).

I find Claude Code nearly unusable. I always have to type in neovim if I’m typing anything more than a few words.

Codex is much better, which is to say, it’s only pretty bad.

Dunno about anthrax, but fully automated weapons are really easy to come by.

You buy it with financing in that case. And you fit your reductions and hires into the same fiscal year.

To be fair, I set up our CORS policy 6-7 years ago. I don’t remember the details anymore. I’ve been busy building product. So, this interview question will make me stumble, even though I’m quite senior. I think asking people about things that are set it and forget it once per project is probably not the best way to screen for relevant knowledge.

I inherited this one. My preferred SPA stack these days involves Porsager’s Postgres library, a simple RPC stack with Zod schemas, and Preact.

Even better is an old school MPA with progressive enhancement.

Yes. I’m dealing with a graphql, urql, Next, Prisma stack at the moment. Something that would be a handful of lines of code in a different stack ends up being hundreds in this one.

The Node ecosystem is full of wrong abstractions.

It seems like we have enough satellites in space that are giving off enough unique signals that we could put together a device that solves the GPS problem without needing GPS itself. I wonder what it would take.

Well, it does seem that memories may be embedded in the nervous system as well as the brain, so I don’t think the OP is wrong. You sometimes hear of heart transplant patients having other people’s memories / preferences. So, it’s not good evidence, but it’s a possibility.

It will be interesting to see if it’s ever completed. I imagine dealing with ship fumes in a tunnel will be a challenge. And, I’m curious what the plan is when a ship runs into the sides or breaks down halfway through.

Super cool project nonetheless. I hope it succeeds.

“They reinvented Erlang OTP.” - Reddit

“Incredible, a mere 35 billion cycles to start up!”

The comments on that video are perfect.

Well, I really don’t like Google, but if they make this a thing, I’d switch to Android and put Graphene on it or whatever just so I could tie into this. This is an excellent idea.