Does this look correct for an `Arb.frequency` function? And would there be any interest for this op...

# kotest

[simon.vergauwen]

Does this look correct for an Arb.frequency function?

And would there be any interest for this operator? You could easily derive choice from this operator.

It was inspired by frequency from ScalaTest, in contrast to choice it takes a weight per Arb.

Untitled

[sam]

We have Arb.choose, is that sufficient for what you need?

[sam]

/**
 * Returns a stream of values based on weights:
 *
 * Arb.choose(1 to 'A', 2 to 'B') will generate 'A' 33% of the time
 * and 'B' 66% of the time.
 *
 * @throws IllegalArgumentException If any negative weight is given or only
 * weights of zero are given.
 */
fun <A : Any> Arb.Companion.choose(a: Pair<Int, A>, b: Pair<Int, A>, vararg cs: Pair<Int, A>)

[simon.vergauwen]

No, that isn’t sufficient for me. I’m using this to create an Arb from a sealed class but I want the create an Arb for every case, and I don’t want them to occur in the same frequency.

Example:

fun <A> pureIO(arb: Arb<A>): Arb<IO<A>> = arb.map { IO.just(it) }
fun failingIO(): Arb<IO<Nothing>> = Arb.throwable().map { IO.raiseError(it) }

fun <A> <http://Arb.Companion.io|Arb.Companion.io>(arb: Arb<A>): Arb<IO<A>> =
 Arb.frequency(5 to pureIO(arb), 1 to failingIO())

[sam]

Thats what choose does?

[simon.vergauwen]

Yes, but based on A not on Arb<A>

[simon.vergauwen]

Pair<Int, A> vs Pair<Int, Arb<A>>

[sam]

Oh

[sam]

I have added to master a version that accepts arbs rather than values. I've also added in aliases to frequency to aid discoverability for those from a haskell quickcheck background. https://github.com/kotest/kotest/issues/1499

[simon.vergauwen]

Awesome 🙌