Pagination

Returning GET /users with 10 million users in one response would be slow, memory-hungry, and mostly useless to a client anyway. Pagination returns a bounded SLICE of the collection per request:

def paginate(items, page, page_size):
    start = (page - 1) * page_size
    return items[start : start + page_size]

users = list(range(1, 26))   # imagine 25 user IDs
print(paginate(users, 1, 10))   # first 10
print(paginate(users, 3, 10))   # the remaining 5

Clients (and pagination UI — "page 3 of 12") often need to know the TOTAL page count upfront, which requires rounding UP — a collection of 10 items at page size 3 needs 4 pages (3+3+3+1), not 10/3 = 3.33 truncated down to 3 (that would silently drop the last item):

def total_pages(total_items, page_size):
    return -(-total_items // page_size)   # ceiling division trick

print(total_pages(10, 3))   # 4

-(-a // b) is a common Python idiom for ceiling division using only integer floor division (//) — negating, floor-dividing, then negating again flips a floor into a ceiling. (math.ceil(a / b) works too, but involves a float conversion this integer-only trick avoids.)