Compare with DIs in frameworks

Aiogram

Implicit VS explicit

Aiogram’s Dependency Injection makes injection in implicit way - they are created outside of handler context, so it is hard to tell where values came from. Also, to make it create only required values you need to do more work using flags, including validating & handling these flags all by yourself. The more your project grow, the more this abstraction leads to unclear, harder-to-debug code.

Let’s take a look:

from aiogram import Router, F
from aiogram.types import Message

from src.models import User


router = Router()

@router.message(F.text = "/whoami", flags={"requires": {User: "user"}})
async def on_whoami(message: Message, user: User):
    pass

This code shows problem in detail: I can’t tell where parameter user is created, neither is it created properly. Also, I can set any values into flags and get no error at bot start.

How this looks with FunDI:

from fundi import from_

from src.models import User
from src.dependencies import require_user

async def on_whoami(user: User = from_(require_user)):
    pass

FunDI makes it clearer. You can understand from where values came and where they are created. No flags, no implicit injection.

Yes, FunDI also supports implicit injection with scope. But also makes it much easier to debug because, it tells you which function from which module requires value that was not passed to scope.

FastAPI

Request dependent VS fully independent

FunDI is similar to FastAPI’s Dependency Injection(because it was inspired by FastAPI). It has only one problem - you can’t use it outside of Request scope. FunDI was created to cover this issue and successfully does this job.

Summary

Is	FunDI	Aiogram	FastAPI
Implicit	Optionally	Yes	Partially
Independent	Yes	No	No