Use async/await for one-shot asynchronous work with a single result (a network call, a file read); use Combine (or AsyncSequence) for streams of values over time — UI events, debounced search text, or continuous updates. They solve different shapes of problem, and modern code often mixes them.
Task cancels its children).map, debounce, combineLatest), with backpressure and declarative composition.// Combine: a live, transformed stream — perfect for search-as-you-type
searchText
.debounce(for: .milliseconds(300), scheduler: RunLoop.main) // wait for pause
.removeDuplicates()
.map { query in APIClient.search(query) } // -> Publisher
.switchToLatest() // cancel stale requests
.receive(on: RunLoop.main)
.assign(to: &$results)
// async/await: a single fetch — linear and readable
func loadDetail(id: Int) async throws -> Detail {
try await api.fetchDetail(id) // one value, then done
}
You rarely need Combine's full machinery anymore. AsyncSequence and for await cover many streaming cases natively, and you can bridge: publisher.values exposes a Combine publisher as an AsyncSequence.
for await value in publisher.values { // consume Combine stream with async/await
process(value)
}
debounce + combineLatest + switchToLatest), or when integrating with @Published and existing Combine code.AnyCancellable lifetime management.Picking the wrong tool leads to Combine pipelines for trivial one-shot calls (overengineering) or hand-rolled state machines where a debounce operator would do. Senior interviewers want to see you match the shape of the data — single value vs stream — to the right abstraction, and know async/await is now the default.
A library of IT interview questions with detailed answers — from Junior to Senior.
Donate