koehr

It's funny to me that they use Claude Code as advertisement for their coding models. This software is mediocre at best, full of useless bloat and really only good enough for the use case (unless, the use case is advertising their models to vibe coders, then it's pretty good and I'm not the target audience)

I'm working on Solace¹, a programming language that targets JavaScript, but is not a superset, like Typescript. The idea is, to allow zero-runtime-cost safeguards, like Optional Values and Error Unions (syntactically they look like Zig's), decent pattern matching and Traits. There's still a lot of work to do, though, because some of the approaches I went for, don't hold in all cases (generics are tricky).

1) https://git.koehr.ing/n/solace

So the longest subway ride in NYC is 24km (15 miles) and 1h long? We have multiple underground lines farther than this in Berlin with the longest being 32km (20 miles). But don't get fooled, the issues are the same.

This reads more like an advertisement for Mythos, on the first glance

I just tried it. It downloaded Qwen3.5 2B on my phone and it's pretty coherent in its sentences, but really annoying with the amount of Ente products mentioned in every occasion. Other than that it's fast enough to talk to and definitely an easy way to run a model locally on your phone.

This is the wrong argument. Claiming that Mozilla is doing it wrong because the technology purist part of their userbase decided they don't want AI is simply short-sighted. The kill switch is the best option, because it let's Firefox be like a typical user would expect, while still giving the option to deactivate things. Deactivate by default and the typical user feels patronized.

For me this really speaks for intermediate text formats like Markdown, that are easy to read and render, while covering most formatting needs.

Historically, it was a rather famous page. It's a watchblog by the German security expert Felix Von Leitner. It exists since 2005, but since May this year, he didn't write anything until one new entry that you can see here: http://blog.fefe.de/?mon=202512

They replied in the meantime and removed the domain from their threat database.

Interesting, I just downloaded the big list and couldn't find it there.

Looks like https://temp.sh/ is blocked as well.

Including a CLI client, an NPM package and some spec refinements, working towards web page generation.

Plaintext Casa - now with working links on its website! Thanks for letting me know!

A decentralized social network that uses plain text files over HTTP. No algorithms. No ads. No data harvesting. But not just microblogging. With rich about sections and multipage feeds, it can be a full blogging platform. Write short posts, long essays, organize by topics. All in plain text.

It's interesting how this sounds like a cutting edge experiment, while this is a common thing to see in Germany and other European states for quite a while now.

I think the syntax is really pretty!

I started playing around with the idea of a new programming language that compiles to JavaScript. Despite syntactic improvements and type-checking, it's goal is to streamline promises, async/await and callbacks into one signal based system I call live variables. Additionally, those could be compiled via different, interchangeable compiler plugins to framework specific code, like React's useState or Vuejs' Refs, to make it more future prove. It's still in a concept stage and maybe the whole project is futile, but if it piqued your interest, check out https://github.com/nkoehring/Solace

Great to see that there is work happening in this direction. I wonder what the difference to usual public-key authentication is.

Have you tried matterrank.ai? It might give you want you want, but as a search engine, instead of a chat interface.

The misconception here is that the internet is not decentralized. The original idea of the internet is to be decentralized. There are just very large centralized silos built on top of its structure.

Funny, even ironic, that the thread now starts with someone saying "I use Arch". Good meme food.

Wow, this is exactly the "solution" that all the tracking companies love to hear from the people.

The EU can still, in theory, sue them because they're serving Europeans. Especially in the beginning many companies became afraid of the possibility so they simply blocked access to see where it goes. Then it probably became clear the European customers are not worth the effort to change back. But actually it's still illegal what they're doing because the GDPR also states that customers have to be treated neutral regardless of their location, as long as it's not about licensing of course.

Honestly, melting two open source editors together (or rather cloning one while integrating the other) and ask for that amount of money for it is in my opinion unacceptable. The only thing that would make this a bit more acceptable is when the developers are promising to spend a lot of their earnings for vim development (and maybe VSCode, too).

*edit:* Just saw that it is actually open source and will be available for non-commercial use for free. I'm more than fine with that model!

I fail to see any mention of or link to the detailed requirements talked about. To save y'all from yet another search engine visit, they are basically Intel 8th Gen Core upwards and comparable AMD CPUs. Here you can find the lists: https://docs.microsoft.com/en-us/windows-hardware/design/min...

After Bradley Taunt published 1MB.club, a good discussion sparked ( https://news.ycombinator.com/item?id=25151773 ).

One big opinion that I definitely share is that 1MB is actually pretty generous, even excessively generous. So I went for a more reasonable number that allows for fast loading everywhere but also some complexity. Another point raised was that not only the size should matter but also the amount of content. 200kb of pure text are probably more valuable than a single word shown with a 200kb webfont.

The page analysis and insertion are (semi-)automated. As currently visible on 1MB.club, the manual process is just not feasable. The pages are downloaded and analysed with Phantomas. They are also rechecked after a week.