Purpose of module grouping in Tortoise ORM

[N1qro]

What is the purpose of specifying multiple module names in Tortoise ORM? Every time I use this library I have to specify not only the list of file paths to discover models from but also I am required to group them by using dictionary and assigning name to it

async def main():
    await Tortoise.init(
        db_url="sqlite://db.sqlite",
        modules={"discord": ["database"]},  # <- here
        timezone="Europe/Moscow"
    )
    await Tortoise.generate_schemas(safe=True)
    ...

What are the cases where we would actually need to have this functionality in our project?
Personally, I cannot think of any

[agammann]

discord is a Tortoise app label, while database is the Python module where it discovers your models. They serve different purposes.

For example, if database.py contains Account, a foreign key to it uses the app label:

account = fields.ForeignKeyField("discord.Account", related_name="events")

One app can also collect models from several Python modules:

modules={"discord": ["database", "database_extra"]}

Multiple apps become useful when groups of models need different default database connections. The full config form lets each app specify its own models and default_connection; the setup documentation has that configuration example. I checked this with Tortoise ORM 1.1.8 and two SQLite connections: discord models used one database and reports models used the other.

For a small project with one database, a single label such as discord is sufficient. You can put every model module under that label; there is no requirement to invent more groups.