mathieudombrock

Gemini protocol exists and is doing exactly that!

https://geminiprotocol.net

Was Terry genuinely racist or did he just like spewing garbage to be edgy?

Qwen 3.6 27b has been the sweet spot for me in terms of local models. I've had good luck using it with Pi harness. Looking forward to this.

What's the intended use case? Porting from framework A->B?

Does it handle responsive design? Keeping the same breakpoints etc?

It seems to me like the common pattern these days is to ask for notification permissions instead.

What I don't understand about the "RSS can't support ads" argument is that most of my RSS feeds don't actually let me read the article. Just a headline and a snippet at most. So if I want to actually read the article I'm going to have to load their page.

I've had deepseek-4v-flash work around my intentional sandboxing/restrictions to do things I explicitly forbid it from doing. I'm honestly not impressed here.

What model is that?

I use opencode as my harness which is pretty ok. I daily drive it with no issues. It's not going to do any orchestration for you but it works.

I'm slowly moving to Pi but it's a very DIY system. Think neovim for agents. I think it has the potential to be where I end up for a while but it's a long road.

That's what I'm not getting about these kinds of posts. What is the point of sharing this? It's just a bunch of nothing.

Why would you choose to have the ai use a language you don't understand? Isn't this basically admitting you had nothing to do with this project and anyone else could pay an ai to make the same thing easily?

Is this something you expect other people to use?

Are you planning to maintain this?

Are you making a point about ai capabilities?

Is this just a joke?

I guess I don't really understand the point of posts like this.

A major difference is that _someone_ knew what was going on (compiler devs).

The same input twice is only nondeterministic if you don't control the seed.

This includes open and closed models ranked by popularity and other metrics.

https://openrouter.ai/rankings

This includes open and closed models ranked by popularity and other metrics.

https://openrouter.ai/rankings

The Amish are generally pacifist.

Ignoring the orientation constraint, one easy way to tell is that one has a flat line on the bottom and the other doesn't.

LLMs are determistic. Just like everything else computers are capable of doing.

Commercial front-ends just hide the random seed parameters.

Vim mode in vscode is not even close to emulating a real neovim setup.

CodeCompanion.nvim is a pretty nice plugin. I use that for quick stuff and opencode in the embedded terminal for larger tasks.

I think the best answer you're really going to get here is that it's cool and fun to learn and use old languages.

This is unfortunately loading very, very slowly for me.

I'm not OP but I've worked on JUCE plugins with React UIs for JUCE8 web view.

The UI load is pretty instantaneous.

Everything uses the native web view. This means that on macOS you get WebKit and on Windows you get Chrome (or sometimes IE).

So it's generally faster than Electron. In the project I worked on it was not noticeably slower than native UI.

Memory usage is really close to what you might expect from any other app running native web view.

Wren is super neat. I've written a few small games for TIC-80 using it. It's a really fun language to write.

I think Google has proven with their recent actions concerning android that they really can't be trusted with big, critical open source projects.

I tried this. It sounds good on paper but the LLM will just "forget" to use it's tools. Either it will decline to query the database and just make stuff up, or it will decline to update the database when it should. The further along the game play gets the more out of sync the game word gets from the external state. I'm sure there is a clever solution but I never found it.

Neat, it seems to me like Lua is a great language for text adventures.

I love text adventures. Collosal Cave Adventure and Zork are some of the coolest programs I've ever seen.

I've always wanted to try writing one and this article might have just inspired me to finally do that.

This is a neat idea and I wish it worked. I've spend hours and hours trying to get LLMs to be a "dungeon master" for text adventures. I've written a good amount custom code trying to facilitate this. Trying to force the LLM to keep it's story straight.

I'm pretty convinced that the current generation of LLMs is nowhere close to being capable of this. No matter how many context hacks you throw at it.

It inevitably derails and ruins the immersion.

Best of luck on this. If you can pull it off it would be really cool I think.

I found this article really interesting. This is pretty much exactly how I feel about LLM programming.

I really enjoy programming and like the author said, it's my hobby.

On some level I kind of resent the fact that I don't really get to do my hobby for work any more. It's something fundamentally different now.