We’ve explored how resource() handles loading, cancellation, and all six status states. But there’s one pattern left that separates a good app from a great one: optimistic updates.

Most applications handle two data states: “loading” and “loaded.” Maybe three if they remember errors. But in a real enterprise dashboard, there are six distinct states — and handling each explicitly is the difference between a polished product and a buggy one.

In the previous post, we declared our resource() with a params computation and a loader. The user types a query, and the resource fetches matching projects.

Welcome to Phase 3B — the deepest technical dive in our series so far. In Phase 3A, we pre-fetched data using resolvers. But resolvers are a routing tool. What about data that depends on user interaction — a search query, a filter, a date picker?