ahknight

Because who hates themselves that much? It's the thing I touch and interact with. That's exactly the part that needs to be sturdy, smooth, and pretty. It's the facade to the beast at the other end.

Yeah, but how many years until 128GB+ is attainable by mere mortals again? My 2021 home server build was 64GB of RAM. My 2025 build was 32GB and zram. :/

I got used to the hub-and-spoke model at home (previously thin terminal, server-client, etc.). Big ole desktop/server and smaller devices that (ab)use it remotely. Roam around with a smaller computer/tablet/phone. Tailscale to bind it all together.

If your computing needs line up, it's a very serviceable approach.

Did you know you can create more than one email address?

Now there's an entry for Fable.

From the look of it, their ops team is having a really bad Friday.

They made the guardrails tighter. Tight enough to capture more false positive requests, now. Yaaaaay.

They won't. They're focusing on different areas for their models.

I can vouch for Nebula. Just follow their recommendation to not get the 1x service if you want to use the data for medical reasons. The more passes they make, the more accurate the result.

Downside: those files are HUUUUUGE. Have a good reason to do it before pulling one down and trying to work with it.

Streaming.

Now if only Ubiquiti could solve the problem where everything is always out of stock.

But the Internet nerds wish to blindly judge something they know nothing about so they can feel better with the assumption that they could have done better somehow. How will they be appeased if the document they will say they have read and understood (without having done either) is not available to point at? How, I ask?!

That was on my short list, but I liked the look of the Seiko SBTM339 more (though it's a JDP model so I had to import it). Citizen has a model in that range as well. Then the Seiko Astron line has both radio time and GPS models if you're willing to pay 5x as much (they do look so much nicer, though).

And I can confirm the radio time app works. (For those curious, it plays the time signal over the speaker and the faint EM from the speaker is more powerful locally than the original radio signal at distance, as well as landing on the correct frequencies.)

I love that there's not only the internals of a standard three-hand, but the automatic mechanism as well. If you're going pure mechanical, an automatic is the best way. As long as you actually move during the day the watch generally stays wound up (though a twist when you put it on is a good measure). There's a ton of great watches out there powered only by walking around, and it's fascinating to look at what they can do with that.

The mechanics of a six-hand are similar, using the mechanism that is described here for the date indicator.

Another fun device, though more from an EE POV, are the solar+radio/GPS versions of the same. Automatics can hold power for a few days and need walking around to wind. Solar needs light (any light, though sunlight is always best) and hold power for over a month. Many higher-end models can self-set over radio time or GPS signals as well.

When the world goes tits-up someday, both classes of watches will suddenly become essential, and are already essential for people who spend a good amount of time "unreachable" for work or pleasure.

Podman is on macOS, FWIW. Uses the existing container framework to run the machine already. Root-full or not.

"Build an Oracle clone in Rust. Make no mistakes."

86% AI generated text. Seriously, does nobody write anymore?

It's that last S in SaaS that builds the moat around companies like this. Even if you can make the Software, it's on you to provide the Service as well if you go that route. Quite a lot of companies simply don't want to self-host yet another thing and make their IT group even larger (or more burdened). It's rarely the Software or the Platform that's the selling point as much as the Service.

It's a good thing chronic amnesia is a feature at the moment.

That's amusing, and I think it's something different than it appears. The models always predict over the existing context. If it's full of a certain tone, then the responses will carry that tone. I've been bored before and start responding in a voice (say, generic honor-bound warrior slaughtering evasive bugs) and I've noticed that comments, variable names, and even documentation starts to carry that tone for the remainder of the session.

The next session sees all of that, calls it unprofessional, and asks to clean it up. At which point I may or may not start in iambic pentameter to see where that takes us.

Prompting is boring.

Kindness is that, yes. Fundamentally, though, it's about being considerate in one's actions so as to not harm others. If someone truly believes that acting a certain way at any point risks their ability to reliably be kind in others, then it's a social kindness to be kind and considerate in all actions.

I'll not reach for the easy response and say "Be kind to the Earth" fails your definition without reaching for pedantry with "the Earth has living things" because the Earth is instead a wet rock that cannot understand kindness, yet we show it.

It's fairly straightforward. Create the source files. Iterate the linter checks until it passes. Finalize.

I'd bet they just spell it out at each step so there's something to point to when they do those final reviews and reject it. "Now go look at step 385 and try again."

The funny thing is that you can use both a style template and content template with the DOCX exporter of pandoc and _export_ custom styles. You just can't import them.

Works great on my MD -> DOCX/PDF/HTML workflows but not the inverse, alas.

If you use Markdown, then it almost doesn't matter because your converted output won't have a line break until you insert a single newline.

This is where I'm at right now and it works rather well. I have a glue script that automates some shorthand in the Markdown into LaTeX for PDF and pre-set styles for DOCX and HTML exports. It works rather well, and being able to tweak my own LaTeX shell for the content is quite nice as well.

It's ePub that gets the weak treatment right now. Which just has me wondering if I can have my tool output the input files this standardebook toolkit wants (probably) and get a reasonable output from it.

This is the question nobody's asking.

"Vibe coding" is a term that means "Prompt the LLM with a request for a project and wait for it to finish". If you're reading and understanding anything it made, that's not vibe coding, that's agentic development.

I also have a 64GB M1 Max and am similarly impressed with what that workhorse can do. The M5 tempted me -- a lot -- but then I looked at what I was already getting done on that machine and just couldn't justify it ... yet. Someday, surely, but not yet. Gemma4 gave all my local projects new life, just like what you did here.

Great job. Long live the M1 Max!

I'm not normally one to share videos as answers, but this particular fellow does a LOT of work with local AIs and Macs and happens to have a nuanced answer. https://youtu.be/XGe7ldwFLSE

Sonnet-level performance at Haiku prices. They know what they have and who the audience is they want.

... and you get what you pay for. Or less.