hi, the following throws exceptions

fun main() {
    withSpark {
        dsOf(
            listOf(mapOf("a" to "b", "x" to "y")),
            listOf(mapOf("a" to "b", "x" to "y")),
            listOf(mapOf("a" to "b", "x" to "y"))
        ).showDS().printSchema()
    }
}

I don't think spark can even handle datasets of lists of maps. No idea how that would even be encoded in a Column/Row structure. Any particular reason you need it?

spark can handle it if, for example, I return a list of maps from a UDF & I specify a custom/explicit DataType. The use case is to store a list of

attribute sets: [ { attr1: v1, attr2: v2}, {attrA: va, attrB:vb, attrX:vx} ]

where the attribute names/keys can be different between the sets and even unknown

Wowzer

Thanks !! 😉