Where would using Record Publisher with Combine be useful?

Question

I am trying to understand Record Publisher.

Here is an example to better understand the case:

var subscriptions = [AnyCancellable]() let pub = Record<Int, Never>(output: [1, 101, 102, 1001, 1002], completion: .finished) // I have used here one of 3 possible inits for that Publisher, and I would like to use exactly that specific one. DispatchQueue.main.asyncAfter(deadline: .now() + 2) { pub.sink { print(">>>1 \($0)") } .store(in: &self.subscriptions) } pub.sink { print(">>>2 \($0)") } .store(in: &subscriptions)

Output:

 >>>2 1 >>>2 101 >>>2 102 >>>2 1001 >>>2 1002 >>>1 1 >>>1 101 >>>1 102 >>>1 1001 >>>1 1002

But I don't understand this: where is the part when we record stream of inputs and later it is called for a new subscriptions? For me it would be totally the same using [1, 101, 102, 1001, 1002].publisher.

Okay. Is “being able to specify an error/failure completion value” not enough of a difference from Sequence.Publisher? — Sweeper
– Sweeper, Commented Jul 3 at 10:10
is it only difference? I thought that there is some mechanism to wait/record stream and somehow later play it on subscribers.. — Bartłomiej Semańczyk
– Bartłomiej Semańczyk, Commented Jul 3 at 10:14
Pretty extensive exploration here: apeth.com/UnderstandingCombine/publishers/publishersrecord.html — matt
– matt, Commented Jul 3 at 13:14

Sweeper · Accepted Answer · 2025-07-03 13:18:04Z

You are expected to do this "recording" manually. If the recording is synchronous, you can use init(record:). If the recording is asynchronous, you can create a Record.Recording and call init(recording:).

init(output:completion:) is useful if you already have the recording in the form of an array, not as a Record.Recording.

How is this functionally different from using publisher on an Array? Well, not very much. It allows you to specify an error as the completion of the publisher.

Record(output: someArray, completion: .failure(NSError()))

is basically the same as

someArray.publisher.setFailureType(to: NSError.self).append(Fail(error: NSError()))

The more important difference is in their semantics. Asking "why use Record when you can just use Publisher.Sequence" is like asking "why use Just(1) if you can just use [1].publisher", or "why use Empty<Int, Never>() when you can use Int?.none.publisher". Record is more readable and indicates the intent more clearly when you are trying to publish a "record" of some kind.

See also the implementation of Record and Publishers.Sequence in OpenCombine. They have basically the same Inner subscription except Record.Inner completes with the specified Subscribers.Completion.

Collectives™ on Stack Overflow

Where would using Record Publisher with Combine be useful?

1 Answer 1

Comments

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Related