What's the way to use `assertSnapshot` in async test?

[valeriyvan]

Please look below example of async test. Test with XCTAssertEqual works as expected but same test with assertSnapshot crashes.

What's the way to use assertSnapshot in async test?

struct AsyncSequenceExample: AsyncSequence {
    typealias Element = Int

    func makeAsyncIterator() -> AsyncIterator {
        print()
        return AsyncIterator()
    }

    struct AsyncIterator: AsyncIteratorProtocol {
        var current = 1

        mutating func next() async throws -> Element? {
            print("enter")
            try await Task.sleep(nanoseconds: 100_000_000) // Simulate an asynchronous operation
            defer { current += 1 }
            guard Int.random(in: 0..<100_000_000) > 1_000_000 else {
                return nil
            }
            print(current)
            return current
        }
    }
}

class MyTests: XCTestCase {
    func testAsyncSequence() async throws {
        let asyncSequence = AsyncSequenceExample()

        var result: [Int] = []
        do {
            for try await element in asyncSequence {
                result.append(element)
            }
        } catch {
            XCTFail("Unexpected error: \(error)")
        }

        // This test works as expected
        // XCTAssertEqual(result, [1, 2, 3])

        // But this crashes
        assertSnapshot(of: result.description, as: .lines)

    }
}

[dwroth]

The problem is the use of XCTWaiter.wait inside of verifySnapshot. Using the synchronous wait function in an asynchronous context is a known issue (that's why Apple introduced await fulfillment(of:).

[tejassharma96]

Is this crash Terminating app due to uncaught exception 'NSInternalInconsistencyException', reason: 'Current context must not be nil'? If so, I am hitting that issue too in an async test :/ is there any workaround that we can use until the library can be updated to support some sort of await assertSnapshot method?

[tejassharma96]

Actually, I did find that marking all the async tests as @MainActor works, so there's that... still frustrating but I suppose it works now :/

[stephencelis]

Tracking with #978.