Loot tables

[GregHib]

Requirements:

• order matters (bones at bottom)
• weighted
• "always" drops
• nest-able - a loot table of loot tables (what happens if one table is empty but gets called?)
• ability to modify weight dynamically - e.g gradually increase the weight for a player for each time they don't get X as a reward
• without - replacement option. Each time a reward is given, then it can't be given again until the table is reset.
• external modifiers - ring of wealth etc..

Drop tables are essentially simplified utility-based systems #58 so why not plan them to cover these use cases instead? If containers #76 are made functionally #92 couldn't we pass add functions as the values of a table and then just apply/transform the choice onto the container?

Further reading

[GregHib]

See Apache math Enumerated Distributions

[GregHib]

import org.apache.commons.math3.distribution.EnumeratedDistribution
import org.apache.commons.math3.util.Pair

val nested = EnumeratedDistribution(
    listOf(
        Pair("Third", 0.5),
        Pair("Fourth", 0.5)
    )
)
val list = listOf<Pair<Any, Double>>(
    Pair("First", 0.2),
    Pair("Second", 0.4),
    Pair(
        nested,
        0.5
    ),
    Pair("Fifth", 0.3)
)
val distribution = EnumeratedDistribution<Any>(list)

fun main() {
    repeat(10) {
        var sample = distribution.sample()
        while (sample is EnumeratedDistribution<*>) {
            sample = sample.sample()
        }
        println(sample)
    }
}

Pros:

• Quick and simple
• Don't have to worry about randomness
• Easy nesting

Cons:

• Tables have to be static - No system to dynamically adjust the weight for each player.
• No tracking of which nested table was selected ("Your ring of wealth glows" message)