Loading strategies¶

Relationship loading is explicit. You choose the path and strategy.

Choose a strategy¶

Strategy	Helper	Use when
Joined	`joinedload("path")`	You want one SQL query with joins and the related row count is controlled.
Select-in	`selectinload("path")`	You want batched secondary queries for collections or larger graphs.
Lazy	`lazyload("path")`	You want to mark a path as loadable later through explicit APIs.
No load	`noload("path")`	You want to prevent a relationship path from loading.

Use loader options¶

from ormdantic import joinedload, selectinload

flavors = await db[Flavor].find_many(load=[joinedload("supplier")])

suppliers = await db[Supplier].find_many(
    load=[selectinload("flavors").sorted_by("name").batched(100)]
)

Nested paths use dot notation:

await db[Supplier].find_many(
    load=[
        selectinload("flavors"),
        selectinload("flavors.reviews"),
    ]
)

Use depth loading¶

depth loads relationship paths by graph distance:

await db[Supplier].find_many(depth=1)

Loader options are preferred for production code because they document the exact paths and strategies.

Load one relationship explicitly¶

supplier = await db[Supplier].find_one("s1")
flavors = await db.load(supplier, "flavors")

This still performs explicit async I/O. It is not hidden attribute access.