Hi, can you provide an example of post processing? Do you have any plans to add PostProcessor like in MVICore? Thanks

Hello.

PostProcessor

is not required in MVIKotlin, because you always has the access to the current state. You are able to

dispatch(Cmd1)

, then get the updated state via

getState()

, and then perform additional actions if needed and dispatch another commands

dispatch(Cmd2)

.

It can lead to too complicated logic in my opinion. Post Processor looks like a good concerns segregation

Well, avoiding

PostProcessor

and

NewsPublisher

was one of the main ideas at the very beginning. I provided some comparison with MVICore in the discussion. The rationale for this is as follows: 1.

PostProcessor

is rarely needed according to my experience with MVICore 2. When it is actually needed - this means that you need to perform additional actions after updating the state, based on the new state. So you can't just

concat

two actions together. But firstly perform the action A, then perform action B based on the new state. 3. Adding

PostProcessor

is quite noisy - a. You need to introduce

Action

classes; b. You need to convert your

Actor

(aka

Executor

in MVIKotlin) from processing Intents to process Actions; c. You need to add a mapper from

Intents

to `Actions`; d. Finally you need to add the

PostProcessor

class. Instead, MVIKotlin provides the ability to dispatch

Cmd

to the

Reducer

, read the new state straight away and continue the job. The segregation of concerns can be addressed by extracting dedicated private methods for different actions.

Thanks. I got your point. So, another question. Is this code legal from the best practices side?

override fun executeIntent(
		intent: MviIntent,
		getState: () -> MviState,
	) {
		when (intent) {
			is SomeIntent1 -> {
				doWork1(getState)
			}
            is SomeIntent2 -> {
                doWork2()
            }
		}
	}

	private fun doWork1(getState: () -> MviState) {
		scope.launch {
            val condition: Boolean = ///
            ///
            if (condition) {
                executeIntent(
                    SomeIntent2,
                    getState
                )
            }
		}
	}

It is good enough. I would call

doWork2

directly instead of

executeIntent(SomeIntent2, getState)

, though.

I just want to keep them unbound directly. As you said, "purity vs boilerplate"

By

purity vs boilerplate

I meant that in MVICore there is no

getState

function, you receive the state as a value. So all functions look purer. But you have to write more code if you need to check the updated state. Personally I don't see anything wrong by calling dedicated functions directly, instead of

executeIntent

. They are all private, I would prefer calling dedicated functions instead of more generic. But it is up to you, no strict rules here.

Okay, thank you