Hey everyone, I have a question regarding Flow. I have an architecture that looks something like this: UI + ViewModel <- Service <- Repository In this pipeline, Service is maping data from the repository for something that can be consumed by the ViewModel using the map {} operator. Now comes my question: Is it a good idea to put code that changes the system state inside this map or is this a bad idea and I should only map data from one type to another inside the flow ? So basically I would like to schedule/unschedule some alarms as data is registered/unregistered in the repository, but I have the feeling that a flow is more like a query to collect data than some method that can change the system state. I would appreciate some input on what is an expected behavior of a Flow and how it can be use to model system behavior. Thank you

Looks like you need MVI pattern. In case of flow you can use

transform

,

flatMapConcat

or

flatMapMerge

to launch coroutine inside of flow and emit data after delay. But this kind of code will be hard to maintain as it complexity grows.

I strongly believe all operator lambdas should be pure functions. Only act on your flows in the collector