Assume a Flutter app built with Clean Architecture and Riverpod. The App has Data-Layer (Data-Sources and Repositories), Domain-Layer (Models and Use-Cases) and Presentation-Layer (Riverpod Providers which provide access to Use-Cases).
The app needs to work offline, thus a local database exists in addition to a remote Web-API. The app gets constant updates via a WebSocket connection, thus the Repositories need to return Streams of data, so that the UI will be kept up-to-date all the time.
Where would you put "Optimistic UI" logic? Should it be placed into a (State)Notifier inside the Presentation-Layer, or should I add an In-Memory Cache to the Repository so that the UI does get an instant response on User action.
Having this logic in the Presentation-Layer would move (or rather duplicate) parts of the business logic to the Presentation-Layer, which I think is not a good idea. However, implementing it with Riverpod would be super simple and Riverpod would also handle disposing the state, when the Response to the User-Action comes back from the Remote or Database through the Stream.