[koin-core] flatten function

[pedrofsn]

I was analyzing an app initialization and I saw Koin consuming a lot of time in flatten function. This logic could be improved?

Flatten function:

@OptIn(KoinInternalApi::class)
fun flatten(modules: List<Module>): Set<Module> {
    fun flat(modules: List<Module>, newModules: MutableSet<Module>){
        modules.forEach{
            newModules += it
            flat(it.includedModules,newModules)
        }
    }
    return mutableSetOf<Module>().apply { flat(modules,this) }
}

I was thinking in process this in IO instead of Main thread. In this case Koin will need to open this to the client pass the flattened list.

Another idea is try to figure out if it possible to apply other algorithm to reduce the time-space complexity.

[pedrofsn]

Chart generated with macrobenchmark library, with Koin consuming a lot.

[pedrofsn]

We did a POC moving our koin to the background, using lazyLoadModules. And it is charming, but don't solve the problem with flatten function.

[pedrofsn]

Thank you @guilhermemagro for providing this analysis and collecting the charts!

[arnaudgiuliani]

there is a first proposal with #1801 - let's try to benchmark results

[arnaudgiuliani]

see #1801