V2: The Last Version of Axe API

TL;DR

V2 doesn't bring any new features at all. I just upgraded dependencies and dropped support for old tech versions. This version will be the last major version of Axe API.

What happened?

I planned to create a REST API framework that helps develop APIs quickly with as little code as possible. That's why Axe API is a success — anyone can create a REST API quickly by writing very little code. I used this project for many different projects over the years. But it also has some cons that are killing the project, and let me explain them all.

1. The Framework Era Is Over

I should have written this framework 10 years earlier. The timing wasn't good enough, and AI just killed it. Even Tailwind has so many issues because of LLMs. Who needs a framework to write code anymore? When was the last time you visited a framework's documentation? Every framework on the market had a chance to explain their philosophy to developers via documentation. That's all gone long ago.

Writing a REST API without a framework is very easy if you use LLMs. Why do you need another layer of it? It doesn't make sense anymore. I believe that not just frameworks, but most open-source projects could die at some point — but that is another discussion.

2. Axe API Has Too Much Magic

Everybody said good things about Axe API. It was a magical tool, because you could do many things by writing very little code. But it's just too much magic. A developer who knows the system (like me, for example) could create APIs very quickly. But beginners had so many issues, because it's not easy to understand how it works. Since I'm not Facebook or Google, very few paid enough attention.

With that said, I had other plans to implement in V2. Instead of using magic, I was hoping to bring some new features that could help us use the same magical power in a more declarative style of coding. But since LLMs arrived, I couldn't find enough motivation to build it.

3. Community Support Issue

I didn't actually believe I could build a community around the framework. Yet I did everything I could. But here is the sad truth about my experience — somehow, I couldn't make people believe in how valuable the project was. Maybe it was because of me; I wasn't good enough at convincing people. Or maybe the project wasn't that valuable. Either way, even though I tried everything I know, there is no real community behind the project.

I wouldn't have cared too much about the community issue if LLMs hadn't arrived, because I could have been the first and last consumer of the framework myself.

4. We're Just LLMs' Training Data

I didn't like it. I wrote a blog post about it, Bye, Bye GitHub, on my personal blog.

Everything we create in the open-source world is just another training dataset that companies will use without our consent. As a result, they will sell the LLM service they created using our data back to us. So why? What's the point of all of this?

Goodbye

I started this project more than 5 years ago. I liked it a lot. It helped me learn many things, and I wrote this project from scratch WITHOUT USING LLMs!

Anyway...

I've just upgraded critical versions of dependencies, so it will buy you some time if anyone out there is still using this framework in any of your projects.

Thank you very much for your support, if you ever provided any.

Farewell.