Matrix 2.0 fr fr

New login

Matrix supports OAuth, but it required some additional setup. I chose Kanidm, as I am a proud Rustacean. It doesn't have a lot of features, but it gets the job done. I was considering other options, but my current setup benefits from minimalist software.

For now, registration is closed, but I will open it soon. I have some UI-related things to think about, as well as making sure everything is secure.

You will be able to create an account on AYEQUE.art and use it to log in to Matrix without any additional passwords. Later, I will work on giving people the opportunity to create posts, comment, and create some groups and communities.

Is it sane?

Maybe not really, but it's what? Fun. I think that Matrix could power most of the posting, messaging, and other interactions between people.

I just like the idea of using someone else's code. Sometimes you code something to learn instead of using an existing project. Sometimes you just think that you are better and can make a better solution.

I do a lot of stuff to understand and explore new topics. At the same time, I would be so happy to build on top of something that exists and is already used by people. I understand the urge to implement everything yourself, but often it is toxic.

So what's the difference?

I wouldn't try to develop my OS kernel to run it instead of something well supported and maintained, such as Linux. I would create a desktop environment, as it's what people interact with and, to some extent, it is even more important than what your distribution is.

I would create my own graphics API, but I wouldn't try to recreate something like winit or SDL, or anything else that handles low-level, OS-specific window interactions. Of course, I would examine a tool, and I would need to do that anyway if I interact with it when it comes to Vulkan development.

I would create a game engine but probably I would try to avoid the implementation of all the tools myself. Maybe try to integrate it into an existing IDE, like Godot IDE (why not?), Zed or make nvim native? For context, my game is in Godot and I have no active plans to make games in a custom game engine for now.

What's the next move?

Well, agentLIBRE was left unattended for some time, so it will receive some more attention soon. I don't wanna brag, but it has already started improving itself.

We will see how Qwen 3.8 27B will perform, but for now my Gemma4 QAT models can interact with frontier models, where something like GPT 5.6 Sol reviews and creates specs, and your local model implements them.

Latency is probably bad if we compare it to some GPT 5.6 Luna as a cheap and fast implementer. I am also considering DeepSeek 4 Flash, but for now OpenAI subsidizes my tokens, so I stay with them. We will let the market decide.

Digital warfare

I really like the fact that there are many people who do the same things I do -- maintain a personal website and blog. It seems like a lot of people have the same idea.

The idea is that we might be really close to losing our rights when it comes to privacy. Some platforms used to be fun, but now almost every platform wants your government-issued document, etc.

Maybe one day I will start "harassing" people by texting and emailing them to try to create something that everyone can use and be in the same environment.

For example, you create a website with any kind of frontend. You want some blogging functionality? Imagine there will be a framework that you just add to your project and know that people can text, comment, etc.

Later, it can be something else, like a payment system, a kanban board, or anything you can come up with. This way, everything can be as decentralized as makes sense, while everyone can keep their autonomy.

I think that a future like that is extremely beautiful, and it is something we should strive for. No more AI-generated videos of cigarette family being homeless in my feed.