Hi,
Thanks for bringing this project so far. Every day LLMs are getting better, but the old metin2 deployment ways are not that ideal for full-llm usage. Now thanks to you I cloned the repo and let GPT 6 Astra work on it. I'm a senior backend developer but there is no way that I can analyze the architecture and make decisions that fast.
It's been working on it like 20 hours and these are what's been done;
- Swtiched from MySQL 5.x to 9.7
- Upgraded python to 3.13
- Now using GCC15/C++20, Go1.27.1
- Now using TLS 1.3 for auth & some requests
- Upgraded LUA to 5.5
- Now using categorized user packet rate limits instead of one. (I guess now the game will not kick me because I tried to use some stuff too fast, )
- Discovered that the core server is mainly stores sessions, affects and other stuff just based on user id, there is no other key like session etc. Fixed that.
- and so much more that I don't understand.
The no brainer goals of mine;
- Upgrade to DX11 and make the client x64
- After successful upgrade to MySQL 9.x, consider PostgreSQL and its extensions if there's an optimization possible
- Make both server side and client side completely modern CI-CD compatible, if possible, write unit - integration tests for each critical function
- Detect chinese - ambigious variable names and make everything readable
- Make a web admin dashboard that we can monitor the server and stats, while we can manage all game functions and files.
Wish me luck and lots of ChatGPT usage reset gifts, maybe one day I will share one-click deployable metin2 server repository with enterprise level architecture. I know lots of private servers achieved great things for years, and their developers will fold me in any matter. But I've played every version of this game like 15 years, and I'm really bored to play with the same client bugs, screen errors over and over.
