Examples of what I mean by modding:

  • minecraft mods: add some jar file into your mod folder
  • skyrim mods: add some .esp file into your mod folder
  • luanti: put some folder with .lua files and config into your .minetest/mods folder

Mods are basically “turing-complete” and can add different types of computation to your application, while integrating with the (GUI) system.

How to design a program that allow for modding?

With interpreted programming languages like python or lua, you can load code as strings at runtime … but is it done that way for these languages (that would be really bad for security)?

eval("print('mod loading ...')")

So roughly how I imagine modding in python to work, a demo in the python repl …

>>> items = {}
>>> newmod = """
... {"name": "holy-mod-sword", "value": 10, "do-damage": lambda x: x.firedamage(1000)}
... """
# loading the mod
>>> items["holy-mod-sword"] = eval(newmod)
>>> items
{'holy-mod-sword': {'name': 'holy-mod-sword', 'value': 10, 'do-damage': <function <lambda> at 0x7f8c710a9d00>}}

is it done that way or similar?

  • @ClemaX@lemm.ee
    16•6 days ago

    Eval is bad for security boundaries and the string based approach is a pain to develop and maintain. An alternative that is equally bad for security but better for development would be dynamic imports using importlib.

    If you want to support custom scripts while enforcing security boundaries, you could use an embeddable interpreter like lua, or create your own.