flatMap vs switchMap usage for side effects

[MFlisar]

You use switchMap for all your side effects, so e.g. if you consider an UpdateSideEffect that triggers e.g. 3 actions and we want those actions to always be executed after each update, we need to use flatMap instead like following:

 fun UpdateSideEffect(companion: Companion, clazz: Class<Action>): SideEffect<State, Action> = { actions, state ->
    actions
            .ofType(clazz)
            // flatMap, because we never want to "lose" any side effect action!
            .flatMap {
               
                Observable.just(item)
                        .flatMap {
                            Observable.fromArray(
								Action1,
								Action2,
								Action3
							)
                        }
            }
}

If we call e.g. 3 update actions, we will only get the side effects of the last action (if we use switchMap), but we will get all side effect actions if we use flatMap like in my example. Of course, if order of side effect actions is important we could also use concatMap as well...

Question
I'm unsure why you don't use flatMap by default, is there any particular reason for that? Or is there something speaking against using flatMap at this position instead at all?

[Tapchicoma]

Using flatMap would be fine if your side effect does not consume actions that it emits, new input action does not happen with higher frequency then sideeffect actions emission. TL;DR use flatMap if you know why you are doing it.

switchMap should be used by default as it more fail safe in cases when sideeffect triggered again, while still running from previous input.