Listing Collections¶

GET /tasks returns many records. A naive findMany() works for ten rows and falls over at ten million.

The Naive Version (and why it's a trap)¶

1const tasks = await prisma.task.findMany();   // ⚠️ returns the entire table
2return NextResponse.json({ data: tasks });

Problems as data grows:

• Huge payloads crush the network and the client.
• Memory spikes on the server.
• No way for the client to ask for "just page 2".

Always Constrain Three Things¶

A production list endpoint controls:

1. 1.Scope — filter to what the caller may see (where: { ownerId }).
2. 2.Volume — limit how many rows come back (take).
3. 3.Order — a deterministic sort (orderBy).

1const tasks = await prisma.task.findMany({
2  where: { ownerId: session.userId },
3  orderBy: { createdAt: "desc" },
4  take: 20,
5});

Return Useful Metadata¶

Clients need to know how to ask for more:

1const [data, total] = await Promise.all([
2  prisma.task.findMany({ where, orderBy, take: 20, skip }),
3  prisma.task.count({ where }),
4]);
5
6return NextResponse.json({
7  data,
8  meta: { total, pageSize: 20, page },
9});

A Default Page Size Is Mandatory¶

Never let a client request "all rows" by omitting a limit. Set a sensible default (e.g. 20) and a hard maximum (e.g. 100). Full pagination, filtering, sorting, and search get a whole chapter (Chapter 8).