The list that crawls

A feed endpoint returns the most recent crumbs, each tagged with its author's name. The response is correct — right crumbs, right authors — but it's slow, and it reports two numbers that explain why: how many database queries it ran, and how many milliseconds it took. For a 40-item feed, the query count is in the forties. Ask for fewer items and the count drops in lockstep; ask for more and it climbs the same way. The number of database round trips is tracking the number of rows, one to one, plus one. That's the smell. A feed of N items should not cost N+1 trips to the database — it should cost a small, constant handful regardless of how long the list is. The list query is fine on its own. It's what happens *after* it — once per row — that turns a quick lookup into a crawl. Return exactly the same feed while keeping the query count flat as the list grows.