MuOxi is a modern library for creating online multiplayer text games (MU* family) using the powerful features offered by Rust; backed by Tokio and MongoDB,. It allows developers and coders to design and flesh out their worlds in a fast, safe, and reliable language. Explore MuOxi API the rustacean way Join us on discord.
The codebase is currently in alpha stage . Majority of development is done on the master
branch. There is a working TCP server that allows
for multiple connections and handles them accordingly. Effort is focused at the moment in
designing the backend database structure using a combination of MongoDB and json files.
Any contributions from the community is appreciated and wanted! No matter your skill level any sort
of effort into this project is extremely welcomed. For those wanting to contribute, fork the master
branch
and submit PR's. Any questions or information, we welcome you at our discord server. Come on by.
The bare minimum TODO features that must be implemented before I would consider it a bare mud game engine.
- Allows for multiple communication protocols (telnet, MCCP, websocket, etc)
- Allows for new player creation
- Asks for a name and password
- saves player info (etc. name, password)
- Implements some basic commands: quit, say, tell, shutdown
Handles players disconnecting or quitting- Implements a periodic message every n seconds
- Implements some rudimentary admin control (eg. muting another player)
- Basic cardinal movement
- Implements a storage based system
The database design is seperated into four different layers, with different levels of abstraction. MuOxi utilizes MongoDB for its storage needs and Redis for its caching and fast retrieval needs. A unique design approach has been taken that allows information to be kept safe from database corruption, brownouts, or blackouts. The ideology is as follows:
Layer 1: JSON Files <--------
| |
\ / |
Layer 2: MongoDB |
| |
\ / |
Layer 3: Cache/Memory |
| |
\ / |
Layer 4: MuOxi Applications--
The entire database actually lives in JSON files from accounts, mobs, players, equipment, spells, skills, etc...
JSON files where chosen because of its close relationship with MongoDB native storage choice, BSON; as well
as it's human friendly format. A seperate process called the watchdog monitors custom defined .json
files in the
/config
directory for any changes to contents themselves. Upon a detected change it triggers an upload piece of logic
that updates MongoDB, which leads us to layer 2 of the design.
This is where all persistent data will live throughout, and past, the life-span of MuOxi. MongoDB naturally stores data in a BSON format, and allows all the goodies that come with any database (indexing, search, upsert, insert, deletion) The database should always be a reflection of what is stored in the flat files, when MuOxi uses data from the database, it gets loaded and we move to layer 3 of the design.
This layer is dominated by Redis, for quick retrieval of information and adding ad-hoc non-persistent data such as combat, triggers, and other various information that would not be detrimental to the engine if the engine or database were to become corrupted and/or shutdown for whatever reason. At initial startup redis will cache the entire database into it's memory where then layer 4 will read from. If changes occur within the MongoDB itself, redis will recache.
This is the layer where MuOxi will actually use all persistent and non-persistent data to drive the actual engine itself. Whether it be handling different states of connected clients, combat data, player information, and any-and-all other memory will be read from the cached database to keep the engine running. Upon an action within MuOxi that would causes a change to the Database, MuOxi will actually write to the flat-files instead of directly to Mongo. This was a throughouly thought out process to keep MongoDB a read-only database, from the perspective of the engine itself. When a change occurs and MuOxi writes to the flat files we began again at layer 1 of the design. It is the responsibility of the WatchDog to monitor changes to the json files and update MongoDB. MongoDB and the JSON files should always be a reflection of each other.
The prototype idea of how the core design is laid out into three seperate objects.
- Staging/Proxy Server (Clients will connect to this server and essentially communicate with the engine via this stage)
- Game Engine (all the game logic lies here and reacts to input from connected clients)
- Database (stores information about entities, objects, and game data)
- Communication ( Each supported comm client (MCCP, telnet, websocket) will act as a full-duplex proxy that communicates with the Staging Server)
The idea is that players will connect via one of the supported communication protocols to the proxy server. In this server, clients are not actually connected to the game, unless they explicity enter. The staging area holds all connected client information such as player accounts, different characters for each player, and general settings. When a client acutally connects to the game itself the server acts as a proxy that relays information from players to the game engine, where the engine will then react to the players input. The engine and staging area will be seperated and communicate via a standard TCP server. The reason for this seperation, is to protect players from completely disconnecting from the game if changes to the game engine is made.
The support for multiple type of connections is a must. Therefore the following shows an example design layout that has the ability to handle multiple communication protocols. Each comm type will have a unique port that must be addressed and acts like a proxy to the main Staging Area.
------------
| Websocket | <---------------- \
------------ \
---------- --------------------- ---------------
| Telnet | ---------------------->|Proxy/Staging Area | <-- TCP --> | Game Engine |
---------- --------------------- ---------------
/
-------- /
| MCCP | <----------------------
--------
This design is still in prototype phase.
The MuOxi library is aimed at creating a very simplistic and robust library for developers to experiment and create online text adventure games. As it stands the engine has the following capabilities:
- Accepts multiple connections from players
- Maintains a list of connected players
- Hold shared states between connected clients
- Removes clients upon disconnection
The project contains two seperate bin that can both be evoked from the command line:
-
(Not working as intended at the moment) cargo run --bin muoxi_web
- Starts the websocket server listening for incoming webclients, default 8001
-
cargo run --bin muoxi_staging
- starts the main Proxy Staging server where all clients will live, this area is where clients will communicate to the game engine. Direct telnet clients can connect this is server via port 8000
-
cargo run --bin muoxi_watchdog
- starts the external process that monitors changes to configuration json files. Once a change has been detected it triggers an update protocol to update MongoDB
-
cargo run --bin muoxi_engine
- Starts the main game engine running in it's own seperate process. The whole game is contained within a TCP listening server that exchanges information back and forth between to the Proxy Server. Right now it is just an echo server
The concept around MuOxi is not just to recreate an existing MUD game engine in Rust, but rather to utilize the performance and safety that Rust has to offer. That being said, this future vision for MuOxi will change over time, but it needs to fulfill some features that I think will make this an outstanding project.
- The Core of MuOxi will be written in Rust, expanding the core will need Rust code
- The game logic, that handles how Mobs interact, expiermental mob AI integration, etc.. will be handled in Python.
- add more here