Load related data and paginate without repeats

In this lesson you will load related data and paginate without repeats. Work with the Equipment Rental API in the downloadable lab. The reference is a complete solution with separate lesson checks, so you can inspect the answer, make a deliberate local change and verify its behavior.

Reading a list needs both deliberate loading and stable pagination. related_rentals uses selectinload to fetch related equipment in an additional bounded query. Without explicit loading, reading equipment on each rental could issue one extra query per row. The reference makes such accidental lazy loading fail instead of quietly hiding the cost. The lesson counts queries while reading related names.

Cursor pagination needs a total order. created_at alone is not unique: multiple rentals can share a timestamp. Order by (created_at, id), encode both values, and use a strict greater-than comparison on the next page. Apply the owner's scope before the cursor and limit, so forbidden records cannot influence a learner's page. The cursor is a position, not an authorization token; every request still applies current scope. This local example does not promise a snapshot of a dataset changing between requests. State that limitation rather than claiming all concurrent inserts are frozen. A returned next cursor may lead to an empty final page when a page is exactly full.

Worked source: equipment/services.py, page_statement.

Run lesson09. Three rows share one timestamp, yet the first two and final one do not overlap. The same test reads each equipment name and requires exactly two queries. Decode a sample cursor in your own local debugger and identify both ordering values. Do not treat it as a secret access grant.

def page_statement(owner_id, cursor=None, limit=10, *, inclusive=False):
    from sqlalchemy import tuple_

    query = select(Rental).where(Rental.owner_id == owner_id)
    if cursor:
        boundary = cursor_values(cursor)
        pair = tuple_(Rental.created_at, Rental.id)
        query = query.where(pair >= boundary if inclusive else pair > boundary)
    return query.order_by(Rental.created_at, Rental.id).limit(min(max(limit, 1), 100))

python run_checks.py -k lesson09