I don't care moving Nim0 repository from GitLab to GitHub if it can help collaboration on that project. I can even share it with another maintainer if someone is interested. For the current time, I've created a clone at <https://github.com/pmetras/nim0>.
But I don't think that this project is worth adding an AST. Wirth created Oberon-0 as an illustration for teaching CS and compiler construction. He wanted to keep it as simple as possible, though touching some important parts of compiler design like records or arrays. Nim0 is a direct port of Oberon-0 compiler with a few enhancements. I kept Wirth's coding style in some parts even if using Nim could make the code nicer or I thought that more modern coding would improve it. It's really a pedagogical toy to learn compilation while reading the book _Compiler Construction_ and trying to do some exercises at the end of chapters. If I wanted to have a more powerful bare compiler with an intermediate representation, I would start from scratch. Nim0 was thought to be a single-pass in-memory compiler with RISC code interpretation (Wirth even provided an FPGA hardware to execute programs). Adding an AST would require changing the whole code generation module and make the whole program more complex for beginners. As it is, it limits the type of language experiments you can do with it. But the goal of Nim0 is not to replace Nim compiler but to make it easier to understand how it works.
