peheje

Vibes. But I was thoroughly disappointed by Sol 6.0 high in today's work session.

Available in play store on Android also. Cheap

one of those games where i wish i could erase my memory just to play it again, still think about it, i think it's like a year ago i played it, maybe more. idk. great game.

I think I'd just separate the two things:

``` Theme ( ) Light (•) Dark

[x] Follow system Dark — currently controlled by system ```

When "Follow system" is on, the theme choice is greyed out.

Feels simpler to me, and it scales better if themes ever become more than just light/dark.

Source

I'm using my unrooted stock s23u with damaged screen as a server but only to me thru tailscale.

Termux. Proot ubuntu. Serving OWUI and syncthing. Works fine. Had to remove pw protection so runit starts services at reboot. Therefore encrypted at rest any sensitive data

I Ssh in into it with ai harness to make changes. Made me not want to sell it after all.

Despite all the trends, I still really like HTML for structure, CSS for styling, and JavaScript for logic.

The boundaries aren’t perfectly clean, and that’s fine. But the separation gives you a useful way to think: structure this first, style it later, add behaviour where needed.

My experience with Compose—though I suspect SwiftUI people will recognise the feeling—is that I have to think about everything, everywhere, all the time.

Then we add some MVVM/UDF flavour. The "ViewModel" knows nothing about the view, despite usually serving exactly one screen. Add some "MutableStateFlow"s, combine them into "ScreenUiState", expose it as a "StateFlow", collect it with lifecycle awareness.

Beautifully decoupled. Extremely testable.

Then the organisation writes no unit tests and relies entirely on two-hour nightly screen tests.

A welcome-page refactor breaks the profile page.

“Didn’t you check the nightly build?”

No. It runs at night.

“Well, that’s your responsibility.”

But you broke it.

“Yes, but it’s your code.”

Then why did we ship it?

Fine. Schedule the postmortem with my mother.

And I’m not blaming mobile developers here. I’m frontend, backend, full-stack, I think AI engineer now. I have personally helped make simple things complicated across the entire stack.

What I like about the web is that one simple screen can be vanilla JavaScript. Another can use Vue. Another can use some specialised spreadsheet component.

People react to that with horror: what if components are duplicated, behave or look slightly different?

Fair concern. But that is real coupling with visible consequences and trade-offs. Somehow we have started treating coupling as something abstract that only exists inside code, rather than something that should produce an actual benefit when removed.

HTML. CSS. JavaScript. Or something close to it.

Maybe I’m getting old.

Shipping. Breaking stuff. Fixing it. And discovering, repeatedly, that the tool sounded much more confident than it had any right to.

Like always. This is why the average carpenter you hire in 2026 still won’t do a perfect job, despite carpentry having existed for thousands of years.

Might just resub. Will experiment with Luna next sessions. 5 h window is not working very well for me. But if I can drop down to Luna at 20-30 % left and comfortably ride out the wave then.. that might just work.

Whole thread mase me lol. Thanks guys.

I did this too. But honestly, I was getting a little tired of it.

Now I use AI to build things I genuinely like and actually use. Nothing particularly fancy.

I built a weather website that I use all the time when planning hikes. I also built a generative hiking-route tool: give it a starting point and a distance, and it tries to find interesting paths for you. I’ve made plenty of small helper tools for work and health, and now I’m wrestling with Open WebUI, trying to make it work exactly the way I want while repurposing an old Android device as the server.

Do I miss nerding out over implementing a B-tree, or learning async programming in F#, Go, Crystal, Nim, C++, C#, JavaScript workers, or Java green threads?

I don’t know, man. Maybe a little.

But it has also never been easier to say: “Hey AI, teach me this concept using five examples. Give me one task, then evaluate my solution. And teach it in this particular style.”

We lost something. I genuinely think we did.

But I also think we gained much more.

It’s okay to mourn what was lost, as long as you can still look forward.

What are some genuinely useful things macros can do?

Well.. awful compared to all others who definitely did not cheat right?

We need Linux on phones. Bank apps not needed as long as I can use browser. But do need some things like wireless cards, popular apps like Sonos and Spotify working.

This is a great read — really — a truly great one — it really is.

These are great

makes sense? commit messages is pre ai era, we need a new git, and people are already working on it

Reminds me of 'Learning to be me' by Greg Egan

Reason people want the best: people want to believe their project is so advanced that they need the most clever LLM possible. To say otherwise is to admit that it's not really frontier or novel in any way. And people don't like that.

I completely trust the government to keep dangerous AI away from bad guys. After all, they did a perfect job stopping illegal guns and drugs.

OpenRouter sometimes chooses a very expensive provider. Try the floor slug or choose directly the provider. I moved to just putting 5 dollars directly on deepseek instead of going through OR.

resistance is futile

there will be a lot of paper clips

I mean indsight is 20/20, but saying that is like saying "everyone will just use the best tools". That's not what we see most places in the world for most types of resources.

We try to keep things simple. Everything has risks. No stall, run async, backward compatible. DB handles rollback via transactions. Happy to expand if interested.

Deploying SQL changes? Why not just let the application do that on startup. Ofcourse be backward and forward compatible. SQL change only deploy.

"Update something in the cloud" <- What do you mean?

first impression: the tui flickers a lot, unpleasent. very laggy to write in.

having issues with truncated output from deepseek v4 pro through openrouter via pi-harness on ptyxis-terminal using ubuntu

trying reasonix with direct api..

Agreed. And the gained time either goes toward 1) more experiments, or 2) leisure, which makes you sharper in the lab and happier overall. Not sure the "I saved $17,000 so far" framing is the most useful way to look at it, but it's a cool project and I love that people are doing this kind of thing.

The future will have LLMs running local at your laptop/devices. If not almost exclusively then at least for 90-95% of the tasks. Speculative decoding is just one technique out of many existing and more to come that will make this even more viable. The gap is closing on both fronts. Software gets faster/more clever. Hardware gets faster and smaller. The single user story is the story. I'm obviously speculating myself, but that's how I see it.